Bookshelf NBK133363 PDF
Bookshelf NBK133363 PDF
Bookshelf NBK133363 PDF
Number 211
Evidence-Based
Practice
Patient Safety
ii
Preface
The Agency for Healthcare Research and Quality (AHRQ), through its Evidence-based
Practice Centers (EPCs), sponsors the development of systematic reviews to assist public- and
private-sector organizations in their efforts to improve the quality of health care in the United
States. These reviews provide comprehensive, science-based information on common, costly
medical conditions, and new health care technologies and strategies.
Systematic reviews are the building blocks underlying evidence-based practice; they focus
attention on the strength and limits of evidence from research studies about the effectiveness and
safety of a clinical intervention. In the context of developing recommendations for practice,
systematic reviews can help clarify whether assertions about the value of the intervention are
based on strong evidence from clinical studies. For more information about AHRQ EPC
systematic reviews, see www.effectivehealthcare.ahrq.gov/reference/purpose.cfm.
AHRQ expects that these systematic reviews will be helpful to health plans, providers,
purchasers, government programs, and the health care system as a whole. Transparency and
stakeholder input are essential to the Effective Health Care Program. Please visit the Web site
(www.effectivehealthcare.ahrq.gov) to see draft research questions and reports or to join an
email list to learn about new program products and opportunities for input.
We welcome comments on this systematic review. They may be sent by mail to the Task
Order Officer named below at: Agency for Healthcare Research and Quality, 540 Gaither Road,
Rockville, MD 20850, or by email to [email protected].
Carolyn M. Clancy, M.D.
Director
Agency for Healthcare Research and Quality
iii
RAND Corporation:
Breanne Johnsen, B.S.
Jody Wozar Larkin, M.L.I.S.
Acknowledgments
The authors gratefully acknowledge the following individuals for their contributions to this
project.
Technical Expert Panel
Alyce Adams, Ph.D.
Kaiser Permanente
Oakland, CA
iv
vi
Additional Contributors
General Contributors
ECRI Institute
Michele Datko, M.S.
Lydia Dharia
Kitty Donahue
Eileen Erinoff, M.S.L.I.S.
Gina Giradi, M.S.
vii
viii
Contents
Executive Summary .................................................................................................................. ES-1
Part 1. Overview ............................................................................................................................. 1
Chapter 1. Introduction ............................................................................................................. 1
Who Will Use This Report, and for What Purpose?........................................................... 2
References ........................................................................................................................... 3
Chapter 2. Methods ................................................................................................................... 5
Topic Development ............................................................................................................. 5
Project Overview ................................................................................................................ 6
Topic Refinement................................................................................................................ 6
Evidence Assessment Framework .................................................................................... 11
Evidence Review Process ................................................................................................. 13
Assessing Quality of Individual Studies ........................................................................... 15
Assessing Strength of Evidence for a Patient Safety Practice .......................................... 16
Summarizing the Evidence ............................................................................................... 16
Setting Priorities for Adoption of Patient Safety Practices ............................................... 18
Future Research Needs ..................................................................................................... 19
Peer and Public Review Process ....................................................................................... 20
References ......................................................................................................................... 20
Part 2. Evidence Reviews of Patient Safety Practices .................................................................. 21
Part 2a. Practices Designed for a Specific Patient Safety Target ................................................. 23
Section A. Adverse Drug Events ............................................................................................ 23
Chapter 3. High-Alert Drugs: Patient Safety Practices for Intravenous Anticoagulants .. 23
Chapter 4. Clinical Pharmacists Role in Preventing Adverse Drug Events: Brief Update
Review .............................................................................................................................. 31
Chapter 5. The Joint Commissions Do Not Use List: Brief Review (NEW*) ............ 41
Chapter 6. Smart Pumps and Other Protocols for Infusion Pumps: Brief Review
(NEW*) ............................................................................................................................. 48
Section B. Infection Control ................................................................................................... 55
Chapter 7. Barrier Precautions, Patient Isolation, and Routine Surveillance for Prevention
of Healthcare-Associated Infections: Brief Update Review ........................................... 55
Chapter 8. Interventions To Improve Hand Hygiene Compliance: Brief Update
Review .............................................................................................................................. 67
Chapter 9. Reducing Unnecessary Urinary Catheter Use and Other Strategies To Prevent
Catheter-Associated Urinary Tract Infections: Brief Update Review............................. 73
Chapter 10. Prevention of Central Line-Associated Bloodstream Infections: Brief Update
Review .............................................................................................................................. 88
Chapter 11. Ventilator-Associated Pneumonia: Brief Update Review ......................... 110
Chapter 12. Interventions To Allow the Reuse of Single-Use Devices: Brief Review
(NEW*) ........................................................................................................................... 117
Section C. Surgery, Anesthesia, and Perioperative Medicine .............................................. 122
Chapter 13. Preoperative and Anesthesia Checklists ...................................................... 122
Chapter 14. Use of Report Cards and Outcome Measurements To Improve Safety of
Surgical Care: American College of Surgeons National Surgical Quality Improvement
Program (NEW*) ............................................................................................................ 140
xi
Chapter 15. Prevention of Surgical Items Being Left Inside Patient: Brief Update
Review ............................................................................................................................ 158
Chapter 16. Operating Room Integration and Display Systems: Brief Review
(NEW*) ........................................................................................................................... 163
Chapter 17. Use of Beta Blockers To Prevent Perioperative Cardiac Events: Brief
Update Review ............................................................................................................... 169
Chapter 18. Use of Real-Time Ultrasound Guidance During Central Line Insertion: Brief
Update Review ............................................................................................................... 172
Section D. Safety Practices Aimed Primarily at Hospitalized Elders ................................... 178
Chapter 19. Preventing In-Facility Falls ......................................................................... 178
Chapter 20. Preventing In-Facility Delirium .................................................................. 201
Section E. General Clinical Topics ....................................................................................... 212
Chapter 21. Preventing In-Facility Pressure Ulcers ........................................................ 212
Chapter 22. Inpatient Intensive Glucose Control Strategies To Reduce Death and
Infection (NEW*) ........................................................................................................... 233
Chapter 23. Interventions To Prevent Contrast-Induced Acute Kidney Injury .............. 248
Chapter 24. Rapid Response Systems (NEW*) .............................................................. 257
Chapter 25. Medication Reconciliation Supported by Clinical Pharmacists (NEW*) ... 270
Chapter 26. Identifying Patients at Risk for Suicide: Brief Review (NEW*)................ 287
Chapter 27. Strategies To Prevent Stress-Related Gastrointestinal Bleeding (Stress Ulcer
Prophylaxis): Brief Update Review .............................................................................. 297
Chapter 28. Prevention of Venous Thromboembolism: Brief Update Review............. 303
Chapter 29. Preventing Patient Death or Serious Injury Associated With Radiation
Exposure From Fluoroscopy and Computed Tomography: Brief Review (NEW*) ...... 310
Chapter 30. Ensuring Documentation of Patients Preferences for Life-Sustaining
Treatment: Brief Update Review .................................................................................. 320
Part 2b. Practices Designed To Improve Overall System/Multiple Targets ............................... 325
Chapter 31. Human Factors and Ergonomics ................................................................. 325
Chapter 32. Promoting Engagement by Patients and Families To Reduce Adverse Events
(NEW*) ........................................................................................................................... 351
Chapter 33. Promoting Culture of Safety ....................................................................... 362
Chapter 34. Effect of Nurse-to-Patient Staffing Ratios on Patient Morbidity
and Mortality................................................................................................................... 372
Chapter 35. Patient Safety Practices Targeted at Diagnostic Errors (NEW*) ................ 385
Chapter 36. Monitoring Patient Safety Problems (NEW*) ............................................ 405
Chapter 37. Interventions To Improve Care Transitions at Hospital Discharge
(NEW*) ........................................................................................................................... 425
Chapter 38. Use of Simulation Exercises in Patient Safety Efforts ................................ 439
Chapter 39. Obtaining Informed Consent From Patients: Brief Update Review .......... 461
Chapter 40. Team-Training in Health Care: Brief Update Review ............................... 472
Chapter 41. Computerized Provider Order Entry With Clinical Decision Support
Systems: Brief Update Review...................................................................................... 480
Chapter 42. Tubing Misconnections: Brief Review (NEW*) ........................................ 487
Chapter 43. Limiting Individual Providers Hours of Service: Brief Update
Review ............................................................................................................................ 493
Part 3. Discussion........................................................................................................................ 499
xii
xiii
Table 3, Chapter 14. Percent of Medicare surgical cases covered by the national surgical
quality improvement program..................................................................................................... 148
Table 4, Chapter 14. List of American College of Surgeons national surgical quality
improvement program collaboratives including type, number of sites, and payor ..................... 149
Table 5, Chapter 14. Example of reductions in complications and associated costs .................. 152
Table 6, Chapter 14. Summary table .......................................................................................... 154
Table 1, Chapter 15. Summary table .......................................................................................... 161
Table 1, Chapter 16. Summary table .......................................................................................... 167
Table 1, Chapter 17. Summary table .......................................................................................... 171
Table 1, Chapter 18. Summary table .......................................................................................... 175
Table 1, Chapter 19. Components of multi-factorial falls prevention trials in hospitals,
1999 to 2009 ............................................................................................................................... 180
Table 2, Chapter 19. Meta-analytic estimate of the effect of multicomponent fall intervention
programs on inpatient fall rates................................................................................................... 184
Table 3, Chapter 19. Abridged evidence tables, adapted from Oliver and colleagues ............... 186
Table 4, Chapter 19. Implementation themes highlighted in implementation studies................ 191
Table 5, Chapter 19. Summary table .......................................................................................... 198
Table 1, Chapter 20. Summary table .......................................................................................... 210
Table 1, Chapter 21. Components of pressure ulcer prevention trials in U.S. hospitals,
2000 to 2011 ............................................................................................................................... 214
Table 2, Chapter 21. Components of pressure ulcer prevention trials in long term care,
2000 to 2011 ............................................................................................................................... 215
Table 3, Chapter 21. Summary table .......................................................................................... 230
Table 1, Chapter 22. Large trials (n > 500) evaluating the health outcome effects of intensive
insulin therapy............................................................................................................................. 234
Table 2, Chapter 22. Summary table .......................................................................................... 244
Table 1, Chapter 23. Summary table .......................................................................................... 254
Table 1, Chapter 24. RRS Summary table: effectiveness ........................................................... 261
Table 2, Chapter 24. RRS Summary table: implementation studies........................................... 264
Table 3, Chapter 24. Summary table .......................................................................................... 267
Table 1, Chapter 25. Studies of medication reconciliation that include assessment of clinically
significant unintended discrepancies and emergency department visits and hospitalizations
within 30 days of discharge ........................................................................................................ 272
Table 2, Chapter 25. Key features of the 12 included medication reconciliation interventions . 276
Table 3, Chapter 25. Medication reconciliation in varying levels of intensity as seen
in published studies ..................................................................................................................... 282
Table 4, Chapter 25. Summary table .......................................................................................... 283
Table 1, Chapter 26. Summary table .......................................................................................... 293
Table 1, Chapter 27. Summary table .......................................................................................... 300
Table 1, Chapter 28. Summary table .......................................................................................... 307
Table 1, Chapter 29. Summary table .......................................................................................... 317
Table 1, Chapter 30. Summary table .......................................................................................... 323
Table 1, Chapter 31. Key characteristics of HFE and its application to patient safety .............. 327
Table 2, Chapter 31. HFE mechanisms between system design and patient safety ................... 331
Table 3, Chapter 31. Examples of HFE design principles .......................................................... 332
Table 4, Chapter 31. HFE issues in selected care settings .......................................................... 339
xiv
xv
Figures
Figure A. Framework for evidence assessment of patient safety practices .............................. ES-3
Figure 1, Chapter 2. Overview of the project ................................................................................. 7
Figure 2, Chapter 2. Framework for evidence assessment of patient safety practices ................. 11
Figure 3, Chapter 2. The evidence review process ....................................................................... 15
Figure 1, Chapter 5. Official do not use list .............................................................................. 41
Figure 1, Chapter 9. Lifecycle of the urinary catheter .................................................................. 74
Figure 2, Chapter 9. Summary of CAUTI and urinary catheter outcomes from 14 studies ......... 79
Figure 3, Chapter 9. Summary of CAUTI and urinary catheter outcomes from 12 additional
studies ........................................................................................................................................... 80
Figure 1, Chapter 13. Screenshot of adoption and diffusion of the WHO surgical safety
checklist ...................................................................................................................................... 133
Figure 1, Chapter 14. Example of observed to expected ratio reporting for American College of
Surgeons national surgical quality improvement program ......................................................... 141
Figure 2, Chapter 14. Geographic distribution of American College of Surgeons national surgical
quality improvement program participating sites ....................................................................... 148
Figure 3, Chapter 14. The 12 steps to implement the national surgical quality improvement
program ....................................................................................................................................... 151
Figure 1, Chapter 19. Multi-systemic fall prevention model ...................................................... 183
Figure 2, Chapter 19. Meta-analysis from Oliver et al. 2006 for multifaceted interventions
in hospital falls (random effects model) ..................................................................................... 185
Figure 1, Chapter 22. Short-term mortality in studies of intensive insulin therapy, by inpatient
setting and condition ................................................................................................................... 237
Figure 2, Chapter 22. Risk for hypoglycemia in studies of intensive insulin therapy in various
inpatient settings ......................................................................................................................... 239
Figure 1, Chapter 25. Overview of medication reconciliation.................................................... 271
Figure 2, Chapter 25. Median and interquartile range for the number of clinically significant
unintentional discrepancies per patient for the 13 included interventions .................................. 279
Figure 3, Chapter 25. Emergency department visits and hospitalizations within 30 days
of discharge in three randomized controlled trials ...................................................................... 280
Figure 1, Chapter 31. SEIPS model of work system and patient safety ..................................... 330
Figure 1, Chapter 34. Hospital organization, nursing organization, and patient outcomes ........ 373
Figure 2, Chapter 34. Despins model on patient risk detection ................................................. 373
Figure 3, Chapter 34. Tourangeaus model on determinants of 30-day mortality...................... 374
Figure 4, Chapter 34. Thornlows model on cascade iatrogenesis: postoperative respiratory
failure .......................................................................................................................................... 375
Figure 5, Chapter 34. Pooled odds ratio of quartiles of death by nurse staffing levels .............. 376
Figure 1, Chapter 35. Interventions by type ............................................................................... 389
Figure 2, Chapter 35. Intervention studies by year ..................................................................... 390
Box
Box 1. Taxonomy of interventions to improve transitional care at hospital discharge .............. 426
xvi
Appendixes
Appendix A. Original List of Patient Safety Practices
Appendix B. AMSTAR: A Measurement Tool To Assess Systemic Reviews
Appendix C. Literature Searches and Topic-Specific Methods
Appendix D. Supplementary Evidence Tables
(*NEW) refers to a new review that was not addressed in the 2001 report, Making Health Care
Safer: A Critical Analysis of Patient Safety Practices.
xvii
Executive Summary
Background
The 1999 Institute of Medicine report To Err is Human: Building a Safer Health System, is
credited by many with launching the modern patient safety movement.1 A year after this report
was published, as part of its initial portfolio of patient safety activities, the Agency for
Healthcare Research and Quality (AHRQ) commissioned a group from the University of
California, San Francisco-Stanford Evidence-based Practice Center (EPC) to analyze evidence
behind a diverse group of patient safety practices (PSPs) that existed at that time.
The resulting 2001 report, Making Health Care Safer: A Critical Analysis of Patient Safety
Practices,2 hereafter referred to as Making Health Care Safer, was both influential and
controversial. A significant number of copies of the report were distributed by AHRQ, and it
became a cornerstone of other efforts (such as the National Quality Forums 34 Safe Practices
for Better Healthcare list)3 to rank safety practices by strength of evidence. However, the low
rankings given to some popular safety practices, such as computerized order entry, raised
fundamental questions about the role of evidence-based medicine in quality and safety practices.
Since the Making Health Care Safer report was published, the safety field has matured.
Regulators and accreditors encourage health care organizations to adopt safe practices and to
avoid adverse events that are considered wholly or largely preventable. A significant amount of
money and person-hours have been invested in efforts to improve safety, and almost all healthcare delivery organizations regard safety as a primary strategic priority.
However, evidence indicates that progress has not matched the efforts and investment. Some
patient safety practices (PSPs) have resulted in unintended consequences, whereas others have
been shown to be highly context dependent, working effectively in a research setting but failing
during broader implementation. In the past 2 years, three studies have found high rates of
preventable harm in hospitals,4-6 one of which found no improvement in adverse event rates from
2003 to 2008.
Against this backdrop, AHRQ commissioned an updated research report on the state of PSPs.
Because many of the project team members and much of the methodology were drawn from the
initial Making Health Care Safer project, and because most of the relevant practices were
reviewed then, we see this report as a natural sequel to the 2001 report. However, because of the
burgeoning literature relevant to patient safety and the limits of budget and time, we chose to
examine a subset of PSPs (chosen through methods described below). Moreover, part of the
maturation of the safety field has included a deeper appreciation of the importance of context in
patient safety practices, a topic examined by our research team in the 2010 report, Assessing the
Evidence for Context-Sensitive Effectiveness and Safety of Patient Safety Practices: Developing
Criteria, hereafter referred to as Context Sensitivity.7 Accordingly, this report emphasizes
matters of context and generalizability, as well as unintended consequences, to a greater degree
than the 2001 Making Health Care Safer report.
Objectives
The goal of this project was to conduct a systematic literature review evaluating the evidence
for a large number of patient safety practices.
ES-1
Analytic Framework
For this report, we adopted the definition of a PSP used in the 2001 Making Health Care
Safer report:
A Patient Safety Practice is a type of process or structure whose
application reduces the probability of adverse events resulting from
exposure to the health care system across a range of diseases and
procedures.
The framework for considering the evidence regarding a PSP was worked out as part of the
report on Context Sensitivity.7 One of the principal challenges in the review of PSPs has been
addressing the question of what constitutes evidence for PSPs. Many practices intended to
improve quality and safety are complex sociotechnical interventions whose targets may be entire
health care organizations or groups of providers, and these interventions may be targeted at rare
events. To address the challenge regarding what constitutes evidence, we recognize that PSPs
must be evaluated along two dimensions: the evidence regarding the outcomes of the safe
practices, and the contextual factors influencing the practices use and effectiveness.
These dimensions are represented in Figure A, which depicts a sample PSP that consists of a
bundle of components (the individual boxes), and the context within which the PSP is embedded.
Important evaluation questions, as depicted on the right in the figure, include effectiveness and
harms, implementation, and adoption and spread. We then apply criteria to evaluate the four
factors that together constitute quality (depicted as puzzle pieces in the bottom half of the figure.
They include:
1. Constructs about the PSP, its components, context factors, outcomes, and ways to
accurately measure these constructs
2. Logic model or conceptual framework about the expected relationships among these
constructs
3. Internal validity to assess the PSP results in a particular setting
4. External validity to assess the likelihood of being able to garner the same results in
another setting
We then synthesize this information into an evaluation of the strength of the evidence for a
particular PSP.
ES-2
The principal results of the Context Sensitivity report included the following key points.
Whereas controlled trials of PSP implementations offer investigators greater control of
sources of systematic error than do observational studies, trials often are not feasible in
terms of time or resources. Also, controlled trials are often not possible for PSPs
requiring large-scale organizational change or PSPs targeted at very rare events.
Furthermore, the standardization imposed by the clinical trial paradigm may stifle the
adaptive responses necessary for some quality improvement or patient safety projects.
Hence, researchers may need to use designs other than randomized controlled trials to
develop strong evidence about the effectiveness of some PSPs.
Regardless of the study design chosen for an evaluation, components that are critical for
evaluating a PSP in terms of how it worked in the study site, and whether it might work
in other sites, include the following:
o Explicit description of the theory for the chosen intervention components, and/or an
explicit logic model for why this PSP should work
o Description of the PSP in sufficient detail that it can be replicated, including the
expected change in staff roles
o Measurement of contexts
o Explanation, in detail, of the implementation process, the actual effects on staff roles,
and changes over time in the implementation or the intervention
o Assessment of the impact of the PSP on outcomes and possible unexpected effects
(including data on costs, when available)
o For studies with multiple intervention sites, assessment of the influence of context on
intervention and implementation effectiveness (processes and clinical outcomes)
High priority contexts for assessing any PSP implementation include measuring and
information for each of the following four domains:
o Structural organizational characteristics (such as size, location, financial status,
existing quality and safety infrastructure)
o External factors (such as regulatory requirements, the presence in the external
environment of payments or penalties such as pay-for-performance or public
ES-3
Methods
We divided the project into three phases: topic refinement, the evidence review, and the
critical review and interpretation of the evidence. The project team performed topic refinement
and conducted the critical review of the evidence jointly with the Technical Expert Panel (TEP),
which had also participated in the Context Sensitivity project. This TEP included many of the
key patient safety leaders in the United States, Canada, and the United Kingdom: experts in
specific PSPs and evaluation methods and persons charged with implementing PSPs in hospitals
and clinics.
Topic Refinement
Because the goals of the project were to assess the evidence of the effectiveness of new safe
practices and the evidence of implementation for current safe practices, most PSPs were eligible
for this review. Thus, our first task was to refine the scope of the topic to fit within the timeframe
and budget of the project, a task undertaken by the project team and the TEP. To accomplish this
task, we created an initial list of 158 PSPs that we considered potentially eligible for inclusion.
Through a process of internal team triage, group discussion with the TEP, and formal TEP votes,
we narrowed the list to 41 PSPs for which a review of evidence was judged likely to be most
helpful to providers, policymakers, and patients. However, this number of PSPs was still too
large for us to review the evidence comprehensively within the timeframe. For that reason, we
asked our TEP whether breadth or depth was likely to be more valuable for stakeholders; in
other words, we asked whether the review should focus on fewer topics in more detail or cover
all topics but with less detail. Our TEP recommended a hybrid approach, in which some topics
would be reviewed in depth, whereas other topics would receive only a brief review.
Topics could be considered as needing only a brief review for several reasons: the PSP is
already well established; stakeholders need to know only whats new since the last time a topic
was reviewed in depth; new evidence suggests the PSP may not be as effective as originally
believed, so it is no longer a priority PSP; or the PSP is emerging with little evidence
accumulated. We ultimately ended up with 18 in-depth reviews and 23 brief reviews.
ES-4
Evidence Reviews
In-Depth Reviews
Overall approach. For many of the 18 topics designated to receive an in-depth review, a
systematic review was likely to exist. Thus, a search to identify existing systematic reviews was
usually the project teams first step. To assess the potential utility of such reviews, we followed
the procedures proposed by Whitlock and colleagues,8 which essentially meant addressing the
following two questions: (1) Is the existing review sufficiently on topic to be of use? (2) Is the
existing review of sufficient quality for us to have confidence in the results?
If an existing systematic review was judged to be sufficiently on topic and of acceptable
quality, we took one of two steps. We either performed an update search; that is, we searched
databases for new evidence published since the end date of the search in the existing systematic
review. Or, we conducted a search for signals for updating. Such searches generally followed
the criteria proposed by Shojania and colleagues.9 The searches involved a search of high-yield
databases and journals for pivotal studies that could be a signal that a systematic review is outof-date. Any evidence identified via the update search or the signals search was added to the
evidence base from the existing systematic review.
Some PSPs had no existing systematic reviews, while other PSPs had prior reviews that were
either not sufficiently relevant or were not of sufficient quality to be used. In those situations, we
conducted new searches using guidance as outlined in AHRQs Methods Guide for
Effectiveness and Comparative Effectiveness Reviews.10
Evidence about context, implementation, and adoption are key aspects of this review. We
searched for evidence on these topics in two ways:
We looked for and extracted data about contexts and implementation from the articles
contributing to the evidence of effectiveness.
We identified implementation studies from our literature searches. Implementation
studies focus on the implementation process, particularly the elements demonstrated or
believed to be of special importance for the success, or lack of success, of the
intervention. To be eligible, implementation studies needed to either report or be linked
to reports of effectiveness outcomes.
Reporting format. We took the format for in-depth reviews from AHRQs Context
Sensitivity report. Table A outlines the format of the in-depth reviews.
ES-5
Brief Reviews
Brief reviews are not full systematic reviews. The goals of the brief reviews covered in this
report varied by PSP according to the needs of stakeholders. The assessment could focus on
either information about effectiveness of an emerging PSP or implementation of an established
PSP; alternatively, the review could explore whether new evidence calls into question the
effectiveness of an existing PSP. Thus, the methods for the brief reviews differed by topic.
However, in general, brief reviews were conducted by a content expert who worked with the
project team. The brief reviews involved focused literature searches for evidence relevant to the
specific need. The evidence was then narratively summarized in a format that also varied with
the particular goal.
Evidence Summary
We judged that users of this report would want a summary of the evidence for each topic.
Such summary messages may facilitate an uptake of the findings. The project team developed
the following summary domains with input from the TEP.
ES-6
Scope of the problem. In general, we addressed two issues: the frequency of the safety problem,
and the severity of each average event. For benchmarks, we regarded safety problems that occur
approximately once per 100 hospitalizations as common; examples include falls, venous
thromboembolism (VTE), potential adverse drug events, or pressure ulcers. In contrast, events an
order of magnitude or more lower in frequency were considered rare; such events include
inpatient suicide, wrong-site surgery, and surgical items being left inside a patient. The scope
must also consider the severity of each event; for instance, most falls do not result in injury, and
most potential adverse drug events do not result in clinical harm. However, each case of inpatient
suicide or wrong-site surgery is devastating.
Strength of evidence for effectiveness. This assessment follows a framework for strength of
evidence that the project team adapted from existing EPC Methods guidance12 to increase the
relevance to patient safety practices. This means we included in strength of evidence assessments
evidence about context, implementation, and the use of theory or logic models, in addition to
standard EPC criteria on inconsistency, in precision, and the possibility of reporting bias.
Evidence on potential for harmful unintended consequences. Most PSP evaluators have not
explicitly assessed the possibility of harm. Consequently, this domain includes evidence of both
actual harm and the potential for harm. The ratings on known or potential harms ranged from
high risk of harm to low (or negligible) risk of harm; in some cases, the evidence was too sparse
to provide a rating.
Estimate of costs. This domain is speculative, because most evaluations do not present cost data.
However, we believed that providing at least a rough estimate of cost would be beneficial
information to include in this report. Therefore, we used the following categories and
benchmarks to provide a rough estimate of cost, noting, where necessary, the factors that might
cause cost estimates to vary.
Low cost. PSPs that do not require hiring new staff or large capital outlays but instead
involve training existing staff and purchasing some supplies. Examples include most fall
prevention programs, VTE prophylaxis, and medical history abbreviations designated as
Do Not Use.
Medium cost. PSPs that might require hiring one or a few new staff members, have
modest capital outlays, or incur ongoing monitoring costs. Examples include some fall
prevention programs, many clinical pharmacist interventions, and participation in the
American College of Surgeons outcomes reporting system ($135,000/year).13
High cost. PSPs that require hiring substantial numbers of new staff, have considerable
capital outlays, or both. Examples include computerized order entry (because it requires
an electronic health record), having to hire many nurses to achieve a certain nurse-topatient ratio, or facility-wide infection control procedures (estimated at $600,000 year for
a single intensive care unit).14
Implementation issues. This section summarizes how much we know about how to implement
the PSP and how difficult it is to implement. To approach the question of how much we know,
we considered the available evidence about implementation, the existence of data about the
effect and influence of context, the degree to which a PSP has been implemented, and the
presence of implementation tools, such as written materials and training manuals.
ES-7
For the question of implementation difficulty, we used three categories: difficult, for PSPs
that require large scale organizational change; not difficult, for PSPs that require protocols for
drugs or devices, such as those needed to reduce radiation exposure or to help prevent stressrelated gastrointestinal bleeding; and moderate, for PSPs falling between the extremes.
Results
We completed 18 in-depth reviews and 23 brief reviews. Table B summarizes the findings
according to the five main issues previously described (scope, strength of evidence, harms, costs,
and implementation). The table is organized into two main sections: PSPs aimed at a specific
(single) patient safety target, such as adverse drug events, or general clinical topics, such as
preventing pressure ulcers; and PSPs designed to improve the overall system or to address
multiple patient safety targets, such as nurse-staffing ratios or computerized provider order entry.
In some cases, the text in the PSP column differs slightly from the chapter heading for that PSP.
This is due to the desire by our TEP to include the target safety problem in the table (if targeted
at a specific safety problem), more specification, or an example of the PSP (e.g., adding such as
a centralized display of consolidated data to the PSP designated as operating room integration
and display systems).
Table B. Summary table*
Patient Safety Practice
Scope of the
Problem Targeted
by the PSP
(Frequency/
Severity)
Practices Designed for a Specific Patient Safety Target
Adverse Drug Events
High-alert drugs: patient safety
Common/Moderate
practices for intravenous
anticoagulants;
in-depth review
Use of clinical pharmacists to
Common/Low
prevent adverse drug events;
brief review
The Joint Commissions Do Not
Common/Low
Use list; brief review
Smart infusion pumps; brief
Common/Low
review
Infection Control
Barrier precautions, patient
Common/Moderate
isolation, and routine surveillance
for the prevention of healthcareassociated infections; brief
review
Interventions to improve hand
Common/Moderate
hygiene compliance; brief review
Strength of
Evidence for
Effectiveness
of the PSPs
Evidence or
Potential for
Harmful
Unintended
Consequences
Estimate of
Cost
Implementation
Issues:
How Much Do We
Know?/How Hard Is
it?
Low
Low-to-moderate
Low
Little/Moderate
Moderate-tohigh
Low
High
Little/Moderate
Low
Negligible
Low
Low
Low
Moderate
Little/Probably not
difficult
Moderate/Moderate
Moderate
Moderate
(isolation of
patients)
Moderate-tohigh
Moderate/Moderate
Low
Low
Low
Moderate/Moderate
ES-8
Scope of the
Problem Targeted
by the PSP
(Frequency/
Severity)
Common/Moderate
Strength of
Evidence for
Effectiveness
of the PSPs
Moderate-tohigh
Evidence or
Potential for
Harmful
Unintended
Consequences
Low
Estimate of
Cost
Low
Implementation
Issues:
How Much Do We
Know?/How Hard Is
it?
Moderate/Moderate
Common/Moderate
Moderate-tohigh
Low
Low-tomoderate
Ventilator-associated
Common/High
pneumonia; brief review
Interventions to allow the reuse
Common/Low
of single use devices; brief
review
Surgery, Anesthesia, and Perioperative Medicine
Preoperative checklists and
Common/Moderate
anesthesia checklists to prevent
a number of operative safety
events, such as surgical site
infections and wrong site
surgeries; in-depth review
The use of ACS-NSQIP report
Common/High
cards and outcome
measurements to decrease
perioperative morbidity and
mortality; in-depth review
New interventions to prevent
Rare/Low
surgical items from being left
inside a patient; brief review
Moderate-tohigh
Low
Low
Low
Low-tomoderate
Low
High
Negligible
Low
A lot/Moderate
Moderate-tohigh
Low
Moderate
Moderate/Moderate
Low
Negligible
Little
Common/Low-tohigh
Low
Negligible
Low if it
simply
involves
more
frequent
manual
counting;
high if RFID
is used
Moderate
Common/High
High (death,
stroke,
hypotension, and
bradycardia)
Low
NA
Common/Low-tomoderate
High evidence
harms may
equal or
exceed
benefits
High
Negligible
Low-tomoderate
A lot/Moderate
ES-9
Moderate-to-difficult/
Not difficult
(implementation of a
bundle)-to-moderate
(understanding
organization culture
and context)
Moderate/Moderate
A lot/Not difficult
Moderate/Moderate
Scope of the
Problem Targeted
by the PSP
(Frequency/
Severity)
Safety Practices for Hospitalized Elders
Multicomponent interventions to
Common/Low
prevent in-facility falls; in-depth
review
Multicomponent interventions to
prevent in-facility delirium; indepth review
General Clinical Topics
Multicomponent initiatives to
prevent pressure ulcers; in-depth
review
Inpatient, intensive, glucose
control strategies to reduce
death and infection; in-depth
review
Interventions to prevent contrastinduced acute kidney injury; indepth review
Rapid-response systems to
prevent failure-to-rescue; indepth review
Medication reconciliation
supported by clinical
pharmacists; in-depth review
Identifying patients at risk for
suicide; brief review
Strategies to prevent stressrelated gastrointestinal bleeding
(stress ulcer prophylaxis); brief
review
Strategies to increase
appropriate prophylaxis for
venous thromboembolism; brief
review
Preventing patient death or
serious injury associated with
radiation exposure from
fluoroscopy and computed
tomography through technical
interventions, appropriate
utilization, and use of algorithms
and protocols; brief review
Ensuring documentation of
patient preferences for lifesustaining treatment, such as
advanced directives; brief review
Increasing nurse-to-patient
staffing ratios to prevent death;
in-depth review
Strength of
Evidence for
Effectiveness
of the PSPs
High
Evidence or
Potential for
Harmful
Unintended
Consequences
Estimate of
Cost
Implementation
Issues:
How Much Do We
Know?/How Hard Is
it?
Moderate
Moderate/Moderate
Moderate
Moderate/Moderate
Common/Low
Moderate
Moderate
(increased use of
restraints and/or
sedation)
Low
Common/Moderate
Moderate
Negligible
Moderate
Moderate/Moderate
Common/Moderate
Moderate-tohigh evidence
it doesnt help
High
(hypoglycemia)
Low-tomoderate
NA
Common/Low
Low
Negligible
Low
Little/Not difficult
Common/High
Moderate
Low
Moderate
Moderate/Moderate
Common/Low
Moderate
Low
Moderate
Moderate/Moderate
Rare/High
Low
Low
Moderate
Little/Moderate
Rare/Moderate
Moderate
Moderate
(pneumonia)
Moderate
Little/Not difficult
Common/Moderate
High
Moderate
(bleeding)
Low
Little/Moderate
Rare/High
Moderate
Negligible
Low
Moderate/Not difficult
Common/Moderate
Moderate
Low
Low
Moderate/Moderate
Common/High
Moderate
Low
High
A lot/Not difficult
ES-10
Scope of the
Strength of
Problem Targeted
Evidence for
by the PSP
Effectiveness
(Frequency/
of the PSPs
Severity)
Practices Designed To Improve Overall System/Multiple Targets
Increasing nurse-to-patient staff
Common/High
Low
ratios to prevent falls, pressure
ulcers, and other nursing
sensitive outcomes (other than
mortality); in-depth review
Incorporation of human factors
Potentially
Not assessed
and ergonomics in the design of
applicable to all
systematically,
health care practices by hiring an patient safety
but moderateexpert or training clinicians in
problems
to-high
human factors; in-depth review
evidence for
some specific
applications
Promoting engagement by
Common
Emerging
patients and families to reduce
practice (few
adverse events (such as patients
studies
encouraging providers to wash
available)
their hands); in-depth review
Interventions to promote a
Common/Low-toLow
culture of safety; in-depth review
high
Evidence or
Potential for
Harmful
Unintended
Consequences
Estimate of
Cost
Implementation
Issues:
How Much Do We
Know?/How Hard Is
it?
Low
High
A lot/Not difficult
Negligible
Moderate
A lot/Moderate
Uncertain
Low
Little/Moderate
Uncertain
Emerging
practice (few
studies
available)
Low
Uncertain
Lowtomoderate
(varies)
Varies
Negligible
High
Moderate/Difficult
Low
Negligible
Moderate-tohigh
Little/Difficult
Uncertain
Moderate
Moderate
Common/Moderate
Moderate-tohigh for
specific topics
Moderate
Negligible
Low
Moderate/Not difficult
Common/High
Moderate
Low
Moderate
Common/Moderate
Low-tomoderate
Low-to-moderate
High
Moderate/Moderateto-difficult
Moderate/Difficult
Common/Moderate
Low
Low
Low
Moderate/Not difficult
Common/High
Common/Low-tohigh
Common/Moderate
Common/Moderateto-high
ES-11
Scope of the
Problem Targeted
by the PSP
(Frequency/
Severity)
Common/Moderate
Strength of
Evidence for
Effectiveness
of the PSPs
Evidence or
Estimate of
Implementation
Potential for
Cost
Issues:
How Much Do We
Harmful
Know?/How Hard Is
Unintended
Consequences
it?
Limiting trainee work hours; brief
Low
Moderate (at
High
Moderate/Difficult
review
least); includes
lack of training
time
Abbreviations: ACS NSQIP=American College of Surgeons National Surgical Quality Improvement Program; NA = not
available; PSP: Patient Safety Practice; RFID = radio-frequency identification.
*In some cases, the text in the PSP column differs slightly from the chapter heading for that PSP. This difference is attributable
to our Technical Expert Panels desire to include the target safety problem (if the practice is in fact targeted at a specific safety
problem), more specification, or an example of the PSP (e.g., adding such as a centralized display of consolidated data to the
PSP designated as operating room integration and display systems).
Rating Scales:
Scope of the problem targeted by the PSP (frequency/severity): frequency = rare or common; severity = low, moderate, or high.
Strength of evidence for effectiveness of the PSPs: low, moderate, or high.
Evidence or potential for harmful unintended consequences: negligible, low, moderate, or high.
Estimate of cost: low, moderate, or high.
Implementation issues: How much do we know? = little, moderate, or a lot; How hard is it? = not difficult, moderate, or difficult.
Discussion
Since the 2001 report, Making Health Care Safer, a vast amount of new information on
PSPs has emerged. Compared with a decade ago, more agreement is now evident on what
constitutes evidence of effectiveness and the importance of implementation and context. In this
review, we determined that the strength of evidence was at least moderate for 20 PSPs, or about
half of those reviewed. For 26 of the PSPs, we judged that evidence of at least moderate strength
was available on how to implement them.
Thus, sufficient evidence exists about effectiveness and implementation to permit our TEP
members to conclude that some PSPs are ready to be strongly encouraged for adoption by
health care providers. Their assessments were based explicitly on the combination of the
available evidence with their expert judgment in interpreting the evidence. The 10 strongly
encouraged PSPs are listed in Table C.
Table C. Strongly encouraged patient safety practices
Preoperative checklists and anesthesia checklists to prevent operative and post-operative events
Bundles that include checklists to prevent central line-associated bloodstream infections
Interventions to reduce urinary catheter use, including catheter reminders, stop orders, or nurse-initiated
removal protocols
Bundles that include head-of-bed elevation, sedation vacations, oral care with chlorhexidine, and subglotticsuctioning endotracheal tubes to prevent ventilator-associated pneumonia
Hand hygiene
Do Not Use list for hazardous abbreviations
Multicomponent interventions to reduce pressure ulcers
Barrier precautions to prevent healthcare-associated infections
Use of real-time ultrasound for central line placement
Interventions to improve prophylaxis for venous thromboembolisms
The TEP members concluded that several other PSPs had sufficient evidence of effectiveness
and implementation, and that they should be encouraged for adoption. The 12 encouraged
PSPs are listed in Table D.
ES-12
The 22 PSPs in Tables C and D represent practices that health care providers can consider for
adoption now. This recommendation particularly applies to the 10 strongly encouraged
practices. For these practices, at least in the judgment of our TEP, there is sufficient knowledge
to implement them, and that doing so will likely result in safer care. Future evaluations will
likely further the knowledge of how best to implement the practices to make them most effective.
However, in the meantime, our TEP believes that providers should not delay their consideration
of adopting these practices, as enough is known now to permit health care systems to move
forward.
Limitations
Because of limited resources and time, the current report does not cover the entire patient
safety field, which has grown exponentially since the last report, both in the number of potential
PSPs and in the amount of data about individual PSPs. For that reason, we used an explicit and
transparent process to select which PSPs to evaluate, and our final list of 41 (from the more than
150 candidates) included the PSPs we felt were of highest priority to policymakers and
providers.
Secondly, we did not perform in-depth reviews for all 41 PSPs. To maximize use of the
available time and resources, we tailored our methods to the needs of our stakeholders. In
particular, we targeted the 18 PSPs that were of the greatest interest to our stakeholders, or for
which we likely had the most new information for in-depth reviews. The remaining 23 PSPs
received brief reviews. It is important to note that the decisions about which PSPs would receive
which level of scrutiny and analysis were made by a broadly representative stakeholder
committee.
Thirdly, the in-depth reviews, although thorough, did not conform to all of the criteria for
conducting an evidence review as presented in the Institute of Medicines report, Finding What
Works in Health Care: Standards for Systematic Reviews,15 or to all the criteria in AHRQs
Methods Guide for Effectiveness and Comparative Effectiveness Reviews9; for example, we
did not publicly post a protocol for each of the individual reviews. We used our collective
experience as EPC team members to adapt existing EPC methods that best preserved the essence
of a systematic review, while allowing for the completion of 18 in-depth reviews within 9
months and within the available budget.
Additionally, over time, we will likely improve our methods for assessing evidence regarding
how patient safety interventions affect health care processes and outcomes. The methods we used
ES-13
for this report incorporate new perspectives regarding the importance of implementation and
context, which was the focus of the Context Sensitivity report; likewise, in the future, we
expect to increase our understanding of the interactions between multiple intervention,
implementation, and organizational variables and how the variables influence safety outcomes. If
future research reveals that these variables interact in ways that our current understanding of
theory and logic models cannot explain, we will need to modify the methods for evaluating PSPs
again.
Lastly, we relied on the judgment of our TEP at every important step of the project.
Therefore, the results are as much a product of these judgments as are our systematic review
methods. Hence, our results might be sensitive to the selection of particular experts on our TEP.
However, we mitigated this potential bias by including more than double the number of experts
on our TEP as we typically would for an EPC review, which allowed us to include a diverse set
of stakeholders from the U.S., Canada, and the United Kingdom. Stakeholders included PSP
developers and evaluators, patient safety policymakers, and experts in design and evaluation
methods. Rather than regarding the tight linkage between the needs of the stakeholders and the
work of the EPCs as a limitation, we view it as a strength that increases the likelihood that the
results of the review will be meaningful to providers, payors, and patients, and that the reports
results will lead to meaningful change.
ES-14
References
1.
2.
3.
4.
5.
6.
7.
ES-15
8.
9.
10.
11.
12.
13.
14.
Shekelle PG. Chapter 34. Effect of Nurse-toPatient Staffing Ratios on Patient Morbidity
and Mortality in Making Health Care Safer
II: An Updated Critical Analysis of the
Evidence for Patient Safety Practices.
Comparative Effectiveness Review No. 211.
(Prepared by the Southern California-RAND
Evidence-based Practice Center under
Contract No. 290-2007-10062-I.) AHRQ
Publication No. 13-E001-EF. Rockville,
ES-16
Part 1. Overview
Chapter 1. Introduction
The modern patient safety movement is believed by many to trace its origins to the 1999
publication of the groundbreaking report, To Err is Human by the Institute of Medicine.1 This
report, which highlighted the 44,000 to 98,000 deaths per year from medical errors in the United
States (U.S.) (the equivalent of the fatalities that would result from the crash of a jumbo jet a
day), galvanized the public and resulted in the focus, of widespread media and legislative
attention, for the first time, on the issue of patient safety. Parallel reports from other countries
were similarly influential.2
As part of its initial portfolio of patient safety activities, the Agency for Healthcare Research
and Quality (AHRQ) commissioned a team from the University of California, San Francisco
(UCSF)-Stanford University Evidence-Based Practice Center to analyze the evidence behind a
diverse group of patient safety practices (PSPs) in useor conceptualizedat that time. The
reportMaking Health Care Safer: A Critical Analysis of Patient Safety Practices2 (MHCS)
was published in 2001.3 The report analyzed nearly 80 different safety practices on several
dimensions, including potential impact, supporting evidence, and costs and complexity of
implementation. Based on these evidence reviews, practices were ultimately rated on both impact
and evidence, as well as prioritization for future research.
MHCS was immediately both influential and controversial. Several hundred thousand copies
of the report were distributed by AHRQ, and it became a lynchpin for other efforts (such as the
National Quality Forums Safe Practices list) to describe PSPs through the lens of evidencebased medicine. The controversy was generated by the reports rankings of PSPsin particular,
the relatively low rankings for certain popular practices such as computerized order entry
which raised fundamental questions about the role of evidence in assessing the value of quality
and safety practices, questions that continue to be debated to this day.4-7
In 2001, hospitals and health care organizations were under relatively little pressure to
implement safety practices. A decade later, the stakes have grown far higher.8 Regulators and
accreditors are pushing health care organizations to adopt various safe practices or to avoid
particular adverse events that are considered wholly or largely preventable. Many payers,
including the Centers for Medicare & Medicaid Services, have embedded patient safety into payfor-performance and no pay for errors initiatives. Billions of dollars and millions of personhours have been invested in a variety of efforts to improve safety, and virtually every health care
delivery organization now identifies patient safety as one of its top strategic priorities.
Yet the evidence indicates that our progress in eradicating medical errors has not matched the
efforts and financial resources invested in implementing PSPs. Studies of some practices that
have tremendous intuitive appeal, such as reducing resident duty hours and implementing rapid
response teams, have yielded conflicting results.9,10 Many examples of unintended consequences
of safety practices have emerged,11 and the successful implementation of safety practices has
been shown to be highly context dependent,12 often working effectively in some hospitals but not
others. Although a national initiative to improve safety in the United Kingdom found some
evidence of improvement, control hospitals improved as much as those that participated in a
vigorous intervention.13 Three recent U.S. studies have demonstrated continuing high rates of
preventable harm in hospitals;14-16 one of these studies showed evidence of no improvement in
adverse event rates from 2003 to 2008.14
Against this backdrop, AHRQ, believing that the time has come to re-examine the state of the
evidence supporting a wide variety of PSPs, commissioned a team led by investigators at RAND
Health, UCSF, and Johns Hopkins to reexamine the evidence behind key PSPs. Many of the
individuals engaged in this task participated in producing the original MHCS report, the MCHS
methodology formed the cornerstone of the present effort, and many of the practices examined
for this report were those previously reviewed in 2001. Thus, we see the present report as a
natural sequel to MHCS.
Because of the burgeoning literature relevant to patient safety and the limits of budget and
time, we selected a subset of PSPs to examine for this present report (chosen through methods
described in Chapter 2) rather than attempt, as we did in 2001, to review all PSPs. Moreover, the
maturation of the safety field has led to a deeper appreciation of the importance of context in
PSPs, a topic examined by our research team in our 2010 report, Assessing the Evidence for
Context-Sensitive Effectiveness and Safety of Patient Safety Practices: Developing Criteria.17
Accordingly, this report emphasizes matters of context and generalizability, as well as
unintended consequences, to a greater degree than did MHCS.
three groups above, but deserve their own designation here, as they may be the most frequent and
intense users of this report as they seek to improve patient safety at their own institutions.
Finally, while this volume is not primarily written for patients and their families, both groups
have become increasingly involved in patient safety efforts in a variety of ways. We welcome
such engagement and believe that patients, families, and their advocates can help advance efforts
to prevent harm.
A decade ago, our early enthusiasm for patient safety was accompanied by a hopeand
some magical thinkingthat finding solutions to medical errors would be relatively
straightforward. Simply adopt some techniques drawn from aviation and other safe industries,
build strong information technology systems, and improve culture, and, the hope went, patients
would immediately become safer in hospitals and clinics everywhere.
We now appreciate the naivety of this point of view. Making patients safe will require
ongoing efforts to improve practices, training, information technology, and culture. It will need
top-down resources and leadership, accompanied by bottom-up wisdom and innovation. It will
depend on a strong policy environment that creates appropriate incentives, while avoiding an
environment in which providers enthusiasm and creativity are sapped by an overly rigid,
prescriptive bureaucracy and set of rules.
While we have become more sophisticated about the challenges of keeping patients safe over
the past decade, the fundamentals have not changed: we need good and well-trained people,
armed with good data, operating under good policies, working under good leaders to do the right
things for patients. We hope this report contributes to these efforts by helping to identify those
right things.
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
Chapter 2. Methods
Topic Development
This topic was nominated by leaders of the Agency for Healthcare Research and Qualitys
Patient Safety Portfolio, part of the Center for Quality Improvement and Patient Safety.
The original goals of the project were stated as follows:
The analysis shall build on and expand upon earlier evidence
reports and current listing of Safe Practices by the National Quality
Forums (NQF) Safe Practices for Better Healthcare 2010
Update. The analysis shall focus on the collection of evidence of
the effectiveness of new safe practices that have been developed
but not included in the 2010 update, evidence of implementation of
current and new safe practices and the adoption of safe practices
by health care providers. This analysis shall include the review of
scientific literature, other appropriate analyses, and extensive peer
review of the draft report. The final report of this project will be
used by AHRQ for strategic planning in its patient safety portfolio
for future project development, implementation of safe practices.
The report will also be used by external organizations such and the
NQF, Joint Commission and others in their patient safety efforts.1
The preliminary Key Questions, pending topic refinement, were organized into three
categories.
What new patient safety practices (PSPs) have been developed since 2001 and/or are not
included in the NQF safe Practice list in 2010?
What is the nature of the safety practice i.e. clinical, organizational, or behavioral?
What is the intended risk that the practice is designed to prevent or mitigate?
Describe how the practice is a bundle of individual components or practices, if
applicable.
What is the intended setting for the practice, i.e., in patient, ambulatory, combination,
specialty, or clinical domain, and organizational setting?
What are the nature, quality, and weight of evidence of the practices effectiveness?
Describe how the practice has been sustained in its use after initial implementation.
Was there any external support for the implementation process, e.g., AHRQ technical
support, use by a collaborative, or quality improvement organization (QIO)?
Adoption/Diffusion
What is the extent to which the practice has been adopted by multiple institutions or
organizations outside the developing institution?
Was there any organized activity or program to support the diffusion of this innovation or
practice?
What, if any, evidence exists on the sustained use of the practice?
Has the practice become a requirement for use by any accreditation or credentialing
agency or organization?
Project Overview
An overview of the project is depicted in Figure 1. A key aspect of this project is the active
participation of a Technical Expert Panel (TEP) comprising a large number of patient safety
stakeholders and evaluation methods experts. We retained the participation of the TEP that had
participated in a prior AHRQ-supported project, Assessing the Evidence for Context-Sensitive
Effectiveness and Safety of Patient Safety Practices: Developing Criteria (hereafter referred to
as Context Sensitivity). The TEP comprised many of the key patient safety leaders in the
United States., Canada, and the United Kingdom, including experts in specific PSPs, as well as
experts in evaluation methods and people charged with implementing PSPs in hospitals and
clinics.
We divided the project into three phases: topic refinement, the evidence review, and the
critical review and interpretation of the evidence. The project team conducted the topic
refinement and the critical review and interpretation of the evidence jointly with the TEP; the
project team performed the evidence review.
Topic Refinement
Because the goals of the project were to assess the evidence of the effectiveness of new safe
practices and the evidence of implementation of currentsafe practices, practically all PSPs
were potentially eligible for inclusion in this review. Thus our first task was to refine the scope
of the topic to something that was achievable within the timeframe and budget for the project;
this task was undertaken by the project team and the TEP. Figure 1 presents an overview of how
this task was accomplished. We first compiled a list of potential PSPs for the review, starting
with the 79 topics in the MHCS report (2001)2 and adding practices from the National Quality
Forums 2010 Update, the Joint Commission, and the Leapfrog Group; practices identified in an
initial scoping search; and those suggested by our TEP. This effort resulted in an initial list of
158 potential PSPs (see Appendix A).
We then conducted an internal project team process that included amalgamation of some topics
and renaming of others, resulting in 96 PSPs. Internal project team triage resulted in our
identifying 35 PSPs we believed must be included, 48 PSPs about which we were unsure, and 13
that we believed could be excluded or folded into other PSPs that were on our include list
(Table 1). As indicated, we incorporated some of those 13 topics into other topics, such as the
monitoring topics. We excluded others that we judged to represent more of a quality issue than a
patient safety issue (such as pneumococcal vaccination interventions and regionalizing surgery to
high volume centers), whereas we judged others to be too late (warfarin interventions, in light of
the emergence of new oral anticoagulants) or too early in development (radio-frequency
identification [RFID] devices attached to wandering patients) for consideration.
Team Comment
Seems this could be part of a broader focus on patient
safety reporting systems.
This is likely to become less important in the future with a
move to non-coumadin-based anticoagulants such as
dabigatrin, which do not require the same degree of
monitoring.
We could include if the policy recommendation (i.e., the
intervention) was to implement this type of policy. This
became a safety practice when Leapfrog included it,
but its just as easy to argue that its quality rather than
safety.
Would bundle in preventing SSI.
We then sought input from our TEP about these decisions, offering them the opportunity to
change any of the include/exclude decisions, and asked for formal votes on the 49 PSPs
classified an Unsure.
This effort resulted in 48 PSPs judged to be of highest priority in terms of the need for an
evidence review of effectiveness, implementation, or adoption, still too large a number of topics
to review comprehensively within the given timeframe. Therefore, we asked our TEP to assess
whether breadth or depth was likely to be more valuable for stakeholdersin other words,
we asked whether the review should focus on fewer topics in more detail or cover all topics but
in less detail. Our TEP recommended a hybrid approach in which some topics would be
reviewed in depth, whereas other topics would receive only a brief review. Topics could be
considered to need only a brief review for several reasons: the PSP is already well-established;
stakeholders need to know only whats new? since the last time this topic was reviewed in
depth; new evidence suggests the PSP may not be as effective as originally believed, so it is no
longer a priority safety practice to implement; or it is an emerging PSP with limited evidence yet
accumulated about it.
For each of the 48 topics, we then solicited formal input from our TEP about the need for an
in-depth review, a brief review, or no review at all. Table 2 presents the results in terms of the
proportion of TEP members who recommended a topic undergo an in-depth review, a brief
review, or no review at all. We designated all topics that received 50 percent or greater support
for an in-depth review to be reviewed in depth; all other topics were designated for brief reviews.
No topic on the list received 50 percent or greater support for no review at all. The list underwent
further modification, as some PSPs originally designated as separate topics were judged to be
sufficiently similar to be covered together in one review; examples included the topics related to
transitions in care and those related to monitoring.
A final set of modifications to this scope occurred during the course of the reviews.
Our PSP topic on pressure ulcers was modified to focus solely on implementation, as an
EPC review of the effectiveness of pressure ulcer prevention interventions is currently
underway.
We combined the topics, diagnostic errors and notification of test results to patients
into a single in-depth review.
The body of literature on simulation methods was sufficiently large that we treated it as
an in-depth review.
The review topics were then divided among the participating EPCs. Weekly teleconference
calls and email were used to promote common practices in the review process.
Table 2, Chapter 2. Proportion of technical expert panelists expressing a preference for the level
of evidence review for each PSP
PSP*
In-Depth
In-Depth
Handoff - (Transitions in care)
Medication reconciliation
Rapid response teams
Fall prevention strategies and
interventions to reduce the use of
restraints
Diagnostic errors - meta-cognition,
computerized decision support
Protocols for notification of test results to
patients
Geriatric/delirium programs
Monitoring for patient safety problems
Preventing ventilator-associated
pneumonia
Pressure ulcer prevention
Promoting a culture of safety
Universal protocol/preoperative checklist
(surgical safety)
Report cards/outcomes measurement
like NSQIP (surgical safety)
Nurse staffing patterns and ratios
Other interventions targeting improved
transitions in care (Transitions in care)
Intensive insulin therapy for glycemic
control
Use of preoperative anesthesia
checklists (Complications due to
anesthesia equipment failures)
Protocols for high risk drugs, e.g.,
nomograms for heparin
Interventions to prevent contrast-induced
renal failure
The patients role in preventing errors
Brief Review
No Review
79%
71%
67%
21%
29%
20%
0%
0%
13%
64%
29%
7%
64%
21%
14%
64%
29%
7%
64%
57%
7%
36%
29%
7%
57%
21%
21%
57%
53%
36%
33%
7%
13%
50%
43%
7%
50%
43%
7%
50%
36%
14%
50%
36%
14%
50%
36%
14%
50%
29%
21%
50%
29%
21%
50%
36%
14%
50%
21%
29%
Table 2, Chapter 2. Proportion of technical expert panelists expressing a preference for the level
of evidence review for each PSP (continued)
PSP*
In-Depth
Brief Review
No Review
Human factors as a general topic,
50%
13%
38%
focus still to be more precisely defined
Brief Review
CPOE and clinical decision support
47%
47%
7%
systems (CDSS)
Bundles and checklists as a general
47%
40%
13%
strategy (not just for specific indications)
Simulator-based training
46%
31%
23%
Prevention of surgical items left inside
43%
50%
7%
patient (surgical safety)
Medication administration
43%
50%
7%
Display systems
43%
43%
14%
Hand washing + interventions to improve
36%
50%
14%
hand washing compliance
Perioperative beta- blockers
36%
57%
7%
VTE prophylaxis and methods for
36%
50%
14%
implementation
Team training/team practices
36%
43%
21%
Limiting individual providers hours of
36%
50%
14%
service
Smart pumps and other protocols for
36%
50%
14%
infusion pumps
Device-related strategies for preventing
36%
43%
21%
tubing misconnections
Clinical pharmacist consultation services
36%
43%
21%
Prevention of nosocomial UTIs
33%
53%
13%
Use of real-time ultrasound guidance
33%
60%
7%
during central line insertion
Patient understanding/informed consent
29%
36%
36%
(possibly includes health literacy)
Interventions for central venous catheter29%
50%
21%
related blood infections
Patient death or serious injury associated
with prolonged fluoroscopy with
29%
36%
36%
cumulative dose
Death among surgical patients with
serious treatable complications (failure to
29%
36%
36%
rescue)
Barrier precautions, patient isolation,
routine surveillance for patients at
21%
71%
7%
admission
Identifying patients at risk for suicide
21%
64%
14%
Sign your site protocols - potentially
14%
79%
7%
part of checklists
Processes related to reprocessing
14%
50%
36%
single-use medical devices
Do not use abbreviations, acronyms,
symbols, and dose designation
14%
79%
7%
campaign
Ensure documentation of patients
7%
50%
43%
preferences for life-sustaining treatment
Strategies to prevent stress-related
7%
50%
43%
gastrointestinal bleeding
*The topic titles listed in this table were the exact titles the TEP considered in their decision-making; some of these PSP topics or
titles underwent further revisions to their final title between that assessment and this final report.
10
The principal results of the Context-Sensitivity project included the following key points.
Whereas controlled trials of PSP implementations offer investigators greater control of
sources of systematic error than do observational studies, trials often are not feasible, in
terms of time or resources. Also, controlled trials are often not possible for PSPs
11
These principles guided our search for evidence and the way we present our findings in this
report (see Table 3).
12
In-Depth Reviews
Many of the 18 topics designated for an in-depth review were likely to have been the subject
of a previous systematic review; thus, the review process usually began with a search to identify
existing systematic reviews. To assess their potential utility, we followed the procedures
proposed by Whitlock and colleagues5 which essentially meant addressing the following two
questions:
Is the existing review sufficiently on topic to be of use? and
Is it of sufficient quality for us to have confidence in the results?
13
Assessment of whether a review was sufficiently on topic was a subjective judgment based
on the patients-intervention-comparators-outcomes-timeframe (PICOT) focus of the existing
review. To assess the quality of the systematic review, we, in general, used the AMSTAR criteria
(see Appendix B).6 If an existing systematic review was judged to be sufficiently on topic and
of acceptable quality, then based on that review, the following searches were undertaken:
A full update search, in which databases were searched for new evidence published since
the end date of the search in the existing systematic review; and/or
A search for signals for updating, according to the criteria proposed by Shojania and
colleagues,7 which involved a search of high-yield databases and journals for pivotal
studies whose results might be a signal that a systematic review is out-of-date.
Based on the results of these searches, the existing review was supplemented with newer
evidence or considered to be up-to-date.
Any evidence identified via the update search or the signals search was added to the
evidence base from the existing systematic review.
For some topics, no systematic review could be identified, or those that were identified were
either not sufficiently relevant or not of sufficient quality to be used. In those situations, new
searches were done using guidance as outlined in the EPC Methods Guide.8
As indicated above, evidence about context, implementation, and adoption are key aspects of
this review. We searched for evidence on these topics in two ways:
We looked for and extracted data about contexts and implementation from the articles
contributing to the evidence of effectiveness;
We identified implementation studies from our literature searches. Implementation
studies focus on the implementation process, especially those elements of the implementation
demonstrated or believed to be of particular importance for the success, or lack of success, of the
intervention. To be eligible, implementation studies needed to either report, or be linked to
reports of, effectiveness outcomes.
Brief Reviews
Brief reviews are explicitly not full systematic reviews or updates. The goals of the brief
reviews varied by PSP, according to the needs of stakeholders. The assessment could focus
primarily on information about effectiveness of an emerging PSP or implementation of an
established PSP; alternatively, the review could explore whether new evidence calls into
question the effectiveness of an existing PSP. Thus, the methods used to conduct the brief
reviews varied according to the various goals of the reviews. . However, in general, brief reviews
were conducted by an expert in the topic in collaboration with the project team, and involved
focused literature searches for evidence relevant to the specific need. This evidence was then
narratively summarized in a format that also varied with the particular goal.
14
but instead relied on the criteria originally chosen for that review, for example the criteria of the
U.S. Preventive Services Task Force.
Implementation studies were not assessed for their quality, as we lacked evidence or expert
opinion about the criteria for such an assessment.
Use of observational study designs of stronger internal validity (controlled before-and after, time
series, statistical process control) (+1)
+4 = High
+3 = Moderate
+2 = Low
If evidence does not permit a conclusion then the strength of evidence = insufficient
16
and the presence of implementation tools such as written implementation materials or training
manuals.
For the question of implementation difficulty, we use three categories: difficult for PSPs that
required large scale organizational change; not difficult for PSPs that required protocols for
drugs or devices such as those to reduce radiation exposure or to help prevent stress-related
gastrointestinal bleeding; and moderate for PSPs falling between the extremes.
18
19
future research needs were highest priority, and classified as high priority those topics receiving
more than 50 percent support.
References
1.
2.
3.
4.
5.
20
6.
7.
8.
9.
10.
11.
12.
21
Brief reviews use a different format, that varies somewhat depending on the topic. The general
format for brief reviews is: What is (are) the patient safety practice(s)?; How has the patient
safety practices been implemented?; What have we learned about the practice(s)? Brief update
reviews are topics that were covered in Making Health Care Safer 2001 and use a format
designed for reader to identify whats new since then.
References. Each chapter is individually referenced for convenience.
22
23
Weight-based nomograms use actual patient body weight to calculate an optimum dose that
is patient-specific. In contrast, physician dosing without nomograms often does not account
accurately for patient characteristics.
This current report systematically reviewed the literature to identify effectiveness studies of
any intervention with a goal to reduce adverse events related to intravenous heparin in the
inpatient setting that had a comparison group and was not a qualitative study. Since this PSP is
currently most often conceptualized as focusing on intravenous administration as the most highrisk route, we did not include subcutaneous or oral anticoagulant administration in this review.
Intravenous anticoagulants are particularly high risk because dosing is complex and the
therapeutic range is particularly narrow. This narrow range increases the opportunity for harm.6-9
Although bleeding can occur even at therapeutic doses of heparin, it is much more likely when
the dose is excessive or inadequately monitored. Unfractionated heparin, which is given
intravenously, is widely used as the drug of choice for a variety of clinical conditions where
rapid and closely monitored anticoagulation is needed, such as acute coronary syndromes.10
However, since the 2001 publication of Making Health Care Safer, low-molecular-weight
heparinswhich have a less complex dosing regimen, are given subcutaneously, and have been
shown to have equivalent efficacy for many indicationshave widely replaced unfractionated
heparin for some clinical conditions such as venous thromboembolism (VTE) prophylaxis.
A wide variety of safety practices are recommended to increase patient safety for intravenous
anticoagulants in general. These practices include limiting the number and dosage of high alert
drugs prescribed (to ensure that only patients who are most likely to benefit receive the
medications or that lower-risk options are used whenever possible), having independent system
checks and balances in place to identify and prevent dosing errors, and having a transparent error
reporting system to aid in the development and implementation of system changes.1,2 Other
practices include removing high-alert medications from nursing units and floor stock,
standardizing medication doses, using single doses or pre-mixed solutions, labeling different
strength solutions clearly to avoid mixups (e.g., Heplock packaging), provider education and
drug-administration protocols and decision support tools that involve double-checking of the
drug and dosing, pump-setting, and dosage.4 Health information technology tools may help
reduce errors associated with high-alert medications by preventing significant overdoses (e.g.
tenfold errors in dosing) and verifying that the correct medication is being administered.11,12
However, the level of effectiveness of health information technology may vary.13,14 Specific
heparin patient-safety practices reviewed here include dosing nomograms and weight-based
dosing interventions, with and without the use of health information technology tools as part of
the intervention.
24
25
telemetry drug library increased from 56.5 percent (SD: 1.5%) pre-intervention to 72.1 percent
(SD: 2.1%) post-intervention (p<0.001), and the number of telemetry manual pump edits
decreased (56.9 [SD: 12.8] to 14.7 [SD: 3.9]; p<0.001).13 Finally, Fanikos and colleagues
assessed the impact of a smart infusion device with a hospital-determined drug library and
programmable software on anticoagulation errors using a pre-post design. After reviewing a total
of 14,012 administered doses of heparin in 3,674 patients, the software generated a total of 501
heparin alerts in 246 patients. No significant difference in anticoagulation errors was found as a
result of the intervention (49 pre- vs. 48 post-intervention).14
Table 1, Chapter 3. Summary tableheparin effectiveness studies
Author, Year
17
Study
Design
Description of PSP
Baird, 2001
Pre-post
Fanikos,
14
2007
Fraipont,
18
2003
Pre-post
Pre-post
Oyen, 2005
19
13
Prusch, 2011 Intelligent infusion devices (IIDs), bar-codeassisted medication administration system,
and electronic medication administration
record system
15
Toth, 2002
Pre-post
Pre-post
Outcomes: Benefits
Dosing and time to
anticoagulation: No statistics
reported
Anticoagulation medication
errors: No significant differences
Time to therapeutic
anticoagulation, complications:
not significant
Therapeutic anticoagulation
significantly improved,
complications not reported
Telemetry drug library monthly
compliance and manual pump
edits: Statistically significant
improvement
RCT
Total complications,
overanticoagulation, time to
anticoagulation all statistically
significant improvement
Pre-post
Time to anticoagulation
significant; complications not
significant
In terms of evidence grading, the strength of evidence for this topic was low. Risk of bias
was high due to study design issues: Only one study was an RCT.14 Results were inconsistent,
with half of the studies reporting no statistically significant findings; several studies were too
small to measure outcomes meaningfully. Many studies did not report patient safety outcomes,
but instead reported the outcomes for process measures such as time to therapeutic
anticoagulation or compliance with a drug library; many studies that did report complications or
errors did not have sufficient sample size. Finally, regarding precision, a number of different
outcome measures were used, so no conclusions could be made (see Evidence Table on risk of
bias in Appendix D).
26
27
Finally, we identified one additional study which did not meet our inclusion criteria for
reporting effectiveness data but took a broader human factors approach to improving heparin
safety by improving administration. Harder and colleagues evaluated the human factors
associated with improving the safety of heparin administration.22 After completing interviews
with the staff, the authors offered suggestions for improving the heparin administration process
in order to make the computerized heparin dosing interface more user-friendly (e.g.,
automatically converting English and metric measurements.) Iterative refinements were made to
the system after the initial modifications, and an educational program was rolled out to inform
providers about the new heparin administration process.22
Low
Evidence or
Potential for
Harmful
Unintended
Consequences
Low-to-moderate
28
Estimate of
Cost
Low
Implementation Issues:
How Much do We
Know?/How Hard Is it?
Little/Moderate
References
1.
2.
3.
4.
5.
6.
29
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
30
Introduction
In our original report, Making Health Care Safer 2001, Kaushal and Bates noted that over
770,000 people were harmed or died in hospitals annually from adverse drug events (ADE),1-4
with incidence rates in hospital-based studies ranging from 2 to 7 per 100 admissions.1,5-7 In the
outpatient setting, as they also noted, one study on adults estimated the ADE incidence rate at 3
percent.8 The purpose of this review is to update the data on the incidence of ADEs in hospital
settings and to review measures aimed at preventing these events, including the role of the
clinical pharmacist. We searched the literature from 2001 to 2011 and included studies most
relevant to clinical pharmacist interventions on medication errors and adverse drug events in
various health care settings. Our focus was on studies that to some degree addressed the possible
association between clinical pharmacist activities and improved prescribing practices and/or
assessed whether such activities might lead to reduced medication errors and adverse drug
events.
31
32
providing drug information, patient counseling, precepting, toxicology case assistance and
various forms of therapeutic consultations, interventions and managements, including medication
error prevention (though included studies were limited in this latter regard).
By and large, these reviews support clinical pharmacist activities in improving medication
management. In general, three issues emerge from the literature. First, clinical pharmacists are
engaged in a multitude of patient level activities, including recognizing, intercepting, and
documenting drug-related problems, as well as assisting in optimizing pharmaceutical choices
for patients and, in some cases, engaging in specific interventions or in specific disease
management practices. Second, it is problematic to accurately capture all that pharmacists do at
either an individual patient level or at an organization level,20 which makes it that much more
difficult to assess their impact, especially since clinical pharmacists do not work in isolation but
rather with other clinicians and, frequently, within hospitals or health care systems or settings.
Third, studies that attempt to show the benefit of pharmacists engaged in various activities from
a larger vantage point (e.g., assessing whether adding a pharmacist to a ward team reduces
medication errors or adverse drug events) often have challenges in their interpretation, including
lack of concurrent control groups, indeterminate definitions of suboptimal prescribing, varying
definitions of medication errors and preventable adverse drug events, different methods of error
and event capture and reporting, and varying clinical outcome assessments. Even so, while
individual studies do not always demonstrate benefits from an organizational perspective, the
body of work suggests that pharmacists provide substantial value to patient care, clinical teams,
institutions, and health care organizations.
Kaushal, 2008
Population and
Controls
Pediatric ICU or
general ward with
paper charting;
matched units did
not receive
intervention
Intervention
Part or full-time
clinical pharmacist
rounding and
monitoring drug
dispensing, storage,
and administration
33
Outcomes
Measured and
Timing
Medication errors
and adverse events
pre/post, identified
by nurse and
reviewed by 2
blinded physician
reviewers; 6-8 weeks
baseline, 3-month
intervention period
Findings
Full-time clinical
pharmacist
decreased
medication errors
(29 to 6 per 1000
patient days);
increase in
medication errors in
controls; part-time
pharmacists did not
decrease error rate.
Wang, 2007
Population and
Controls
Intervention
22
Pediatrics unit of a
community teaching
hospital
Addition of CPOE to
existing clinical
pharmacist system
23
Inclusion of clinical
pharmacist in
rounding
Clinically important
drug-drug
interactions pre/post
over a 10-week
period
Cardiac ICU
Rounding and
participation in
patient-oriented
activities (e.g., taking
medication histories,
discharge
counseling), and
provider level
activities (e.g., giving
in-service talks to
house staff and
communicating with
physician and
nursing staff)
Pharmacist testing/
recommendations
regarding patients on
antipsychotics who
had movement
disorder complaints
or who were taking
drugs to counter
movement disorders
Medication error
interventions (e.g.,
dose or medication
changes, missing
medications, allergydrug
contraindications)
pre/post over 5 years
Pharmacist-run
education program
on medication orders
and IV fluid review
implemented at
month 4 of 12
months plus other
process changes
Medication errors
pre/post; case
finding by incident
reporting
Rivkin, 2011
24
LaPointe, 2003
25
Stoner, 2000
Simpson, 2004
Outpatient
psychiatric setting
(235 sets of
evaluations in 83
patients on antipsychotics)
26
Neonatal ICU
34
Outcomes
Measured and
Timing
Medication errors,
near misses, and
adverse events over
a 3-month period
Movement disorder
(extrapyramidal)
symptoms
Findings
Clinical pharmacist
intercepted 78% of
111 potentially
serious prescribing
errors but none of 32
harmful
administrative errors
and few of the
transcribing (6/25) or
monitoring errors
(3/7)
Drug interaction
rates decreased
significantly ( 65%)
when compared
retrospectively
(historically) to a 10week period earlier
in the year
Incidence of
medication errors
increased from
around 15 to nearly
24 per 100
admissions, and a
higher trend was
seen during times of
house staff transition
A majority of
recommendations
(82% of 130
evaluations) were
followed by
clinicians; of these,
93% led to a
resolution or
reduction in
extrapyramidal
symptoms
Significant decrease
in medication errors
(from 24 to 5 per
1,000 neonatal
activity days/month);
error rate increased
during summer
staffing change
Population and
Controls
Intervention
Outcomes
Measured and
Timing
Adverse drug
reactions (ADRs)
Bond, 2006
27
584 hospitals
encompassing
>35,000 Medicare
patient stays
Pharmacy staffing
and presence or
absence of various
pharmacy services
Bond, 2007
28
14 different clinical
pharmacy services
and several staffing
models
Severity-adjusted
mortality rates
Review of
medication orders
and identification of
errors via
retrospective review
by an independent
reviewer.
Pharmacists also
documented their
interventions.
Medication Errors, 1
month when
pharmacist was not
present to check
medication orders
versus 1 month
when pharmacist (s)
was (were) present;
time periods for
assessment were
one year apart
Brown, 2008
29
35
Findings
Pharmacist
involvement in 8
services (in-service
education, drug
information services,
adverse drug
reaction
management, drug
protocol
management,
cardiopulmonary
resuscitation teams,
medical rounds and
completing
admission drug
histories) as well as
higher staffing rates
decreased ADRs;
however, pharmacist
participation in total
parenteral nutrition
teams increased
ADRs
In-service education,
drug information,
adverse drug
reaction monitoring;
participation in drug
protocol
management,
cardiopulmonary
resuscitation teams
and medical rounds;
and completing
admission drug
histories were
associated with
reduced mortality as
were two staffing
variables
Pre-post analysis
showed significant
decrease (66.6%)
from error rates of
approximately 16 to
5 per one hundred
medications orders
Rothschild, 2010
Cesarz, 2012
31
Population and
Controls
30
Intervention
Four academic
Emergency
Departments
Observational study
in which pharmacy
residents conducted
226 sessions (787
hours) of observing
pharmacist activities;
the study included
over 17,000
medications ordered
or administered to
nearly 6,500 patients
An academic
medical centers 32bed Emergency
Department, serving
pediatric and adult
populations
Prospective
observational study
looking at activities
of four pharmacists
during relevant shifts
in reviewing
discharge
prescriptions. Data
collection was over a
3 week period and
used standardized
forms for reporting
interventions. All
recommendations
were provided to the
ordering physician
who made the
determination to
change a
prescription
Outcomes
Measured and
Timing
Identification of
medication errors at
various stages of
prescribing or
administration by
unblinded,
continuous
observation. Data
collection was via
templated forms.
Captured elements
included errors of
interest, ranging
from those
intercepted before
reaching the patient
to caught after
reaching the patient
but before harm
could occur to
ameliorated adverse
events (collectively
these together were
known as recovered
medication errors) .
Case reviewers
independently
assessed suspected
error interventions.
Self-report of
interventions on
discharge
prescriptions. An
independent
reviewer determined
whether the
intervention was
categorized as error
prevention or
therapeutic
optimization
Findings
Pharmacists
identified over 500
recovered
medication errors,
with an overall rate
of about 3 per 100
medications or about
8 per 100 patients.
Approximately 90%
were intercepted
before reaching the
patient.
Of 674 discharge
prescriptions
reviewed, ED
pharmacists
intervened on about
10%; roughly half of
the 68 interventions
(54%) concerned
error prevention.
A number of the studies contained design flaws that prevented ruling out the contribution of
other process modifications or even secular changes to the observed results. Nevertheless,
overall, these newer studies continue to support the important roles of clinical pharmacists in
reducing prescribing mishaps as well as in improving several patient-level outcomes in various
36
settings. With the exception of one study, studies in which pharmacists participated in a greater
number of clinical processes seemed to show stronger effects.
37
medication errors and preventable ADEs. The existing data for other inpatient and for outpatient
care settings are also supportive of a role for pharmacists but less robust than in intensive care
units. Data from nursing homes are not as clear as for other settings, but, logically speaking,
since medication and prescribing errors occur in this setting, and patients are elderly and more
prone to polypharmacy, it is likely by analogy that drug safety in nursing homes will be
improved by clinical pharmacist interventions. Similarly, evidence from emergency departments
is limited but given the high intensity of care activities and of prescription utilization, it is logical
that benefits will accrue from pharmacist interventions. More and better designed studies should
help determine the magnitude of the benefit(s), to the extent that such benefits exist, in various
health care settings. A summary table is located in Table 2, Chapter 4.
Table 2, Chapter 4. Summary table
Scope of the
Strength of
Problem Targeted by Evidence for
the PSP
Effectiveness
(Frequency/Severity) of the PSPs
Common/Low
Moderate-tohigh
Evidence or
Potential for
Harmful
Unintended
Consequences
Low
Estimate of
Cost
High
Implementation Issues:
How Much do We
Know?/How Hard Is it?
Little/Moderate
References
1.
2.
3.
4.
38
5.
6.
7.
8.
9.
10.
11.
12.
Ried LD, McKenna DA, Horn JR. Metaanalysis of research on the effect of clinical
pharmacokinetics services on therapeutic
drug monitoring. Am J Hosp Pharm. 1989
May;46(5):945-51. PMID 2729302.
13.
14.
15.
16.
17.
18.
19.
39
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
32.
30.
33.
31.
40
Introduction
Medication errors stem from a variety of causes, including miscommunication between
prescribers and pharmacists in the form of misunderstood and/or illegible abbreviations. The
potential hazards of certain abbreviations started receiving heightened attention approximately
twenty years ago.1 Most notably, as one of its National Patient Safety Goals, the then named
Joint Commission on Accreditation of Healthcare Organizations (JCAHO, hereinafter referred to
as the Joint Commission for consistency) in 2003 announced that nine abbreviations and/or
shorthand notationsa Do Not Use list--should be banned in its accredited hospitals by April
2004.2,3 The list included the following inappropriate abbreviations: U or u instead of unit;
IU instead of International Unit; Q.D. or similar instead of once daily; Q.O.D or similar
instead of every other day, MS, MSO4 and MgSO4 instead of writing morphine sulfate or
magnesium sulfate; and use of zeros, either when trailing an ordinal number (1.0 instead of 1) or
lack of a zero before a decimal point (.9 instead of 0.9)2,4 (See Figure 1).
Figure 1, Chapter 5. Official do not use list
41
42
Unfortunately, in all of these areas, the relevant United States literature is sparse, and
implementation efforts have had mixed results.
43
44
system. The older, locally derived system had automatic conversion of inappropriate
abbreviations installed on some computers; it also allowed for free text entries on the ordering
template. It had minimal clinical decision support and did not send prescriptions directly to
pharmacies. The newer system had a commercially available clinical decision support package,
but did not auto-correct abbreviations. The system was able to send prescriptions to pharmacies.
The newer system included two alerts to providers when they entered and completed a
prescription containing an inappropriate abbreviation. In this yearlong study, data were available
on seventeen physicians in the academically affiliated clinic. Rates of inappropriate
abbreviations (per 100 prescriptions) fell from about 24 at baseline to just under 11 at 6 months
and then to approximately 6 at 1 year after implementation (p-values < 0.001). Interestingly,
non-abbreviation error rates rose at 12 weeks, but were similar at one year postimplementation.15
45
these systems and the cost and effort is not insubstantial to newly utilizing electronic prescribing.
Another alternative would be enforcing a zero tolerance policy on handwritten prescriptions and
medication orders. However, this might create a substantial burden for prescribers and
pharmacists, particularly in the outpatient and retail pharmacy areas, not to mention mail out
facilities. In the meantime, a low-cost approach of implementation, such as through ongoing
education and/or feedback, focused on avoiding selected harmful abbreviations whenever and
wherever possible seems reasonable and feasible. A summary table is located at Table 1, Chapter
5.
Table 1, Chapter 5. Summary table
Scope of the
Strength of
Problem Targeted by Evidence for
the PSP
Effectiveness
(Frequency/Severity) of the PSPs
Common/Low
Low
Evidence or
Potential for
Harmful
Unintended
Consequences
Negligible
Estimate of
Cost
Low
Implementation Issues:
How Much do We
Know?/How Hard Is it?
References
1.
2.
3.
4.
5.
6.
7.
46
8.
9.
10.
11.
12.
13.
14.
15.
47
Introduction
Medication errors represent a serious issue affecting the U.S. health care system, accounting
for the largest category of patient safety incidents within the larger category of medical errors.
One report estimated that at least 1.5 million preventable medication errors occur in the U.S.
each year.1 A list of high-alert medications (those with the highest potential for patient harm if
used in error) published by the Institute for Safe Medication Practices (ISMP) includes several
medications delivered by intravenous (IV) infusion (e.g., insulin, propofol, heparin).2
Because IV delivery is more rapid and leads to higher systemic concentrations of drugs
compared with other delivery methods, adverse drug effects tend to be more rapid and severe
when associated with IV infusion. Because traditional infusion pumps are typically programmed
in milliliters per hour (mL/hr) and volume-to-be-infused (VTBI) in mL, they are particularly
vulnerable to errors in drug administration and monitoring.1 Such errors include administration
of the wrong dose or the wrong drug as well as erroneous infusion to the wrong patient.
48
volume manually).8 A significant drawback is that inappropriately programmed hard alerts may
impede delivery of care, and circumvention of hard alerts can lead to serious errors.9
Smart pumps with DERS plus BPOC can additionally prevent drug delivery to the wrong
patients.10,11 As long as users comply with such alerts and prompts, smart pumps have the
potential to reduce the number of infusion errors. Compliance with safety features can be
improved by programming prompts that increase ease of use, and by emphasizing a culture of
safety within the organization. Smart pumps also contain a data log that can be used to identify
programming errors or show that the pump prevented adverse events.9
However, the basic limitation of smart pumps is that they can correct only errors of
administration; other types of medical errors can occur during ordering or prescribing,
dispensing, transcribing, and monitoring of patient response.10 For this reason, smart pumps
function best not as standalone devices but when integrated into a larger medication safety
system that connects them with computerized provider order entry (CPOE), BPOC, and
electronic medication administration records (eMARs).9 Such interconnected systems can target
not only errors of administration but also errors of ordering, dispensing, and transcription.10
49
dosage limits may have explained the lack of advantage of smart pump decision support in this
study.13 This study used an early version of smart pump technology that was opt-in rather than
opt-out, which made it easier for users to skip the library rather than look for it.
Larsen et al. (2005) performed a retrospective before-after study in pediatric patients that
compared medication infusion errors 12 months before and 12 months after adopting a new
protocol using a combination of smart pumps, standard drug concentrations, and humanengineered (user-friendly) medication labels. The smart pumps included a modifiable drug
library and provided soft alerts to users who attempted to use doses that exceeded the safety
limits. The infusion error rate dropped from 3.1 to 0.8 per 1000 doses from the pre-intervention
to the post-intervention period, a risk reduction of 2.3 (95% CI 1.1-3.4, P <0.001).14 However,
since this was a combination of three interventions, it is unclear what percentage of the error
reduction can be attributed to smart pumps alone. Data were obtained from the hospital-wideincident-reporting system, which tends to underreport errors, but the reported pre- and postintervention error rates should be representative of the relative number of errors.14
Adachi and Lodolce (2005) conducted a retrospective before-after study (one year preintervention, one year post-intervention) to determine whether a new intervention (revised
standard order sets and smart pumps with soft alerts) could reduce IV dosing and administration
errors. Although they found that only a small reduction occurred in overall dosing errors (59 to
46), a larger reduction occurred in pump-related errors (24 to 10, or from 41% to 22% of dosing
errors). Standard concentrations eliminated errors related to the wrong drug concentration. Nine
out of the 10 post-intervention pump programming errors occurred because users did not use the
pump software.15
Three uncontrolled studies illustrate compliance issues associated with smart pump soft
alerts. Eckel et al. (2006) reported a high frequency of programmings (44.4%) due to users
bypassing the drug library when selecting a drug. Furthermore, users overrode 88.5% of soft
alerts.16 Fields and Peterman (2005) reported 506 medication errors due to users overriding soft
alerts.17 However, a third study (Breland 2010) reported that a community hospital was able to
improve compliance with pump alerts from 33% (when smart pumps were first introduced) to
97% three years later.18
50
Smart Pumps With Soft and Hard Alerts Plus Barcode Technology
Trbovich et al. (2010) conducted a simulation study comparing nurses ability to avoid
medication errors using a traditional pump, a smart pump, and a pump with an integrated bar
code scanner (the latter two had soft and hard alerts). The study was conducted in a laboratory
setting using patient mannequins with bar-coded wristbands and medication bags with bar-coded
labels containing patient ID; errors were assessed by type. Wrong drug errors did not differ
significantly by pump type. Patient ID errors were remedied by significantly more nurses using
pumps with barcode scanners (88%) than with the smart pumps without barcode scanners (58%)
or traditional pumps (46%). Significantly more nurses remedied critical overdose errors when
using pumps with barcode scanners (79%) and smart pumps without barcode scanners (75%) due
to hard alerts than with traditional pumps (38%). Wrong dose soft alerts did not result in
significant differences in fixing overdose errors among different pumps (errors remedied by 75%
of nurses using pumps with barcode scanners, 63% with smart pumps without barcode scanners,
and 50% with traditional pumps). This was because many nurses overrode soft alerts.7 While this
study provides perspectives on error rates, it does not faithfully simulate a clinical environment:
auto-programming in a clinical setting is limited at this time but is typically accomplished
through interfaces with BPOC systems instead of through printing medication labels with patient
ID.
Smart Pumps With Soft and Hard Alerts Integrated With Barcode
Technology and eMARs
Prusch et al. (2011) conducted a prospective before-after study evaluating a program
integrating intelligent infusion devices (IIDs) with a BPOC system and an eMAR system.21
Monthly compliance with the telemetry drug library increased from 56.5% pre to 72.1% post
intervention (p<0.001) and the number of telemetry manual pump edits decreased (56.9 to 14.7;
p<0.001). Pump programming errors related to i.v. unfractionated heparin occurred at a rate of
16.9 events/10,000 opportunities pre-implementation and 11.3 events /10,000 opportunities postimplementation, but the rate decrease was not statistically significant (P = 0.17). However, smart
pumps were used before and after the implementation period, the only difference being that the
smart pumps became fully integrated with BPOC and eMAR in the post-implementation period.
Therefore, the true impact of smart pumps on infusion error rates is unclear from this study.
None of the studies described above identified harms to patients that could be attributed
specifically to the use of smart pumps in place of traditional infusion pumps.
51
Ownership of the process at the executive level (assessment of culture and budget
resources, forming a multidisciplinary team, performing a Failure Mode and Effects
Analysis [FMEA] to identify barriers to compliance)
Technological readiness (ensure that information technology [IT] systems can interface
with pumps and that IT staff levels are sufficient, update drug libraries and download
medication safety information efficiently [preferably via a wireless network], consider
wireless network communication upgrade if it is unavailable prior to smart pump
implementation)
Physical environment and equipment (ensure sufficient number of pumps, policies for
cleaning, storage, and distribution, short-term pump rental from outside vendors [if
necessary], ensure rental pumps are programmed with the renting facilitys drug library
and dose limits, ensure sufficient number of electrical outlets for pump operation in
patient areas and for recharging internal batteries when not in use)
Staff education (plan for several weeks of staff education, train super-users, ensure
ongoing education, explain purpose of and procedures for soft and hard stops, inform
staff about drug library updates, develop champions in each clinical area devoted to
safety culture, do smart-pump simulation exercises, emphasize benefits of smart pump
technology)
Specialized patient care areas (make plans to address needs of specific therapies or
patient care areas such as pediatrics/nursery, pain management, operating room,
oncology, emergency department, and patient transport)
Vendor support (to help define implementation timetable, provide sample drug libraries,
online tutorials, live telephone assistance, post-implementation follow-up visits,
assistance in data evaluation, and external support groups)
Rollout (prioritize sequence of patient care areas receiving pumps, select areas with
adequate staff and resources, select educators and champions from pilot units, vendor
support should be available, evaluate rollout process)8
Creation of safe and effective customized drug libraries is essential for proper utilization of
smart pumps. Institutions must evaluate their clinical practice when determining what drugs and
dosage limits to select for their library. Drug libraries should at least include all high-alert drugs
with standard concentrations as well as soft and hard stops for various dosage limits. Once drug
libraries have been developed, considerable time must also be devoted to maintaining and
updating the libraries. Wireless communication technology in an organizations infrastructure
allows easier adjustment or updating of drug libraries, which otherwise would require manually
updating each pump separately.8
Breland (2010) reported that a community hospital was able to increase compliance rates
with pump alerts from 33% at baseline (when smart pumps were first introduced) to 97% three
years later. This was done by having nursing directors and managers stress the importance of the
safety software and how it could improve patient safety. Compliance data were shared with staff
nurses and unannounced twice-weekly inspections were performed by pharmacy to determine
why safety software was not being used in individual cases. Continual reeducation and
customization of drug libraries for the needs of specific critical care areas (CCAs) also helped to
improve compliance. Compliance rates for individual CCAs were distributed to nursing
directors, who also emphasized to the staff the legal liability entailed in noncompliance. In
addition, a review of edits and overrides led to a drug library revision to eliminate unnecessary
52
alerts by changing some dosage limits to reflect actual dosing practices (which were determined
to be safe).18
Low
Evidence or
Potential for
Harmful
Unintended
Consequences
Low
Estimate of
Cost
Moderate
Implementation Issues:
How Much do We
Know?/How Hard Is it?
Moderate/Moderate
References
1.
4.
2.
5.
3.
6.
53
7.
8.
9.
10.
11.
12.
13.
14.
54
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
Introduction
Healthcare-associated infections are linked to high morbidity, mortality, and costs
worldwide. In 2002, an estimated 1.7 million healthcare-associated infections were seen in U.S.
hospitals, resulting in approximately 99,000 deaths.1 In 2005, 18,650 patients with methicillinresistant Staphylococcus aureus (MRSA) died, more than the number of Americans who died
from HIV/AIDS in that same year.2 In 2007, Clostridium difficile was ranked among the 20
leading causes of mortality among Americans over 65 years of age.3 Despite decades of infection
control interventions, health care-associated infections continue to be a major burden on U.S.
hospitals.4
Currently, there is a rising wave of new emergent healthcare-associated infections, including
multi-drug resistant strains of Acinetobacter baumannii and Klebsiella pneumoniae.
Additionally, reports of vancomycin-resistant S. aureus have appeared sporadically across the
Nation.5-7 No effective antibiotics are available for some strains of these pathogens, and few new
antibiotics are in the developmental pipeline. For example, since 2007, only two new antibiotics
have been developed. Thus, prevention, not treatment, is the most sustainable strategy to control
health care-associated infections.
55
This update review focuses on what we have learned about infection prevention measures
and their effectiveness since the publication of the original report. We conducted a search of the
health care and health services literature for the time interval 2001 to 2011 and reviewed all
studies relevant to this topic.
56
methicillin-resistant S. aureus infections increased rapidly with a high mortality rate.2,7 However,
since 2007 rates of healthcare-associated MRSA have begun to decline.14 Second, communityassociated MRSA infections caused by the USA300 clone emerged between 1999 and 2001.15
USA300 MRSA has caused severe infections in previously healthy people with no prior contact
with the health care system, thus alarming both health care professionals and the general
public.15 Additionally, USA300 MRSA infections have not replaced healthcare-associated
MRSA infections (e.g. USA100), rather they have occurred as a separate epidemic leading to an
increasing number of MRSA infections.16 Third, isolated cases of vancomycin-resistant S. aureus
(VRSA), first recognized in 2002, have led to fears that failure to control VRE and MRSA
transmission may lead to a new epidemic of VRSA, which will be very difficult to treat.5-7
57
58
institutions, while the two studies with negative results only assessed their interventions in ICUs
or surgical wards. Each study implemented a unique bundle in which the only common factor in
all four bundles was active surveillance plus contact precautions. For example, both the Harbarth
and Robicsek included nasal decolonization while the other two studies did not. The studies also
varied in how their laboratory testing was performed. For example, in the Veterans Health
Administration study, surveillance samples were tested at the local clinical microbiology
laboratory. In contrast, in the Star*ICU study, all surveillance samples were mailed to the
Clinical Microbiology Laboratory of the National Institutes of Health Clinical Center.
Interestingly, when comparing all four of these studies, the studies with negative results had
stronger study designs.
The studies above assessed active surveillance among ICU patients. Admission to the ICU is
a large risk factor for healthcare-associated infections, therefore, it may be cost-effective to
target only ICU patients for active surveillance rather than the entire hospital.38 The high cost of
active surveillance has led to multiple cohort studies with the goal of establishing a rule to
predict which patients are at high risk for MRSA or VRE colonization.39 A prediction rule would
help infection prevention staff determine which patients are likely to carry MRSA or VRE and,
thus, could transmit MRSA or VRE to other patients or could acquire an MRSA or VRE
infection. Ideally, screening the patients identified as high risk of colonization would be more
cost-effective and take less time than testing all patients for MRSA or VRE using traditional
active surveillance. Many prediction rules include recent admission to the hospital, which is a
strong predictor of MRSA and VRE colonization, with sensitivities ranging from 44% to 77%
and specificities ranging from 46% to 98%.38,40-45 Prediction rules have also included risk factors
for colonization such as prior operation, hemodialysis, prior history of MRSA or VRE, transfer
from long-term care facility, age, antimicrobial use during the past year, and a current wound. If
these prediction rules were applied, the proportion of MRSA or VRE colonized patients who
would be missed ranged from 15% to 43%.38,40-45 Thus, current prediction rules have had varying
success.
Similarly, three studies have created prediction rules to predict patients at high risk for C.
difficile infection.46,47 The first prediction rule included age, C. difficile infection pressure, recent
admission to the hospital, severity of illness score, days of high-risk antibiotic use, low albumin
level, ICU admission, and receipt of laxatives, gastric acid suppressors or antimotility drugs.46
The second rule only included the Waterlow score, a nursing tool routinely used to assess a
patients risk of developing a pressure ulcer.47 The third rule included age, hemodialysis and
length of ICU stay.48 The sensitivity of the C. difficile infection prediction rules ranged from
60% to 70% and the specificity ranged from 89% to 95%.46,47
Horizontal approaches to infection control could utilize contact precautions without the use
of expensive laboratory surveillance tests. A single ICU, quasi-experimental study of a bundle
which included universal contact precautions found that not only did this bundle stop an outbreak
of multidrug-resistant Acinetobacter baumannii, it also led to a decrease in MRSA acquisition
from 14% to 10%, and VRE acquisition from 21% to 9%.11 Two quasi-experimental studies
compared universal gloving (wearing a new pair of gloves for each patient) to active surveillance
and contact precautions in a single ICU.49,50 Active surveillance and contact precautions included
VRE and MRSA surveillance cultures on admission and every 4 days with contact precautions
for patients colonized or infected with VRE or MRSA. Both studies found no difference in
MRSA or VRE colonization no matter which intervention was implemented. However, one study
59
found an increase in nosocomial infection rates during the universal glove period, potentially due
to decreased compliance with hand hygiene after removal of gloves.49
Another horizontal approach would be to place patients at high risk for acquiring a
healthcare-associated infection under pre-emptive contact precautions.51,52 One ICU found that
their intubated patients were eight times more likely to acquire healthcare-associated MRSA
compared with non-ventilated patients, thus they performed a quasi-experimental study to assess
an intervention where all intubated patients were placed under pre-emptive contact precautions.
In the first phase of the study, active surveillance for MRSA was performed at ICU admission
and weekly with contact precautions for MRSA positive patients. In the second phase of the
study, active surveillance and contact precautions for MRSA remained, however all intubated
patients were also placed on contact precautions. This study found a decrease in healthcareassociated MRSA infections for both intubated patients (p=0.02) and in all ICU patients
(p<0.05).52
Less is known about optimal methods to prevent C. difficile transmission compared with
VRE and MRSA.53 Most studies of C. difficile prevention are simple quasi-experimental studies
that test a bundled intervention. Multiple recommendations and guidelines suggest contact
isolation for symptomatic C. difficile infected patients only.17,53,54 Contact isolation for C.
difficile infected patients should include single rooms with private toilets if possible.17 According
to the SHEA/IDSA Expert Panel, the only two approaches to preventing C. difficile with good
evidence to support them are wearing gloves when caring for an infected patient and
antimicrobial stewardship.17,54 No data currently support isolating asymptomatic C. difficile
carriers.53,54 An unresolved issue is whether to place symptomatic patients with a history of C.
difficile infection under contact precautions.17
60
The cost-effectiveness studies estimated that the cost of active surveillance and contact
isolation strategies for MRSA or VRE to range from $1,913 to $10,545 per month.63,65,66 The
mathematical model found that the average cost of targeted active surveillance of high risk
patients ranged from $4,100 to $12,508 per infection adverted depending on MRSA prevalence
and screening test used, while the average cost of universal active surveillance ranged from
$5,799 to $21,195 per infection adverted.67 When these costs were itemized, 13% to 99% of the
total cost was spent on specimen collection and laboratory testing while the remaining proportion
was spent on isolation (e.g., gowns, gloves, nurse time to don gowns and gloves).63,65-67 The vast
differences in these proportions were due to how labor costs were accounted for. Studies varied
as to how they assessed the cost of laboratory technologists, cost of nursing time to collect
swabs, and cost of nursing time to don and remove gowns and gloves.
Although cost-effectiveness analyses have not been performed for universal contact
precautions to prevent healthcare-associated infections, an analysis by Wenzel et al., compared
the cost-effectiveness of active surveillance and contact precautions for MRSA to a populationbased infection control approach. This analysis assumed that active surveillance for MRSA
would cost approximately $600,000 while the population-based approach would cost
approximately $300,000. If the active surveillance program reduced MRSA infections by 50%
and the population-based approach reduced healthcare-associated infections by 50%, then the
active surveillance program would save $245 million to $980 million nationally while the
population-based intervention would save $1.75 billion to $7 billion nationally.26
As with all health care interventions, health care worker support and implementation of the
intervention is necessary for the intervention to be successful. The STAR*ICU trial noted
suboptimal implementation of their interventions. That study demonstrated that when contact
precautions were specified, gloves were used for 82% of contacts, gowns for 77% of contacts,
and hand hygiene was performed after gloves were removed for 69% of contacts. Additionally,
when universal gloving was specified, gloves were used for 72% of contacts and hand hygiene
was performed after gloves were removed for 62% of contacts.13 The Veterans Health
Administrations MRSA initiative includes a dedicated MRSA coordinator at each acute care
hospital responsible for implementation of the initiative. From the beginning of the initiative in
2007 to the end of the study period in 2010, compliance with surveillance nasal screening for
MRSA increased, with the percentage of patients who were screened at admission rising from
82% to 96%, and the percentage who were screened at transfer or discharge rising from 72% to
93%. However adherence to contact precautions was not reported.24 Two studies by Bearman
and colleagues found that observed compliance was higher during a universal glove intervention
compared with observed compliance with contact precautions (gowns and gloves) during an
active surveillance plus contact precaution intervention. However, the studies found conflicting
results as to when hand hygiene compliance was greater. The first study found that the active
surveillance and contact precautions intervention was associated with greater compliance with
hand hygiene compared with hand hygiene compliance during the universal gloving
intervention.49 The second study, which included hand hygiene in-service trainings, found that
compliance with hand hygiene was higher during the universal gloving phase compared with the
active surveillance and contact precautions phase.50
Upcoming Studies
Of late, two multicenter cluster-randomized trials of contact precautions have been
implemented. The Cluster Randomized Trial of Hospitals to Assess Impact of Targeted versus
61
Universal Strategies to Reduce MRSA in ICUs (REDUCE MRSA trial) recently finished
collecting data on the effectiveness of the following strategies: (1) MRSA active surveillance of
ICU admissions, followed by contact isolation if positive, (2) MRSA active surveillance of ICU
admissions followed by nasal decolonization if positive, and (3) universal nasal decolonization
of ICU admissions without screening.68 The Benefits of Universal Glove and Gowning Study
(BUGG Study) is currently comparing the effectiveness of universal contact precautions to
standard of care in multiple ICUs in order to determine whether universal gowns and gloves
decrease the overall burden of healthcare-associated pathogens in the ICU setting. The results of
these studies should be available soon and will add to the growing body of evidence on
interventions to control healthcare-associated infections.12
Moderate
Evidence or
Potential for
Harmful
Unintended
Consequences
Moderate
(isolation of
patients)
Estimate of
Cost
Moderate-tohigh
Implementation Issues:
How Much do We
Know?/How Hard Is it?
Moderate/Moderate
References
1.
2.
62
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
63
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
64
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
50.
51.
52.
53.
54.
55.
47.
56.
48.
57.
49.
58.
65
59.
60.
61.
62.
63.
64.
66
65.
66.
67.
68.
69.
Introduction
Healthcare-associated infections account for approximately 80,000 deaths per year in the
United States.1-3 A worldwide systematic review found that the incidence of healthcareassociated infections ranged from 1.7 to 23.6 per 100 patients. Hospital costs directly related to
healthcare-associated infections ranged from $28.4 to $33.8 billion in 2007 U.S. dollars.4 Yet
these infections are frequently preventable through hand hygiene.
Substantial epidemiologic evidence supports that hand hygiene reduces the transmission of
healthcare-associated pathogens and the incidence of health-care associated infections.5 The link
between hand hygiene and improvements in healthcare-associated infections is hard to prove
definitively in modern-day health care. However, the importance of hand hygiene is universally
acknowledged by organizations such as the Joint Commission, World Health Organization
(WHO) and Centers for Disease Control (CDC), which recommend or require hand hygiene
practices and interventions to improve hand hygiene compliance in order to reduce health careacquired infections.5-7 This review will therefore focus on interventions to improve compliance
with hand hygiene, rather than on the efficacy of hand hygiene for reducing healthcareassociated infections.
Compliance with hand hygiene practices among health care workers has historically been
very low, averaging 39 percent.5 The review on hand hygiene compliance and interventions
aimed at improving it that was conducted for the original 2001 Making Health Care Safer
report found that poor compliance has been documented in studies across hospital unit types and
in various other settings. Workers tend to underestimate the importance of compliance and often
overestimate their compliance with hand hygiene procedures.1 The report concluded that future
research studies needed to identify reasons for poor compliance and design sustainable
interventions that target these factors. The aim of this review is to assess the evidence for the
impact of interventions on hand hygiene compliance since that report.
67
Monitoring health care workers compliance with hand hygiene practices is vital for
evaluating whether interventions are successful. WHO recommends using a validated
methodology for training observers to directly monitor hand-hygiene using My five moments
for hand hygiene.5 Other methods for monitoring include patient-observations, measuring of
hand hygiene product consumption (either by volume of product used or through electronic
counting devices), and electronic hand hygiene compliance monitoring systems (e.g. real-time
location systems, dedicated monitoring systems or video monitoring).8
Hand hygiene interventions include both single and multi-level interventions. These
interventions include staff and/or patient education and involvement, feedback initiatives,
cultural change, organizational change, social marketing, additional sinks or alcohol dispensers,
or a combination of the above.1,9
Advocates of hand-hygiene improvement interventions recommend that multimodal
interventions are needed to induce sustained hand-hygiene practice improvements, and should be
based on theories of behavior change. On the individual level, the intervention should target
provider education and motivation regarding hand-hygiene practices; on the interpersonal level,
patient empowerment and cues to action should reinforce proper hand-hygiene practices; and on
the organization level, organizational structure and philosophy needs to be supportive of proper
practices.5
69
solution availability, education, and multifaceted interventions, including feedback. Ten studies
found a statistically significant increase in compliance, and four did not. Three studies evaluated
longer-term results and found that compliance rates decreased after the intervention ended.1
Patient Engagement
A 2011 review by McGuckin and colleagues found evidence of the importance of patient
engagement or empowerment and multi-model strategies in hand-hygiene interventions. The
authors found that patient empowerment comprised patient participation, knowledge, skills, and
a facilitating environment for their participation in hand hygiene. The majority of patients agreed
that they would ask their health care workers to wash their hands (80% to 90%), especially if
encouraged to ask. However, the authors found scarce literature on the efficacy of patient
empowerment interventions to improve health care worker hand hygiene and were unable to
conduct a traditional evidence-based review.17
70
waning over the long term. A recent systematic review focusing on higher quality evidence
found only four studies, three of which showed a significant impact. Another recent review
found mixed results for the impact of hand hygiene interventions on rates of healthcareassociated infections. A variety of interventions to improve hand hygiene are being implemented
and promoted by various U.S. and international organizations, particularly educational programs,
monitoring, and feedback. Interventions should be multimodal, addressing providers
knowledge, attitudes, and beliefs regarding hand hygiene, as well as strategies for behavioral
change, and should ideally be tailored to institutional needs as well as different provider groups
and health care situations. Health care administrators embarking on a hand hygiene intervention
should take advantage of the tools developed by the CDC and the WHO. New strategies, such as
patient engagement in hand-hygiene interventions, are an emerging area with only a few studies
assessing their effectiveness, and need further research on how best to implement them
effectively. Finally, research may be directed toward understanding the effectiveness of specific
elements of hand hygiene interventions, and the context in which they are implemented, in order
to understand which combinations lead most reliably to success. A summary table is located in
Table 1, Chapter 8.
Table 1, Chapter 8. Summary table
Scope of the
Strength of
Problem Targeted by Evidence for
the PSP
Effectiveness
(Frequency/Severity) of the PSPs
Common/Moderate
Low
Evidence or
Potential for
Harmful
Unintended
Consequences
Low
Estimate of
Cost
Low
Implementation Issues:
How Much do We
Know?/How Hard Is it?
Moderate/Moderate
References
1.
2.
3.
71
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
Institute for Healthcare Improvement. Howto Guide: Improving Hand Hygiene. 2006.
www.ihi.org/knowledge/Pages/ToolsHowto
GuideImprovingHandHygiene.aspx
Accessed November 16, 2011.
14.
15.
72
16.
17.
Introduction
Urinary tract infection has long been considered the most common healthcare-associated
infection (HAI), with the vast majority of these infections occurring after placement of the
convenient, often unnecessary,1-3 and easily forgotten urinary catheter.4 With an estimated one
million catheter-associated urinary tract infections5 (CAUTIs) per year, associated with an
additional cost of $676 per admission (or $2836 when complicated by bacteremia),6 it is not
surprising that CAUTIs were among the first hospital-acquired conditions selected for nonpayment by Medicare as of October 2008,7 and have been further targeted for complete
elimination8 as a never event, with a national goal to reduce CAUTI by 25% and reduce
urinary catheter use by 50% by 2014.9,10 These national initiatives renewed public and research
interest in the prevention of CAUTI, prompting updates of several comprehensive guidelines11-14
and reviews of strategies to prevent CAUTI released since the 2001 Making Health Care Safer
report.15
73
18
Meddings J, Saint S. Disrupting the Life Cycle of the Urinary Catheter. Clin Infect Dis. 2011; 52(11): 1291-3 by permission of
Oxford University Press.
Because avoiding unnecessary urinary catheter use is the most important goal in prevention
of CAUTI, this chapter reviews the evidence on two types of interventions that target
unnecessary urinary catheter use: (1) protocols and interventions to decrease unnecessary
placement of urinary catheters (catheter lifecycle stage 1), and (2) interventions that prompt
removal of unnecessary urinary catheters (catheter lifecycle stage 3).
The evidence summarized in this chapter was generated using a literature search conducted for a
prior systematic review and meta-analysis19 along with a focused update of the published peerreviewed literature (from August 2008 to February 2012) through a MEDLINE search for
intervention studies to reduce use of unnecessary urinary catheters in the acute care of adults. A
CINAHL database search was also performed for interventions developed and implemented by
nurses related to urinary catheter use. Studies were included if at least one outcome involving
catheter use or CAUTI events (Table 1) was reported as a result of the intervention, and with a
comparison group (either pre- vs. post-intervention or a separate control group).
74
19
Table 1, Chapter 9. Description of outcomes evaluated (adapted from the prior meta-analysis )
Number of CAUTI episodes per 1,000 catheter-days was recorded and a rate ratio was
calculated to compare pre- vs. post-intervention. When rates of both asymptomatic and
20
symptomatic CAUTI were reported separately, the rates of symptomatic CAUTI were used
19
for the meta-analysis.
Cumulative risk of CAUTI during hospitalization (i.e., the percentage of patients who
developed CAUTI) was also extracted for each study, and a risk ratio was calculated to
19
compare risks before and after the intervention for the meta-analysis.
Mean number of days of urinary catheter use per patient was recorded before and after
the intervention, and a standardized mean difference (SMD) was calculated to compare the
19
two groups for the meta-analysis.
Measures of Urinary Percentage of patient days in which the catheter was in place was calculated before
Catheter Use
and after the intervention, and a standardized mean difference (SMD) was determined for
19
each study for the meta-analysis.
Daily catheter prevalence, defined as the number of patients with catheters in place
during a specific time period, is reported for some of the more recent studies.
Need for Catheter
Re-catheterization need was extracted as the number and percent of patients who
Replacement
required replacement of a catheter after prior removal of an indwelling catheter.
The table in Appendix D summarizes the intervention studies described in this review, including study designs, patient
populations, and the interventions employed to avoid unnecessary catheter placement or to prompt catheter removal.
Meddings J, Rogers MA, Macy M, et al. Systematic review and meta-analysis: reminder systems to reduce catheter-associated
urinary tract infections and urinary catheter use in hospitalized patients. Clin Infect Dis. 2010;51(5):550-60 by permission of
Oxford University Press.
Measures of
Catheter-Associated
Urinary Tract
Infection (Cauti)
Development
75
Table 2, Chapter 9. Indications for indwelling urethral catheter use (from the 2009 CDCs
11
guideline )
A. Examples of Appropriate Indications for Indwelling Urethral Catheter Use
Patient has acute urinary retention or bladder outlet obstruction
Need for accurate measurements of urinary output in critically ill patients
Perioperative use for selected surgical procedures:
Patients undergoing urologic or other surgery on contiguous structures of genitourinary tract
Anticipated prolonged surgery duration; catheters inserted for this reason should be removed in post-anesthesia
care unit
Patients anticipated to receive large-volume infusions or diuretics during surgery
Need for intraoperative monitoring of urinary output
To assist in healing of open sacral or perineal wounds in incontinent patients
Patient requires prolonged immobilization (e.g., potentially unstable thoracic or lumbar spine, multiple traumatic
injuries such as pelvic fractures)
To improve comfort for end of life care if needed
B. Examples of Inappropriate Uses of Indwelling Catheters
As a substitute for nursing care of the patient or resident with incontinence
As a means to obtain urine for culture or other diagnostic tests when patient can voluntarily void
For prolonged postoperative duration without appropriate indications (e.g., structural repair of urethra or contiguous
structures, prolonged effect of epidural anesthesia, etc.)
The technology required to implement catheter placement restrictions ranges from low
technology strategies such as a hospital or unit policy on appropriate catheter placement or preprinted catheter orders with limited indications to higher technology strategies such as
computerized orders22,23,25 for catheter placement. Catheter restriction protocols have been a
common component of successful multi-modal interventions to decrease catheter use and/or
CAUTI rates, including hospital-wide23 interventions and interventions tailored for specific
environments such as the emergency department,21,26 inpatient units17 (including general
medical25,27,28-surgical29 wards and ICU29-33), and in the peri-procedural32 setting. Urinary
retention protocols22,28,29,32-34 are a type of catheter restriction protocols that often incorporate the
use of a portable bladder ultrasound22,28,32,34,35 to verify retention prior to catheterization, and
recommend use of intermittent catheterization rather than indwelling catheters to manage a
common and often temporary issue.
Reminder interventions can be generated by nurses, physicians or electronic order sets, and can
be targeted to remind either nurses or physicians about the catheter. Some reminder interventions
have employed nurses dedicated to detecting unnecessary catheters.23,35 Reminder interventions
can also serve to remind clinicians of appropriate catheter indications.
Unfortunately, reminder interventions can also be easy to ignore43 and catheters may remain
in place without action. The next type of intervention to prompt removal of unnecessary
catheters, which goes a step further, is a stop order that requires action. Stop orders prompt the
clinician (either nurse or physician) to remove the catheter by default after a certain time period
has elapsed or condition has occurred, unless the catheter remains clinically appropriate. For
example, catheter stop orders can be configured to expire in the same fashion as restraint or
antibiotic orders, unless action is taken by a clinician. Stop orders directed at
physicians23,25,28,30,42 require an order to be renewed or discontinued on the basis of review at
specific intervals, such as every 24 to 48 hours after admission or post-procedure. Stop orders
directed at nurses either require the nurse to obtain a catheter removal order from
physicians,27,32,46 or can empower nurses to remove the catheter without requesting a physician
order20,28,30,34,47-49 on the basis of an appropriate indication list. Admittedly, implementing a
nurse-empowered catheter removal protocol may be less effective than anticipated, as early
qualitative research of nurse-empowered interventions indicate some nurses are uncomfortable
with this autonomy49 and might not remove catheters as expected.
Stop order
2.8
5.6
11.2
2.0
4.1
8.2
77
Overall
2.6 (95%CI, 1.63.6)
5.2 (95%CI, 3.2-7.2)
10.4 (95%CI, 6.4-14.4)
This meta-analysis19 also suggested that the mean duration of urinary catheterization
decreased by 37 percent, with 2.61 fewer days of catheterization per patient in the intervention
vs. control groups. The pooled standardized mean difference (SMD) in the duration of
catheterization was -1.11 overall (p=0.070); a statistically significant decrease in duration was
observed in studies that used a stop order (SMD -0.30; p=0.001) but not in those that used only a
reminder intervention (SMD -1.54; p=0.071).19 An update of the literature review since this
meta-analysis yielded 12 additional studies with reminder and/or stop order interventions. Figure
2 illustrates the major findings of the 14 studies for catheter use and CAUTI events as reported in
the prior meta-analysis;19 Figure 3 illustrates the major findings for the 12 subsequent studies,
including eight that reported measures of catheter use, and eight that reported CAUTI events.
78
Figure 2, Chapter 9. Summary of CAUTI and urinary catheter outcomes from 14 studies
Note: Summary comes from the 14 studies20,25,28-30,33,36-41,43,45 included in the 2010 meta-analysis.19
*Difference of p<0.05 reported between intervention and comparison group.
79
Figure 3, Chapter 9. Summary of CAUTI and urinary catheter outcomes from 12 additional studies
Note: Summary comes from 12 additional studies23,27,31,32,34,35,42,44,46-49 since the prior meta-analysis.19
*Difference of p<0.05 reported between intervention and comparison group.
80
generated by a research nurse to remind physicians which of their inpatients had urinary
catheters,43 found that the intervention was either cost-neutral or modestly cost-saving depending
on the assumptions made. More recently, a study35 of five hospitals in the Netherlands employed
a multi-modal intervention including reminders in four hospitals, and a stop order in the fifth
hospital. The program was found to be cost-saving, with the mean amount saved being 537 (or
~$700) per 100 hospitalized patients.
82
and/or a desire to avoid the inconveniences and increased frequency of patient contact required
for the care of incontinence and catheter alternatives. Other nurses report they simply do not feel
comfortable49 removing the catheter without explicit orders from the physician, which is ironic
considering that many nurses place catheters without orders. Nursing comfort with catheter
removal can be increased49 with peer support and education, and may be facilitated by directly
addressing the workload concerns associated with the removal of indwelling catheters. Indeed, a
survey of nurses27 during implementation of a nurse-empowered catheter removal protocol
indicated increased nursing and patient satisfaction, despite the expected increase in workload.
Even though CAUTI is a very common healthcare-associated infection, Krein and colleagues
reported that CAUTI preventive practice use is lagging behind efforts to prevent central lineassociated bloodstream infection and ventilator-associated pneumonia,54 with room for
improvement in adopting catheter removal and CAUTI preventive strategies demonstrated again
in two recent large surveys of hospitals55 and ICUs.56 Fortunately, many resources exist
(www.Catheterout.org) to help hospitals develop and implement programs to decrease catheter
use and prevent CAUTI, including a range of tools and educational materials to address
implementation challenges. Hospital and unit-level leadership also play a key role in preventing
infection.57
83
hospital-acquired infection in the United States, with some unique but not impossible challenges
for implementation. Furthermore, reducing indwelling catheter use addresses the noninfectious
complications of urinary catheter use such as catheter-related patient discomfort and immobility
(Table 4).
Table 4, Chapter 9. Summary table
Scope of the
Strength of
Problem Targeted by Evidence for
the PSP
Effectiveness
(Frequency/Severity) of the PSPs
Common/Moderate
Moderate-tohigh
Evidence or
Potential for
Harmful
Unintended
Consequences
Low
Estimate of
Cost
Low
Implementation Issues:
How Much do We
Know?/How Hard Is it?
Moderate/Moderate
References
1.
2.
3.
4.
5.
6.
7.
84
8.
9.
10.
11.
12.
13.
14.
23.
15.
24.
25.
26.
27.
28.
29.
30.
31.
16.
17.
18.
19.
20.
21.
22.
85
32.
41.
33.
42.
34.
43.
44.
45.
46.
47.
48.
49.
50.
35.
36.
37.
38.
Apisarnthanarak A, Thongphubeth K,
Sirinvaravong S, et al. Effectiveness of
multifaceted hospitalwide quality
improvement programs featuring an
intervention to remove unnecessary urinary
catheters at a tertiary care center in
Thailand. Infect Control Hosp Epidemiol.
2007;28(7):791-8. PMID 17564980.
39.
40.
86
51.
52.
53.
54.
55.
56.
57.
58.
87
59.
60.
Introduction
Central venous catheters (CVCs) are intravascular access devices that terminate within the
great vessels of the neck (superior or inferior vena cava, brachiocephalic veins, subclavian vein
or internal jugular vein), or a site proximal to the heart. CVCs are vital for the care of
hospitalized and critically ill patients, as they provide reliable venous access for clinical activities
such as blood sampling, infusion of medications, and hemodynamic measurement. However,
CVCs are also the leading cause of healthcare-associated bloodstream infections (BSIs) and are
frequently implicated in life-threatening illnesses.1,2 Infections associated with CVCs are
categorized in the literature as either central-line associated bloodstream infection (CLABSI),
or catheter-related bloodstream infection (CRBSI), based on whether surveillance or
ascertainment of infection is the desired goal. For instance, the Centers for Disease Control and
Preventions (CDC) National Healthcare Safety Network (NHSN) uses the CLABSI definition
for surveillance purposes, defining the term as a laboratory confirmed BSI in any patient with a
CVC present either at the time of, or within a 48-hour period before the detection of infection.3,4
Thus, the CDC-NSHN definition overestimates the true incidence of CRBSI, as some BSIs may
be due to infection at other sites (e.g., pneumonia or urinary tract infection) or at sites that are
difficult to detect (e.g., translocation from the gastrointestinal tract or mucositis following
chemotherapy). In contrast, CRBSI is a more precise and rigorous definition that requires either
(a) isolation of the same organism from the catheter and the peripheral blood, (b) simultaneous
quantitative blood cultures with a ratio of 5:1 or higher of those from the indwelling CVC
compared with peripheral blood, or (c) a differential time to positivity of CVC-derived versus
(vs.) peripheral blood culture positivity of more than 2 hours.5 The CRBSI definition is thus
largely used within the context of clinical care and research, whereas the term, CLABSI, is
implemented for epidemiologic surveillance.6 For the purposes of this review, we use the term
CLABSI to encompass both of these operational definitions.
Of the approximately 249,000 BSIs that occur in U.S. hospitals each year, 80,000 (32.2%)
occur in intensive care unit (ICU) settings.2 Because CVCs are more frequently used in ICUs
than in other areas of the hospital, and the strongest predictor of developing a BSI is the presence
of a CVC, the epidemiology of CLABSI has traditionally focused on the critically ill. With over
15 million catheter days in ICUs annually, the potential impact of CLABSI is substantial in this
population alone.6,7 However, in a survey of major medical centers, CVC use was identified in
24.4 percent of patients outside the ICU.8 Thus, millions of patients both in and out of ICU
settings are potentially at risk of developing CLABSI. Although the frequency of CLABSI
outside the ICU is largely unknown, Weber and colleagues found that the incidence of CLABSI
decreased when patients transitioned from ICUs to step-down units or non-ICU floors.9 Data
from the CDC-NHSN also suggest lower CLABSI rates in patients on hospital wards compared
with those in an ICU setting.10 Furthermore, recent evidence suggests that the incidence of
CLABSI in ICUs is significantly lesser than reported in 2001, likely due to a number of efforts
aimed at preventing this infection.11 These efforts notwithstanding, the increasing use of CVCs
88
such as peripherally inserted central catheters (PICCs) outside of ICUs may reflect an important
shift in the epidemiology of CLABSI to non-ICU settings.12 This change is highly relevant, as
lack of a uniform patient-care team and absence of comprehensive surveillance efforts in nonICU settings represent substantial obstacles to addressing CLABSI in these areas.
The economic burden of CLABSI is substantial. A recent analysis estimated that each
CLABSI episode independently increases length of hospitalization from 7 to 21 days, and adds
an attributable cost of about $37,000 (2002 dollars) per patient.13 The annual national cost of
caring for patients who develop CLABSI is estimated to range from $0.67 to $2.68 billion.13-15
Similar trends exist in European nations, where the incremental expenditure related to CLABSI
is estimated at 9,154 (18,241 [$24,558 in 2012 dollars vs. 9,087 [$12233]) per patient.16
Given this clinical and economic cost, investigators, policy-makers, and regulatory agencies in
the U.S. and abroad have devoted great efforts to curtail CLABSI over the past decade.17-19
CLABSIs are potentially preventable through the use of evidence-based practices.20 The
original Making Health Care Safer report examined the prevalence, strategies, and costs
associated with CLABSI prevention, and found that certain practices (e.g., the use of maximal
sterile precautions) were associated with both a decrease in CLABSI risk and reduced cost,
whereas others (e.g., intravenous antimicrobial prophylaxis) added expense without clear
benefit.21,22 In this review, we provide an update to the original report by highlighting the most
clinically and cost effective strategies associated with CLABSI prevention. To compile this
report, we performed a systematic review of the literature and searched multiple databases to
identify relevant studies published between 2000 and 2012 using terms such as Bacteremia,
Catheterization, Central Venous, and central line-associated bloodstream infection. Our
search strategy yielded a total of 1,087 unique manuscripts of which 337 articles were relevant
for this report.
89
Table 1, Chapter 10. Categories and recommendations for CLABSI reduction practices from the Healthcare Infection Control Practices
*
Advisory Committee of the Centers for Disease Control and Prevention
Recommendation
Hand hygiene prior to catheter
insertion
All inclusive catheter carts or kits
Chlorhexidine cleansing
CVC dressing
Chlorhexidine sponge dressing
Description
Decontaminate hands with either antiseptic-containing soaps or alcohol-based
gels/foams before inserting, repairing, replacing, or dressing a CVC
A catheter kit or cart contains all the equipment necessary for CVC insertion
(needle, guidewire, introducers, etc.), and ensures sterility by minimzing
interruptions during line placement
Use a cap, mask, sterile gown, sterile gloves, and a sterile full body drape when
inserting CVCs and PICCs or performing guidewire exchange(s)
Prepare clean skin with chlorhexidine preparation with alcohol before CVC
insertion and during dressing changes
Chlorhexidine/silver sulfadiazine or minocycline/rifampin-impregnated CVCs are
recommended only if the catheter is expected to remain in place 5 days or more
AND the CVC will be inserted in an environment where the CLABSI rate remains
high despite a comprehensive reduction strategy
Whenever possible, use the subclavian site, rather than the jugular or femoral
sites in adults
Minimize contamination risk by scrubbing the access site with an appropriate
antiseptic (chlorhexidine, povidone iodine, or 70% alcohol) prior to accessing the
CVC
Daily evaluation and prompt removal of CVCs that are no longer clinically
warranted is an important aspect of CLABSI prevention; routine replacement of
CVCs, PICCs, or hemodialysis catheters is not recommended
Daily cleansing using a 2% chlorhexidine solution or impregnated washcloth
rather than soap and water in ICU- and hemodialysis patients is recommended
Use either sterile gauze or sterile, transparent, semipermeable dressing to cover
the CVC site
The use of chlorhexidine-impregnated sponge dressings is recommended for
patients >2 months of age if the CLABSI rate is not decreasing despite
adherence to basic prevention measures, including education and training,
appropriate use of chlorhexidine for skin antisepsis and use of maximal sterile
barrier precautions
Topical antibiotic use may promote fungemia or bacteremia in non-dialysis
populations and is recommended only for hemodialysis catheter dressing
Instillation of supra-physiologic doses of an intravenous antibiotic or anti-infective
solution into a catheter lumen between periods of CVC access is recommended
only in those at high baseline risk for CLABSI
The use of oral or intravenous antibiotic therapy either during insertion or
following placement of a CVC is not recommended
90
Category of Recommendation
Category IB
Category IB
Category IB
Category IA
Category IA
Category IB
Category IA
Category IA
Category II
Category IA
Category 1B
Category IB
Category II
Category IB
Table 1, Chapter 10. Categories and recommendations for CLABSI reduction practices from the Healthcare Infection Control Practices
*
Advisory Committee of the Centers for Disease Control and Prevention (continued)
Recommendation
Educational interventions
Description
Education regarding appropriate indications, method of placement, and
surveillance for CLABSI are a critical component of a comprehensive CLABSI
prevention program
The use of five practices in unison at the time of CVC insertion, the bundle, is
recommended. These interventions include hand hygiene prior to insertion; use
of maximal sterile barrier precautions; chlorhexidine for skin antisepsis;
avoidance of the femoral site of insertion; and prompt removal of catheters when
no longer indicated
The use of trained personnel dedicated to the placement of CVCs in ICU and
hospitalized patients is recommended
Category of Recommendation
Category IA
Category IB
91
92
associated with a 50 percent reduction in the subsequent risk of CLABSI compared with
povidone iodine.45 A formal economic evaluation by the same authors projected that although
costlier initially, the use of chlorhexidine over povidone iodine for insertion site disinfection and
CVC care would lead to a 1.6 percent decrease in the incidence of CLABSI, a 0.23 percent
decrease in the incidence of death, and a savings of $113 per catheter.46 Existing HICPAC
guidelines endorse the use of chlorhexidine gluconate for skin antisepsis prior to CVC insertion
(Category IA).20
Antimicrobial catheters. The utility of catheters impregnated with a variety of substances
including chlorhexidine-silver sulfadiazine, minocycline-rifampin, benzalkonium chloride, and
silver have been evaluated in more than 20 randomized controlled studies and four recent
systematic reviews and meta-analyses.47-51 Meta-conclusions from these reviews remain limited,
due to heterogeneity arising from differences in the population, design, and conduct of the
pooled studies. For example, in a study involving a pediatric burn population, Weber and
colleagues found significant reductions in CLABSI with the use of minocycline and rifampin
antimicrobial coated catheters over non-coated catheters.52 However, a prospective, doubleblinded, randomized study in adults failed to show a reduction in CLABSI with a secondgeneration CVC coated with chlorhexidine and silver sulfadiazine.53 Due to the initial acquisition
cost, variation in benefit according to patient populations, and potential concern for inducing
antimicrobial resistance, routine use of antimicrobial CVCs is not recommended.20,54 However,
in facilities where high-rates of CLABSI persist despite deployment and compliance with
comprehensive CLABSI prevention efforts, the use of antimicrobial CVCs is considered
reasonable by current HICPAC guidelines (Category IA).20
The subclavian vein as the insertion point of choice. The site of CVC placement may
influence the risk of CLABSI, owing to the differing density of bacterial skin colonization at
each entry site. In a multicenter study of 289 patients randomized to undergo venous
catheterization using either the femoral or subclavian site, CVC placement in the femoral area
was associated with a substantially greater risk of CLABSI than was subclavian insertion (20
versus 3.7 per 1,000 catheter days).55 In a recent Dutch multicenter study involving 3,750 CVCs
and 29,003 CVC days, insertion into the femoral and jugular vein was independently associated
with an increase in the risk of subsequent CLABSI.56 In a study directly comparing the
subclavian to the internal jugular and femoral sites, the subclavian site was associated with the
lowest risk of infection (0.97 versus 2.99 and 8.34 per 1,000 catheter days, respectively).24 For
this reason, whenever medically feasible, the subclavian vein is the preferred site for venous
catheterization in the current HICPAC guidelines (Category IB).6,20,57-59 However, this
recommendation remains the subject of on-going debate, as some rigorous studies have found
that the risk of CLABSI from femoral vein CVC insertion is not greater than that associated with
insertion into the subclavian or internal jugular veins.60-62
93
Disinfect hubs, needleless connectors, and injection ports prior to CVC use. Contamination
of the catheter hub due to non-sterile access technique is a recognized path for developing
CLABSI. Minimizing contamination by wiping the catheter hub with an appropriate antiseptic
specifically recommended by the device manufacturer, or swabbing the membranous septum of a
CVC with 70% alcohol have been shown to reduce both risk of catheter contamination and
incidence of CLABSI.63-65 The practice of disinfecting access sites prior to CVC use,
colloquially dubbed scrub the hub, is linked to decreases in both bacterial colonization at
access sites and rates of incident CLABSI.66-68 Educational efforts targeting providers
responsible for CVC care (such as bedside nurses) are an important component in ensuring
dissemination and compliance with this practice.69 Although current HICPAC guidelines
emphasize minimizing the risk of contamination by scrubbing the hub with an appropriate
antiseptic (Category IA),20 several in vitro studies have demonstrated that even with strict
attention to decontamination, pathogenic organisms can persist in crevices or inside CVC access
valves and/or require prolonged duration of contact with an antimicrobial to significantly
decrease the level of colonization of CVC valves.69-72 In response, innovative technologies such
as those that incorporate antimicrobial compounds in the matrix of the CVC access valve, or
devices that bathe the valve with antimicrobials are being developed and tested.70,72-74 In the
absence of significant clinical experience with these novel devices, recommendations regarding
their widespread use are not possible.
Remove nonessential CVCs. Each day with a CVC increases the risk of developing
CLABSI.75,76 Prompt removal of CVCs that are no longer warranted is thus an important practice
to reduce CLABSI. This action necessitates both awareness of CVC presence and an ongoing
risk-benefit assessment of continued central venous access. In a study tracking temporary CVC
use in hospitalized patients, Chernetsky-Tejedor and colleagues reported that patients who
underwent PICC placement for venous access paradoxically also had 5.4 concurrent days with a
peripheral intravenous line (P<0.001), and had more days in which the only justification for the
CVC was intravenous administration of antimicrobial agents (8.5 versus 1.6 days; P=0.0013).
The authors therefore concluded that a substantial proportion of CVC-days might have been
unjustified in this cohort.65 In a recent survey conducted in a European hospital, neither the
bedside nurse nor the treating physician knew why a CVC was in place for 8.3 percent of nonICU patients.77 Importantly and relatedly, routine replacement of CVCs at pre-determined time
intervals has not been shown to reduce the risk of CLABSI and is not recommended based on the
available evidence (Category IA).20
Chlorhexidine cleansing. Daily bathing of patients with a chlorhexidine-based solution in ICUor advanced care settings may lower CLABSI incidence. In a crossover study conducted in a
medical ICU, daily washing with a 2% chlorhexidine-impregnated washcloth significantly
reduced subsequent BSI compared with using soap and water (4.1 vs. 10.4 per 1,000 patientdays, P<0.05).78 A study in a surgical ICU also found that daily bathing with a 2% chlorhexidine
gluconate impregnated cloth led to significant reductions in CLABSI (12.07 vs. 3.17 CLABSIs
per 1,000 days; 73.7% rate reduction, P= 0.04).79 The benefits of chlorhexidine baths may also
extend to high-risk patients outside of ICU settings. In a quasi-experimental before and after
study of the effect of daily washing with 2% chlorhexidine solution on CLABSI incidence,
Munoz-Price and colleagues reported a 99 percent reduction in the CLABSI rate in a long-term
acute care facility.80 However, a recent retrospective study involving patients in a surgical ICU
94
suggested that the benefit from chlorhexidine bathing might not apply to all patients.81 As the
evidence base for this practice is limited and conflicting, current HICPAC guidelines cautiously
endorse the use of chlorhexidine washes (either in solution form or as a chlorhexidine
impregnated wash cloth), for daily skin cleansing as a means to prevent CLABSI with a
Category II recommendation.20 However, the level of evidence for this recommendation may
soon be upgraded, as a recent meta-analysis pooling 12 studies found significant reductions in
CLABSI risk in studies that evaluated chlorhexidine cleansing in a medical ICU setting (OR
0.44, 95% confidence interval, 0.33 to 0.59).82
CVC dressing, chlorhexidine sponges and topical antibiotic use. The type of dressing and use
of topical antibiotic ointments or creams at the catheter site may affect the risk of CLABSI. In a
meta-analysis of seven studies comparing clear dressings to gauze for CVCs, transparent
dressings were associated with greater risk of catheter tip colonization (Relative Risk [RR] 1.78,
95% confidence interval [CI] 1.30 to 2.30, P<0.05), but not CLABSI (RR 1.63, 95% CI 0.76 to
3.47).83 Another meta-analysis of randomized controlled trials comparing gauze and tape to
transparent dressings found no significant differences between dressing type and risk of
CLABSI.84 Thus, for CLABSI prevention, existing guidelines do not endorse one type of
dressing over the other and leave the choice of CVC dressing to provider/patient preference and
clinical scenario.20
The use of a chlorhexidine gluconate sponge over the site of CVC insertion has been
associated with a decrease in the frequency and cost of CLABSI. In a study involving 1,636
patients with venous and arterial catheters, Timsit and colleagues reported that chlorhexidine
gluconate sponge placement at the site of catheter insertion substantially reduced the incidence
of CLABSI (1.4 to 0.6 per 1,000 catheter days, hazard ratio 0.39, P<0.03). However, severe
contact dermatitis was observed in eight low birth-weight infants (5.3 per 1,000 catheter days),
and the potential for this adverse effect remains an important limitation in the use of
chlorhexidine gluconate sponges in this population.85 In a recent economic evaluation,
chlorhexidine-impregnated sponge use in patients with CLABSI was estimated to save $197 per
patient using a 3-day dressing change strategy vs. $83 using a 7-day standard dressing change
strategy.86 In another cost-benefit analysis, a hypothetical 400-bed hospital inserting 3,078 CVCs
annually would avoid a projected average of 35 CLABSIs, 145 local infections, and 281 ICU
days with the systematic use of a chlorhexidine-impregnated foam dressing; potential annual
hospital net savings were projected at over $895,000.87 Owing to important differences in study
design and outcomes involving primarily pediatric populations, current guidelines recommend
the use of chlorhexidine-impregnated sponge dressings only in situations where the CLABSI rate
is not decreasing despite adherence to other prevention measures (Category IB).20
The use of topical antibiotic ointment or creams at the insertion site (e.g. povidone iodine) is
recommended only for patients with hemodialysis catheters, where its use has been associated
with suppression of BSI.88,89 Interestingly, a recent prospective, non-blinded crossover study
found that chlorhexidine sponge dressings were not protective against BSI in patients with
hemodialysis catheters.90 Conversely, topical antibiotic dressings are not recommended for
CLABSI prevention in non-dialysis patients as their use may paradoxically increase fungemia
and antimicrobial resistance in this category of patients (Category IB).20,91,92
Antibiotic/anti-infective locks in high-risk patients. A catheter lock refers to the instillation
of supra-physiologic doses of an intravenous antibiotic or anti-infective solution into a catheter
95
lumen between periods of CVC access. Several studies have examined both the utility of specific
antibiotic or anti-infective agents (e.g. vancomycin, cephalosporins, taurolidine, EDTA, ethanol)
and the targeted use of antibiotic locks in high-risk patient populations. In a systematic review
and meta-analysis, vancomycin-based antibiotic locks in patients deemed high-risk for CLABSI
(planned, long-term central venous catheter duration or those with a history of prior CLABSI)
significantly decreased the risk of this outcome (RR 0.34, P=0.04).93 A more recent systematic
review also reported reductions in the risk of subsequent CLABSI using this approach as an
adjunctive treatment, specifically in patients with poor venous access where catheter salvage was
key.94 In view of concerns regarding the potential for inducing antibiotic resistance, several
novel compounds have been tested as anti-infective locks. For example, a recent study found a
solution containing minocycline and EDTA to be highly efficacious in preventing CLABSI in
patients with hemodialysis catheters.95 In patients receiving prolonged home parenteral nutrition
via a CVC, the antineoplastic compound taurolidine was found to reduce the risk of CLABSI
when used as a catheter lock in a before and after study.96 Even though several studies have
found reductions in CLABSI incidence in specific populations, generalizations beyond these
groups are difficult and not appropriate.97-100 Thus, due to important differences in study design,
type of catheter, agent used, and patient population, the use of antibiotic locks should be limited
to those who are at high baseline risk for CLABSI (Category II).20
Systemic antibiotic prophylaxis. Routine systemic antibiotic prophylaxis during or after CVC
insertion to reduce the risk of CLABSI is not recommended (Category IB).20 A recent Cochrane
meta-analysis involving patients with cancer found no convincing evidence that prophylactic
peptidoglycan administration prior to CVC insertion was associated with reduced CLABSI
incidence.101 A recent study examining the effect of prophylactic cefazolin on CLABSI
following port placement similarly found no benefit associated with antibiotic treatment.102
96
implemented at the time of CVC insertion could reduce CLABSI. Notably, these five practices
were selected because they each had strong evidence supporting their efficacy in CLABSI
reduction and the lowest barriers to implementation. These five practices were hand hygiene
prior to insertion; use of maximal sterile barrier precautions; chlorhexidine for skin antisepsis;
avoidance of the femoral site of insertion; and prompt removal of catheters when no longer
indicated. Following implementation of this intervention, the mean rate of CLABSI dropped
from 7.7 per 1,000 catheter days at baseline to 1.4 per 1,000 catheter days at 16 months across
participating sites.33 The use of these five interventions in unison has been called the checklist
or the bundle. The use of the bundle and variations thereof has been associated with a
sustained decrease in the incidence of CLABSI, not only within the U.S., but internationally as
well.38,111-116 The bundle has also been found to be cost-effective both in the U. S. and abroad,
leading to its widespread acceptance as a key strategy with which to reduce CLABSI.20,117 The
HICPAC guidelines categorize the use of bundled interventions during CVC insertion as
performance improvement initiatives and recommends this practice to reduce CLABSI (Category
IB).20
Specialized CVC insertion teams. Data from several studies suggest that CVC placement by
specialized teams dedicated to this role leads not only to greater placement skills and reduced
insertion complications, but also to reduced rates of institutional CLABSI.25,33,38,111 The use of
dedicated and trained staff ensures predictable adherence to evidence-based practices such as
hand hygiene and maximal sterile barriers. The advent of nursing-led PICC teams represents an
important transformation in the placement of CVCs in both critically ill and hospitalized patients.
Preliminary studies suggest that these teams are associated with high rates of insertion success
and low rates of mechanical complications in a variety of patient settings.118-120 However, no data
comparing the risk of CLABSI in patients who undergo PICC placement by nursing PICC teams
compared with other providers (such as hospitalists or radiologists) are currently available. The
HICPAC guidelines recommend the use of trained personnel to insert CVCs (Category IA).20
97
patients in non-ICU settings and those receiving hemodialysis.124 In similar fashion, the Agency
for Healthcare Research and Quality (AHRQ) funded and launched an implementation program
called On the CUSP: Stop BSI. This national venture includes Federal agencies (e.g., CDC),
State organizations, and various professional societies, and aims to reduce the mean CLABSI
rate to less than 1 per 1,000 catheter days in each of the 50 United States.
98
99
can sustain these behaviors beyond the intervention stage.113 Ongoing performance measurement
and process improvement must thus come to represent a fundamental facet of national and local
efforts directed towards CLABSI prevention.
Moderate-tohigh
Evidence or
Potential for
Harmful
Unintended
Consequences
Low
Estimate of
Cost
Low-tomoderate
100
Implementation Issues:
How Much do We
Know?/How Hard is It?
Moderate-to-difficult/
Not difficult
(implementation of a
bundle)-to-moderate
(understanding
organization culture and
context)
References
1.
2.
3.
4.
5.
6.
7.
8.
101
9.
10.
11.
12.
13.
14.
15.
16.
17.
25.
18.
26.
27.
28.
29.
30.
31.
32.
19.
20.
21.
22.
23.
24.
102
33.
41.
34.
42.
35.
43.
44.
45.
46.
36.
37.
38.
39.
47.
40.
48.
103
49.
50.
51.
52.
53.
54.
55.
56.
104
57.
58.
59.
60.
61.
62.
63.
64.
65.
73.
66.
74.
67.
75.
76.
77.
Zingg W, Sandoz L, Inan C, et al. Hospitalwide survey of the use of central venous
catheters. J Hosp Infect. 2011
Apr;77(4):304-8. PMID 21288595.
78.
79.
80.
81.
68.
69.
70.
71.
72.
105
82.
83.
84.
85.
86.
87.
88.
89.
90.
106
91.
92.
93.
94.
OHoro JC, Silva GL, Safdar N. Antiinfective locks for treatment of central lineassociated bloodstream infection: a
systematic review and meta-analysis. Am J
Nephrol. 2011;34(5):415-22. PMID
21934302.
95.
96.
97.
98.
99.
100.
107.
108.
109.
101.
102.
110.
103.
111.
104.
112.
113.
114.
105.
106.
107
115.
116.
117.
118.
119.
120.
121.
122.
123.
108
124.
125.
126.
127.
128.
129.
130.
131.
132.
133.
134.
135.
136.
137.
138.
139.
140.
109
Introduction
Ventilator associated pneumonia (VAP) is defined as a hospital-acquired pneumonia that
develops within 48 to 72 hours after endotracheal intubation; the diagnosis hinges on a lack of
evidence suggesting that the infection developed prior to intubation. VAP is the most common
intensive care unit (ICU)-acquired infection, accounting for 25 percent of all ICU infections and
50 percent of ICU antibiotic use. At least 250,000 VAPs occur in the United States (U.S.) each
year. This condition causes complications in 8 to 28 percent of mechanically ventilated patients
and carries a mortality risk of approximately 10 percent (range 6% to 27%), resulting in a
possible 25,000 VAP-attributable deaths every year. Patients who develop VAP stay, on average,
4 days longer in the ICU. The per-case cost of VAP is estimated to be $23,000, and the total
incremental costs to the U.S. health care system are high: $2.19 to 3.17 billion USD per year1-3
The wide range of these estimates results from the lack of universally accepted, reliable
diagnostic criteria for VAP is present. The diagnosis of VAP may be based on any of a variety of
definitions, including a surveillance definition, a clinical definition, a microbiologically
confirmed definition, or a combination of the three methods. Microbiologically confirmed
definitions also may be differentially based on blind tracheal aspirates, directed bronco-alveolar
lavage, or even protected brush specimens.2
The original Making Health Care Safer report examined four interventions related to VAP:
variation of position (semi-recumbent positioning and continuous oscillation), continuous
subglottic suctioning, selective decontamination of the gastro-intestinal tract and the use of
sucralfate. While the data in favor of semi-recumbent positioning was limited (reduced VAP but
did not change mortality), the practice was judged to be easy to implement and had essentially no
cost or adverse effects. Oscillation was less clear in its benefit secondary to poor methodological
quality of the studies. While no evidence for harm was found, there were increased costs,
estimated to be about $100/day at the time of that report. Subglottic suctioning was judged to be
a promising strategy. At the time of the report, it was infrequently used and there were only a
few studies. Harmful effects were felt to be negligible but there were incremental costs for the
specialized endotracheal tubes required for this strategy. Selective gastro-intestinal
decontamination was found to have strong benefit for reducing VAP, though cost-effectiveness
was unclear. Of the trials examined, none reported adverse events from this practice; however,
there is continued concern that this practice may have a deleterious effect on antibiotic sensitivity
in general, leading to more resistant organisms over time in individuals as well as on a
population basis. Sucralfate as a VAP prevention strategy was judged to be inconclusive.
Additionally, sucralfate is inferior compared with H-2 blockers for preventing gastro-intestinal
bleeding. Given the increased risk of mortality with gastrointestinal bleeding and the increased
costs should this complication occur, sucralfate can no longer be recommended for VAP
prevention and H-2 blockers are the preferred agent for preventing gastro-intestinal bleeding in
critically ill patients.
This updated review focuses on four strategies as well; elevation of the head of the bed,
sedation vacations, oral care with chlorhexidine and subglottic suctioning.
110
What Are the Patient Safety Practices for Preventing VentilatorAcquired Pneumonia?
We conducted a systematic review of the literature to update a 2001 review conducted for the
original report. A recent study estimates that 14,000 to 20,000 lives could be saved each year in
the U.S. if best practices to prevent VAP were universally applied to all patients on mechanical
ventilation.3 The four primary recommended practices include: elevating the head of the bed to
30 degrees, sedation vacations, oral care with chlorhexidine (CHG), and subglottic suctioning
endotracheal tubes. Ventilator bundles usually include other elements such as deep venous
thrombosis (DVT)/pulmonary embolism (PE) prophylaxis and Peptic Ulcer Disease prophylaxis,
but these procedures are designed to prevent other ventilator-associated conditions and do not
address VAP prevention specifically. In fact, Peptic Ulcer Disease prophylaxis may increase the
risk of VAP. Other VAP-specific preventive interventions may include use of closed suctioning
circuits, scheduled circuit changes and a preference for orotracheal over nasotracheal intubation.
The remainder of this section describes the evidence in support of the four primary VAP specific
practices.
Head-of-Bed Elevation
The practice of head-of-bed elevation to prevent VAP has been recommended by several
medical groups including the Canadian Critical Care Trials Group, the American Thoracic
Society and Infectious Diseases Society of America, and the Centers for Disease Control and
Prevention. This recommendation is based on early data showing that being supine was an
independent risk factor for VAP.4 Importantly, a study in 1999 by Drakulovic and colleagues5
demonstrated a reduction in VAP with patients in the semi-upright position.5
A recent systematic review by Niel-Weisse and colleagues that applied strict inclusion
criteria (randomized or quasi-randomized trial, published as a full paper and not an abstract, state
the outcome measures used and present data sufficient to calculate the risks in both groups) and
included only three of 208 potential studies, representing a total of 337 patients questioned
whether patients head elevation can be maintained continuously above 30 degrees in ICUs, and
point prevalence assessments used in many studies may overstate how often the goal is met.4 The
effect of head-of-bed elevation on the incidence of both clinically diagnosed and
microbiologically confirmed VAP was found to be non-significant (RR = 0.47, 95% CI = 0.19 to
1.17 and RR = 0.67, 95% CI =0.23 to 2.01, respectively). A second study used broncho-alveolar
lavage, and the other two used tracheal aspirates for the microbiological assessment. The third
study found no significant increase in harm (decubitus ulcers); other potential harms (such as
DVT) were not assessed. These same three trials also found no significant impact on mortality
(pooled RR = 0.90, 95% CI = 0.64 to 1.27). The data were also judged to be of low quality for
methodological reasons. Despite these findings, an evaluation of the results using an online
Delphi process recommended the practice of keeping the head of the bed elevated by greater than
30 degrees to prevent VAP (most studies had actually used 45 degrees as their target).4 The
favorable point estimates (all favored the intervention despite lack of statistical significance) and
the lack of measurable harm may have influenced this recommendation.
Sedation Vacations
The use of sedation vacations, or sedation holds, has been shown to help patients wean from
mechanical ventilation more quickly than when these techniques are not employed.6,7 Further,
sedation vacations reduce patients exposure and subsequent risk of VAP as well as several other
111
112
This practice also significantly reduced the duration of mechanical ventilation and length of stay
in the ICU, although it had no impact on ICU- or hospital mortality.
estimated that 11 people need to be treated (number needed to treat) in order to prevent one
VAP. Although the cost of these endotracheal tubes is approximately $18 USD, one model of
continuous washing tubes (inflow/outflow ports with pumping system) costs about $200 USD. In
comparison, the cost of a standard endotracheal tube is approximately $1 USD. If the number
needed to treat is accurate, these special endotracheal tubes (even the most expensive versions)
are cost-effective, especially if reserved for patients likely to remain intubated for more than 48
to 72 hours (the risk for VAP in those requiring intubation less than 48- to 72 hours is considered
low). This conclusion is further supported by Hallais and colleagues,19 who compared the cost of
these tubes to the cost of VAP, using very conservative values. The authors found that averting
only three VAPs would offset the cost of the special tubes. Based on this cost analysis, any ICU
with at least 3 VAPs per year would find that switching to these tubes reduces harm as well as
costs.
Ventilator bundles have also recently been evaluated for their cost-effectiveness. A Danish
study20 retrospectively examining ventilated patients in a single ICU found that the cost of
preventing one VAP was 4451 (approximately $6,000 USD), and the cost of preventing one
death was 31792 (approximately $42,000 USD). While the cost and incidence of each VAP
varies across patient populations, the study concluded that the ventilator bundle would likely be
cost-effective in most environments.
114
Moderate-tohigh
Evidence or
Potential for
Harmful
Unintended
Consequences
Low
Estimate of
Cost
Low-tomoderate
Implementation Issues:
How Much do We
Know?/How Hard Is it?
Moderate/Moderate
References
9.
10.
11.
12.
13.
14.
15.
1.
2.
3.
4.
5.
6.
7.
8.
115
16.
20.
17.
21.
18.
22.
23.
19.
116
Introduction
Many hospitals choose to reprocess single-use devices (SUD)those intended by the
manufacturer to be discarded after one use--for reuse in additional patients.1 Reprocessing
includes cleaning, sterilization, and if necessary, refurbishing. Specific information on the size of
the reprocessing industry is not available.2 According to the Association of Medical Device
Reprocessors Web site, disciplines that commonly use reprocessed devices include:
cardiovascular; arthroscopic/orthopedic; general surgery; gastroenterology; laparoscopic surgery.
A wide variety of SUDs are reprocessed. Commonly reprocessed SUDs include: arthroscopic
shavers; biopsy forceps; blood pressure cuffs; clamps and dissectors; compression sleeves;
electrophysiology catheters; external fixation devices; laparoscopic scissors and forceps; opened
but unused items; orthopedic drill bits and burrs; phaco tips; pneumatic tourniquet cuffs; pulse
oximeter sensors; scissors and staplers; soft tissue ablators; trocars.1 Opened but unused items
are not technically reused but also must be reprocessed.
Using SUDs is money-saving and generally thought to be safe.1 However, SUDs are only
required to be demonstrated to the Food and Drug Administration (FDA) as safe for one use;
some manufacturers and the Medical Device Manufacturers Association contend that reusing
SUDs is unsafe because the devices are frail and cannot be adequately cleaned and resterilized.1,3
Potential risks to patients include infection, toxicity, particulate contamination, and mechanical
failure.1
Although reuse of single-use devices is common and perhaps even pervasive, little evidence
on its safety and efficacy has been published. To gather the available evidence, a literature search
of PubMed was conducted for English language articles published between January 1, 2001 and
November 2, 2011.
What Are the Practices for Assuring the Safety of Reused Devices?
Reprocessing used SUDs is subject to FDA oversight. On August 14, 2000, FDA issued a
policy on the reuse of single-use medical devices making hospitals and third-party reprocessors
subject to all the requirements of the Federal Food, Drug, and Cosmetic Acta requirement
formerly imposed only on original equipment manufacturers (OEMs).4 In response, many
hospitals that had been reprocessing SUDs in-house began using third parties to reprocess the
devices. Unused items are not subject to FDA oversight.5
The Medical Device User Fee and Modernization Act of 2002 expanded regulatory
requirements. Premarket notification submissions (510(k)s) for certain reprocessed SUDs
identified by FDA must now include validation data. Validation data include cleaning,
sterilization, and functional performance data, which confirm that each SUD will remain
substantially equivalent to a predicate device after the maximum number of times the device is
reprocessed.4 In addition, the reprocessor must be indicated with a mark or label for each
reprocessed SUD.
117
According to FDA regulations, a third-party or hospital reprocessor must comply with the
same requirements that apply to original equipment manufacturers, including:
Submitting documents for premarket notification or approval for each device and model
reprocessed
Registering as a manufacturer with FDA and listing all products
Submitting adverse event reports
Tracking devices whose failure could have serious outcomes
Correcting or removing from the market unsafe devices
Meeting manufacturing and labeling requirements.4
The FDA considers hospitals that reprocess devices as device manufacturers subject to the
requirements of the Quality System (QS) Regulation.6 However, most hospitals who use
reprocessed SUDs obtain them from third parties who perform the reprocessing.4 An estimated
95% of reprocessing in the U.S. is completed by two firms,1 Stryker Sustainability Solutions
(formerly Ascent Healthcare Solutions, Phoenix, AZ) and SteriMed Inc. (Minneapolis, MN).3
According to their Web site, Stryker claims the majority of the reprocessing market and has over
2,000 hospital members. Their service includes delivering orders and picking up used equipment
(which hospital personnel leave in marked bins), and at their facility, sorting, cleaning,
refurbishing/repairing, repackaging, and resterilizing equipment. The FDA has cleared or
approved a variety of sterilizing agents that can be used in reprocessing7 and inspects
reprocessing facilities for compliance with regulations, with steep penalties for violators.5
118
119
engineering, administration, central sterile supply, surgery, finance, and physician(s) advocating
reuse.1 A third party reprocessor should be selected based upon registration with the FDA and
compliance with FDA regulations; the types of devices to be reprocessed; and support (including
logistical support such as device pickup).1 Reprocessors usually provide template policies and
procedures to hospitals to support implementation.1
Low
Evidence or
Potential for
Harmful
Unintended
Consequences
Low
Estimate of
Cost
Low
Implementation Issues:
How Much do We
Know?/How Hard Is it?
A lot/Not difficult
References
1.
2.
120
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
121
14.
15.
16.
17.
18.
19.
20.
21.
checklist can potentially be used to clarify the details of these three steps. The Universal
Protocol is intended to prevent wrong surgery not just in the operating room but
anywhere an invasive procedure is performed (e.g., interventional radiology unit).12
Checklists specifically intended to check anesthesia equipment. The WHO checklist also
contains a specific item about preoperative anesthesia (Is the anaesthesia machine and
medication check complete?). This single item could itself be addressed by a subchecklist. In 2008, the American Society of Anesthesiologists provided general
guidelines about items that should be checked before surgery, and institutions can
implement the guidelines to tailor the checklist to their specific equipment and clinical
settings.13
Checklists have also been developed, implemented, and assessed outside of the realm of
surgery. The Michigan ICU checklist (also referred to as the Keystone project) has been shown
to prevent central line-associated bloodstream infections (CLABSI).14,15 This program involved a
multifactorial intervention at 108 Michigan ICUs. Data showed a reduction in CLABSI from 7.7
infections per 1,000 catheter-days before the program to only 1.4 infections per 1,000 catheterdays at 16 to 18 months after program initiation follow-up. A 2001 study by Dixon-Woods and
colleagues15 proposed six reasons for this reduction, including creating a densely networked
community with strong horizontal links that exerted normative pressures on members and
harnessing data on infection rates as a disciplinary force. A recent systematic review of this
program (and other PSPs to prevent hospital-associated infections) was conducted by the Blue
Cross and Blue Shield Evidence-based Practice Center; please refer to that report for further
information about the Keystone project.16
123
(for example, the SURPASS checklist does not specifically mention any of the following: pulse
oximetry, difficult airway, risk of blood loss (although it asks whether blood products are
available), team introductions, and anticipation of critical events).
In January 2004, the Joint Commission launched the first version of the Universal Protocol
(UP) for Preventing Wrong Site, Wrong Procedure, Wrong Person Surgery.11,21 It comprises
three sets of steps: pre-operative verification process, marking the operative site, and a time
out immediately before the operation. The preoperative verifications (of person, procedure, and
site) are supposed to occur not only in the operating room, but also (if applicable) when the
procedure is scheduled, when the patient enters the health care facility, and anytime care is
transferred between caregivers. Site marking should involve only the operative site and should be
visible before the patient is draped. The time out is to occur before incision and involve the
entire operating room team. The Universal Protocol is not a checklist,12 but it could be
implemented using one or more checklists. Both steps 1 and 3 specifically mention the potential
use of a checklist.
Anesthesia safety guidelines and standards are actively reviewed and modified globally
through organizations such as the WHO and the World Federation of Societies of
Anaesthesiologists (WFSA).22 The latest WFSA standard, which was developed as part of the
Safe Surgery Saves Lives project, recommends that an appropriate pre-list check be
performed prior to the start of each operating list and an appropriate pre-patient check be
performed prior to each anesthetic. In addition, individual anesthesia societies are developing
guidelines for pre-anesthesia checks, including the American Society of Anesthesiologists (ASA)
and the Association of Anaesthetists of Great Britain and Ireland (AAGBI). The latest U.S. preuse checkout guidelines, entitled Recommendations for Pre-Anesthesia Checkout (PAC)
Procedures, were published in 2008 by the ASA.13 These guidelines were a result of a multiyear effort by an ASA task force consisting of members from the ASA, the American
Association of Nurse Anesthetists (AANA), the American Society of Anesthesia Technicians
and Technologists (ASATT), and major anesthesia system manufacturers. The latest AAGBI
revision was published in 2004 and has been adopted by many institutions around the country.23
Similar to the WFSA guideline checklist, the full ASA and AAGBI checklists were designed to
be used at the start of the day with a subset of the full checklists performed prior to each
procedure. These societies sample checklists were developed as a basis for institutions to
develop their individual checklists.
Additional background information about preoperative checklists, including how they were
developed and modified, and the overlap between different checklists, appears in Appendix A.
124
125
SURPASS Checklist
An empirical test of the 90-item SURPASS checklist was reported in a 2010 study by
DeVries and colleagues.7 The design was a 6-month interrupted time series with concurrent
controls; six hospitals using the checklist were matched with five other hospitals that did not, and
researchers measured the rates of surgical complications in both groups. The 11 hospitals were
distributed through the Netherlands and comprised six tertiary hospitals, three academic
hospitals, and two regional hospitals; numbers of beds per hospital ranged from 3801002. These
hospitals had already been measured for their safety performance, so the potential Hawthorne
effect is lower than it would have been in hospitals just starting to be measured. Regarding
implementation, authors stated that the SURPASS checklist involved extensive time and effort.
A random sample of cases generally revealed good compliance with the checklist (median 80%).
The 3-month period after the checklist was initiated (compared with the 3 months before)
saw numerous improvements: decreases in the percentage of patients with complications (from
15% to 11%), in-hospital mortality (1.5% to 0.8%), patient temporary disability (9.4% to 6.6%),
and reoperations (3.7% to 2.5%). No such improvements were found among the control
hospitals. Interestingly, the extent of improvement was associated with greater compliance with
the checklist: the 566 patients whose surgery involved greater checklist compliance had 7.1
complications per 100 patients, which was considerably lower than the 18.8 per 100 experienced
by the 580 patients whose surgery involved less checklist compliance. This finding provided
greater confidence that the checklist itself was the reason for the improvements. A subsequent
retrospective review of 294 medical claims10 estimated that 40 percent of deaths and 29 percent
of liability incidents might have been prevented if the SURPASS checklist had been used.
126
of wrong site, wrong level surgery.27 Therefore, the preventive benefits of a checklist to prevent
wrong-site surgery, are generally assumed based on clinical expertise.
127
compliance with the checklist; one simply reported 97 percent compliance, and the other two
reported improvement over time (from approximately 60% to approximately 80% in one study,
and from 85% to 95% in another study). One study reported that it took about two minutes to
complete the checklist, and that 20 percent of respondents believed it caused an unnecessary time
delay.
Feedback from surgical teams was generally positive, but support tended to be greater from
nurses and anesthetists than from surgeons. Two studies reported increases in certain attitudinal
variables such as the degree to which people felt familiar with others in the operating room, the
quality of communication, the anticipated safety of patients, and the usefulness of the checklist in
either elective or emergency cases. Behaviorally, one study reported that after 3 months, team
briefings were occurring in 77 percent of operations and time-outs in 86 percent. Another study
reported improvements in anesthetists knowledge about patients, their check of anesthesia
equipment, and staff knowledge of patient identity/procedure/site.
Reasons cited for success included good training and staff understanding, a local champion,
support from upper management, being able to modify the checklist, distribution of
responsibility, the feeling of ownership by team members, and enhanced communication and
teamwork. Barriers to implementation included general surgeon resistance to changing habits,
the belief that they were already checking those things, awkwardness of self-introductions, steep
interpersonal hierarchy, and a fear of legal responsibility if a complication occurred after they
had signed a form.
One ongoing research project, funded by the Agency for Healthcare Research and Quality, is
entitled Factors Associated with Effective Implementation of a Surgical Safety Checklist.34
This 2010-2013 project will examine implementation processes in a large group of U.S. and
international hospitals to identify factors supportive of effective implementation. Further, the
team will determine how teamwork helps explain the impact of the checklist.
The WHO Web site (www.who.int/patientsafety/safesurgery/en/) provides advice to hospitals
for implementing the checklist.35 This advice includes statements such as The key to successful
implementation is to start small. Start with a single operating room on day 1 and see how it
works. This will guide you to strategies for altering the checklist to fit your needs, as well as
identify potential barriers to adaptation.35 Other implementation advice from WHO is available
in the Frequently Asked Questions section
(www.who.int/patientsafety/safesurgery/faq_introduction/en/index.html), the 20-page Starter
Kit for Implementing the Surgical Safety Checklist
(www.who.int/patientsafety/safesurgery/testing/participate/starter_kit-sssl.pdf), and the
Checklist Adaptation Guide
(www.who.int/patientsafety/safesurgery/checklist_adaptation.pdf). Regarding checklist
modification, the Web site states, Do not hesitate to customize the checklist for your setting as
necessary, but do not remove safety steps just because you are unable to accomplish them. Also,
regarding feasibility, the WHO states that It should take no more than a minute to complete
each section of the checklist (i.e., three minutes in total). The pilot study reported that at various
sites, introduction of the checklist took only 1 week to 1 month.6
The Web site www.safesurg.org also provides additional materials relevant to the WHO
checklist. Those interested in implementing the checklist are encouraged to register with the Web
site. The Web site provides a template for the 2009 checklist (www.safesurg.org/templatechecklist.html) in Microsoft Word format (Microsoft Corporation, Redmond, Washington, U.S.),
and an implementation manual is available (www.safesurg.org/implementation-manual.html).
129
The Web site also provides a list of other institutions modified checklists
(www.safesurg.org/modified-checklists.html), where institutions can submit their modifications
of the WHO checklist to be made publically available. On October 3, 2011, the publicallyavailable list contained 79 checklists from 25 countries.
The site also provides several downloadable videos (www.safesurg.org/videos.html): one on
how to use the WHO checklist; one on how not to use it; two from the National Patient Safety
Agency in the United Kingdom; one from University Health Network Hospital in Toronto; one
from the Surgical Care and Outcomes Assessment Program in Washington State; two from Great
Ormond Street Hospital in the United Kingdom; one in French from Fattouma Bourguiba
Hospital in Monastir, Tunisia; one in Spanish from la Agencia de Calidad de Andalucia; one
Spanish translation of the WHO how-top video; and two from Auckland City Hospital in New
Zealand.
SURPASS Checklist
We performed a citation search to determine if the SURPASS checklist has yet been
attempted outside the Netherlands; however, no such attempts were identified. The SURPASS
Web site (www.surpass-checklist.nl/home.jsf?lang=en) describes an electronic version of the
checklist (called SURPASS Digital) that can be used by any web-connected computer. The
electronic version allows one to modify the checklist, although the designers of SURPASS
strongly encourage users to avoid modification (www.surpasschecklist.nl/content.jsf?pageId=FAQ&lang=en).
130
The Swedish study38 involved two hospitals, each of which had a recent wrong-site surgery
incident, and a root-cause analysis suggested that a time-out procedure might help. A time-out
checklist was implemented, and one year later, a questionnaire was sent to all 704 team
members. Of the 331 responders, 93 percent expressed the belief that the checklist contributes to
increased patient safety (either without a doubt, or probably). When asked about eight
specific components of the time-out checklists, the percentage of respondents who believed the
component was very important varied widely, from a low of 14 percent for the introduction of
team members to highs of over 80 percent for patient identity, correct procedure, and correct
side.
The English study39 was conducted at a childrens hospital in which staff had incorporated an
eight-item correct-site surgery checklist into an existing preoperative checklist. Five people were
required to sign the documentation: marking surgeon, operating surgeon, ward nurse, scrub
nurse, and anesthetist. Comparing 2008 to 2006, correct completion was unimproved for four of
the eight items (ward nurse signed, operating surgeon signed, scrub nurse, signed, and operating
department practitioner signed) but was improved for the other four (mark site documented, no
mark required documented, entries legible, and marking surgeon signed).
The North Carolina study37 implemented a checklist to prevent wrong-site surgery that was
tailored to the hospitals preferences and procedures. Previously, the staff was using a
cumbersome form to document their compliance with the Universal Protocol. Champions
demonstrated the checklist during educational staff meetings, and new staff were given a primer.
Staff commented favorably that they no longer had to remember everything.
The Association of PeriOperative Registered Nurses (AORN) Comprehensive Surgical
Checklist (www.aorn.org/uploadedImages/Images/Images/comprehensive_surgical_checklist_
RGB961.jpg) was a collaborative effort between AORN, the developers of the WHO checklist,
and the Joint Commission. The Web site states that the checklist, created in April 2010, was
created to support a facilitys need to use a single checklist that includes the safety checks
outlined in the World Health Organizations (WHO) Surgical Safety Checklist, while also
meeting the safety checks within The Joint Commissions Universal Protocol in order to meet
accreditation requirements. Our searches identified no empirical studies of this checklist.
Another combined checklist (called a crosswalk) combining the WHO checklist and the
Universal Protocol was published in November 2011 by the Pennsylvania Patient Safety
Reporting System.40 This document also addresses checking preparedness for surgical fires, as
well as two intraoperative checks specifically for spinal surgery involving precise locations. Due
to the recency, no studies exist yet on this crosswalk.
131
modified, or rejected according to clinical needs and constraints and are not intended to replace
local institutional policies. In addition, practice guidelines are not intended as standards or
absolute requirements, and their use cannot guarantee any specific outcome. Practice guidelines
are subject to revision as warranted by the evolution of medical knowledge, technology, and
practice. They provide basic recommendations that are supported by a synthesis and analysis of
the current literature, expert opinion, open forum commentary, and clinical feasibility data.41
The ASA encourages institutions to submit their version of the Pre-Anesthesia Checkout
(PAC) for publication on the ASA Web site (www.asahq.org/For-Members/ClinicalInformation/2008-ASA-Recommendations-for-PreAnesthesia-Checkout/SampleProcedures.aspx). Currently, sample PACs are posted for the following anesthesia system
models: (1) General Electric AESTIVA, (2) Draeger Apollo, (3) Draeger Narkmoed GS,
(4) Draeger 6000, (5) Draeger B/C/GS, and (6) Draeger Fabius GS. Eight U.S. hospitals are
currently represented on the ASA collection of sample checkouts.
As an additional international example of implementing anesthesia checklist guidelines, the
Columbian 2009 version of their Minimum Safety Standards in Anaesthesia states that
anaesthesiologists and surgeons must collaborate in completing an overall check list, which is to
include at least the items in the WHO checklist. In addition, before applying anaesthetic, the
anaesthesiologist must complete a pre-anaesthetic checklist.42
In referencing earlier implementation strategies for aviation checklists, a 2000 article by
Blike and Biddle28 propose the three Ps for successful implementation of their anesthesia
machine electronic checklist. They refer to the three Ps as a guiding philosophy, with
procedures designed to achieve the goal of the philosophy using consistent policies for
implementation. They concluded that the earlier AACR was deficient in that the associated
published checklist had no supporting philosophy.
Regarding staffing, the 2008 ASA guidelines identify particular aspects of the PAC that
could be performed by a qualified anesthesia and/or biomedical technician. However, regardless
of the level of training and support by technicians, the anesthesia care provider is ultimately
responsible for proper function of all equipment used to provide anesthesia care.13
132
An additional potential cost benefit relates to reduced litigation claims. With regard to
anesthesia, comparing the period prior to the1990s to the period from 1990 to 2003, the
proportion of claims with substandard care decreased (from 39% to 22%), and payments were
made less frequently (from 58% to 42% of the time).3
Are There Any Data About Adoption and Diffusion of This Patient
Safety Practice?
On February 22, 2012, the WHOs Surgical Safety Web Map indicated that as of February 1,
2012, 4,120 hospitals had expressed interest in using the checklist and 1,790 of these hospitals
have used the checklist in at least one operating theatre (Figure 1). On the map, red crosses
represent those expressing interest, and yellow crosses represent previous/current users.
Figure 1, Chapter 13. Screenshot of adoption and diffusion of the WHO surgical safety checklist
Note: This figure is a screenshot taken on 2/22/2012 of the WHO Surgical Safety Web Map
(http://maps.cga.harvard.edu:8080/Hospital/). Red crosses represent hospitals who have expressed interest in using the WHO
Surgical Safety Checklist (as of 2/1/2012); yellow crosses represent hospitals that have used the checklist in at least one operating
theatre. Using the right-hand panel, the map can also be configured to display locations of endorsing organizations, international
endorsing organizations, pilot sites, and countries with nationwide implementation. Granted permission by the World Health
Organization.
Our searches found that a number of professional organizations have recommended adoption
of the WHO checklist (Table 1).
133
Web site
www.ihi.org/offerings/
MembershipsNetworks/
MentorHospitalRegistry/Pages/
SurgicalSafetyChecklist.aspx
National Patient
Safety Agency
(NPSA) in the UK
www.nrls.npsa.nhs.uk/alerts/
?entryid45=59860
France
NA
Canadian Patient
Safety Institute
(CPSI)
NA
Washington State
Surgical Care and
Outcome
Assessment Program
49
(SCOAP)
www.scoap.org/
checklist/index.html
Details
In December 2008, then-president Donald Berwick
issues the Surgical Safety Checklist Challenge: to have
each hospital use the checklist in at least one
operating room by April 1, 2009. To assist facilities in
implementing the checklist, the IHI Web site provides a
list of eight mentors throughout the United States who
have already implemented the checklist. The
demographics of these eight sites are provided to
enable facilities to match themselves up with similar
mentor facilities.
NPSA mandated in February 2010 the use of the
45
checklist in all of its Trusts in England and Wales.
The NPSA Web site contains downloadable materials,
videos, and three tailored WHO surgical checklists (for
radiological interventions, cataract surgery, and
maternity cases). Also, the Surgical Checklist
Implementation Project, funded by the National Health
Service (NHS), involves four studies of implementing
the WHO checklist at 20 NHS Trusts
46
(www.safesurgery.org.uk/). The topics of the four
studies are (1) perception of the checklist and possible
barriers to use; (2) additional quantitative data on staff
perceptions of the checklist; (3) how the checklist is
actually used in operating rooms; (4) the impact of
46
checklist use on clinical outcomes.
France mandated the use of the WHO checklist in all
47
its 8000 hospitals in 2010.
The CPSI has endorsed the checklist, and checklist
implementation is now an accreditation standard for
48
Canadian hospitals.
The Canadian province of Ontario mandated use of the
47
checklist in 2011.
SCOAP stated a goal of having all of its hospitals use
the WHO checklist in every operating room by the end
of 2009 . The February 2010 SCOAP version of the
WHO checklist is available on the Web site. The Web
site also states that According to the Washington
State Hospital Association, 100% of Washington State
hospitals have either implemented a standardized
surgical checklist or are in the process of doing so.
Hospitals can also order a 2x3 foot laminated SCOAP
checklist from the Web site.
The SCHA, in the fall of 2010, planned to institute the
checklist in all the states hospitals over the next few
47
years.
Ireland and Jordan each plan to require checklist
50
implementation in all its hospitals.
The Spanish Ministry of Health and the Spanish
51
Association of Surgeons have joined the initiative.
NA
134
Web site
NA
Details
These organizations developed a modified version of
52
the WHO checklist. The checklist was launched in
August of 2009 with the endorsement of national
government health departments in both countries.
The webcast event Check a Box, Save a Life was launched on October 22, 2009 to
promote the use of the WHO checklist.53 The event, run mostly by medical students, involved
182 hosting sites from 121 medical institutions and an estimated 1,400 online viewers. A
Facebook page had enrolled 111 medical students who agreed to host the event at their
institutions. At the Institute for Healthcare Improvement forum 6 weeks later, 15 case reports
were presented that detailed checklist-related projects.
In January 2010 in the UK, just before the mandatory requirement to use the WHO checklist
was instituted by the NPSA, Sivathasan and colleagues54 conducted telephone interviews with
238 hospitals in the UK (randomly selected from some 540 hospitals, therefore representing
about 44% of UK hospitals). Almost all (99%) of the hospitals had heard of the checklist, and its
use was already compulsory in 65 percent of them. In hospitals where it was not required, 81
percent used it voluntarily, and 75 percent had a plan to make it mandatory in the future.
However, some operating rooms reported partial use of the checklist, i.e., intentionally skipping
items or skipping the entire checklist because of time constraints.
In June 2009, the journal OR Manager received online data from 136 subscribers regarding
use of the WHO checklist.55 About half (48.5%) said they had implemented the checklist, and
64 percent said the checklist has improved safety in the operating room. However, 11 percent of
respondents stated that the checklist was not well accepted by surgeons, and another 63 percent
said surgeons did accept it but with reservations. Nurses were believed to have a somewhat
greater degree of acceptance, with only 2 percent not well accepted and 52 percent accepted
with reservations.
A survey in October/November 2009 of 12 oral and maxillofacial consultants in Yorkshire
England found that all were aware of the WHO checklist, but only 5/12 were actually using it.56
Ten of 12 expressed the belief that it would improve patient safety, but four of 12 said it would
not improve team communication.
Regarding the Universal Protocol, accredited hospitals are required to comply. Therefore the
diffusion of the Universal Protocol is large, by mandate. However, as stated earlier, the
Universal Protocol is not a checklist. We found no published information on how many hospitals
actually use a checklist in their efforts to comply.
135
High
Evidence or
Potential for
Harmful
Unintended
Consequences
Negligible
Estimate of
Cost
Low
Implementation Issues:
How Much do We
Know?/How Hard Is it?
A lot/Moderate
References
1.
2.
136
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
137
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
34.
35.
36.
37.
38.
39.
40.
41.
27.
28.
29.
30.
31.
32.
33.
138
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
139
54.
55.
56.
57.
140
The intervention comprises five basic components. First, a trained surgical clinical reviewer
(SCR) prospectively collects data on preoperative and clinical variables and on 30-day outcomes
(outcomes are described in more detail, below). The number of variables depends on the
particular Program option chosen (which in turn depends on the needs and size of the particular
hospital) but is typically 46 or 69 for each case. A predetermined number of cases is reported,
which again depends on the Program option (also referred to as Use Option) the hospital has
elected to use. The second component is development and maintenance of models of expected
mortality and morbidity by risk and types of procedures performed. The third component is the
calculation of the observed-to-expected (O/E) 30-day mortality and morbidity ratios. Data are
then fed back to individual sites as observed-to-expected ratios of, typically, 21 morbidities, such
as wound, respiratory, central nervous system, urinary and cardiac complications, as well as
mortality. Data are provided alongside blinded national results from the other participating sites.
Sites are designated as being a high (worse than expected) or low outlier (better than expected)
for each category of morbidity and for mortality.
Figure 1, Chapter 14. Example of observed to expected ratio reporting for American College of
Surgeons national surgical quality improvement program
Figure 1 accessed from the ACS NSQIP Web site, December 2011. Reprinted by permission of American College of Surgeons
NSQIP.
Lastly, institutions then identify areas where they are a high outlier and improvement is
needed. Auditing by the ACS NSQIP staff occurs randomly and for cause, that is, site reports
many high risk patients but a low complication rate. Individual surgeon-level data are provided
to the participating hospital if they request it. While the responsibility for making changes and
addressing areas in need of improvement remains with the individual sites, the administrative
ACS NSQIP body provides support in terms of case reports, best practices, national meetings,
and monthly supportive conferences calls with the surgeon champions and surgical clinical
reviewers.
questions as to the management and efficiency.9 In 1914, Codman started his own hospital in
Boston, the End-Result Hospital, to study the quality of surgical care.
Two precedents helped inform the rationale for ACS NSQIP. In 1994, in response to
concerns about high complication rates in the VA, a system to collect and report clinical
variables and outcomes across all VA sites was established (ultimately coined VA NSQIP).10,11
To compare results to non-VA hospitals or among VA sites fairly, these investigators needed to
correct for how sick patients were. Lisa Iezzoni coined the phrase, the algebra of effectiveness,
which means that outcomes are a function of patient factors, effectiveness of the care provided,
and random variation.12 Patient factors include the severity of target illness plus their
comorbidities. Thus, the VA developed a database of preoperative risk factors (severity of illness
and comorbidity) along with the database used to collect postoperative outcomes. In an attempt
to level the playing field for cross-institutional comparisons, the investigators developed riskadjustment models. The perceived success of the VA program led to its adaption for non-VA
hospitals, now known as the ACS NSQIP.
The second precedent for ACS NSQIP was the apparent success of programs like the New
York State Cardiac Surgery Reporting System (CSRS).13 In this program, which began about 20
years ago, the State Department of Health collects and distributes data from all New York State
hospitals performing coronary artery bypass grafting, with the aim of promoting accountability
and quality improvement. While originally envisioned as a confidential feedback system, public
and court pressure led to the public release of results, and the CSRS became a public reporting
system. Evaluation of the impact of the CSRS has been complicated by secular trends and the
lack of a widely-accepted comparison group, but systematic reviews of public reporting have in
general concluded that the preponderance of the evidence supports a causal relationship between
the CSRS and greater declines in cardiac surgery morbidity and mortality than would be
expected from the long-established secular trends alone.14,15 The basic concepts of the New York
State CSRS have been adopted by other states (California, Pennsylvania) and have expanded
across the U.S. through the efforts of the Society of Thoracic Surgery (STS) Registry, which
incorporated public reporting. Measurement and reporting of cardiac surgery outcomes has also
spread to England.16
The rationale for how NSQIP improves care is multifaceted. To improve care and reduce
complications, surgeons must first know the outcomes of their own procedures. The data used to
provide this feedback must be high quality and reliable, and the method of risk-adjustment must
be adequate to allay concerns about comparing apples to oranges. Lastly, the program must
establish an impetus for change: In this case, it is the comparison of care at ones own site to that
of other representative sites, that is sites with similar elements and the same risk-adjustment.
This comparison allows surgeons and hospitals to see how they compare in terms of poor
outcomes, which in turn promotes accountability and stimulates work to correct the problems.
Most sites, 59 percent of those surveyed, were unaware of their hospitals adverse event rates, let
alone how they compare to other hospitals, until after they enrolled in ACS NSQIP.4
One of the great strengths of the ACS NSQIP program is that it relies on prospective clinical
data. Administrative and claims data are limited, as they lack sufficient clinical data elements
and vary considerably in terms of quality (coder variations, subjective reporting, focus on
payment rather than on outcome reporting). A study comparing administrative and claims data
collected by the University Health System Consortium (UHC) program showed that the ACS
NSQIP identified 61 percent more total complications than were identified by UHC, including 97
percent more surgical site infections and 100 percent more urinary tract infections.17
142
Furthermore, NSQIP focuses on 30-day outcomes and is not limited to adverse events
associated with the index admission. Studies show that more than 50 percent of complications
happen after discharge. For colectomies, 45 percent of deep surgical site infections, 39 percent of
organ space infections, and 28 percent of deep venous thrombosis (DVT) occur after patients
have left the hospital.18 Identifying complications that occur outside the hospital is the
prerequisite first step to developing changes in care to help prevent those complication, which in
turn should result in reduced morbidity and mortality and save costs.19
143
University of Virginia
Massachusetts General
27
Hospital
Complication
19.3% SSI in diabetics; 8%
in non-diabetics
VTE 3.4% (2008)
Intervention
Glucose control protocol
VTE risk assessment and
order set
Hospital A
Hospital B
VTE 17.6%
Hospital C
Unplanned reintubation 3%
(O/E=1.56)
Ventilator >48 hrs 3.84%
(O/E=1.71)
Hospital D
Impact
Reduction of SSI O/E 1.31
to 0.78
Reduction of VTE rate
3.4% to 0.2% (2008-2009)
Reduction of SSI from
17.6% to 11.2% (36%
reduction)
(2003-2006)
Reduction of UTI 7.0% to
1.8%.
Morbidity O/E ratio went
from 1.19 [99% CI 0.93 to
1.48] to 0.93[(99% CI 0.67
to 1.48]
(76% reduction)
(2003-2004)
Organ space infection
increase attributed to
increased leak rates and
identified surgical
technique issues and saw
improvements, but rate still
high.
(2005-2010)
VTE decreased from 17.6
to 2.3%; O/E 1.88 to 1.05
(2006-2010)
TBD
Almost all these studies have a pre-post design, with no control groups, and therefore have
all the limitations common to studies of that design, including regression to the mean. Yet, in
aggregate, these reports consistently show that hospitals identified as high outliers in some
particular outcome that respond by implementing a targeted intervention experience a decrease in
that outcome. The magnitude of some of these decreases cannot be explained by regression to the
144
mean or confounding; for example the decrease in DVT rate from 10 percent to 1 percent in one
study or from 3.2% to 0.2% in another study.
Two published longitudinal studies have reached divergent conclusions on the effects of
reporting. The first study looked at changes over three years (2005 to 2007) in ACS NSQIPparticipating sites (N=183) for all outcomes measured and surgical specialties using riskadjustment and accounting for hospital procedure volume (Hall B, 2009).19 For the most recent
time period, 2006 to 2007, 118 hospitals were enrolled and were participating long enough to
produce clinically useful data. The authors found that 82 percent (97 of 118) of hospitals had
improved morbidity and 66 percent (78 of 118) had improved mortality. The adjusted absolute
difference in observed to expected (O/E) ratio was -0.114 for morbidity and -0.174 for mortality
(negative numbers meaning less morbidity and mortality). Similar results were seen when the
researchers accounted for institutional volume. They also found that the number of high outliers
(those with worse outcomes) decreased over time and the number of low outliers (those with
better outcomes) increased. Additionally, high outliers were more likely to improve and had
larger mean changes in outcomes. For large hospitals, it was estimated that an average of 200 to
500 complications and 12 to 36 deaths may have been avoided.19 This study was conducted by
investigators affiliated with ACS NSQIP.
The other study compared ACS NSQIP to another private sector collaborative, which was
based at the University of Michigan Medical Center.28 The data file provided by ACS NSQIP
included Michigan and non-Michigan hospitals. The Michigan Surgical Quality Collaborative
(MSQC) includes 34, largely community (68 percent) hospitals, unlike the ACS NSQIPparticipating hospitals, which are primarily academic/teaching. Sixteen MSQC hospitals were
included in the analysis, which assessed two time periods: April 2005 to March 2007 compared
with April 2007 to December 2007. Results were also compared with the 126 non-Michigan
NSQIP hospitals over the same time period. All hospitals used a similar data reporting system.
This analysis found that the MSQC hospitals had a decrease in morbidity from 10.7 percent to
9.7 percent (9% reduction, P=.002; odds ratio=.898) over the 3 years, whereas morbidity did not
change for the ACS NSQIP hospitals in either time period (12.4%; odds ratio=1.0). Mortality
rates did not change for either group of hospitals.28 This study was conducted by investigators
affiliated with MSQC. One possible explanation for the difference in results between the two
studies is that the length of time the hospitals were enrolled in the program may have been too
short to see improvements, especially in the larger hospitals, which predominated among the
non-Michigan ACS NSQIP hospitals.
145
performs a small volume of procedures might need longer follow-up, possibly more than a year,
to accurately assess quality.32 Concerns have also been raised that surgeons could alter treatment
plans or surgical options for patients based on their operative risk rather than give the patient the
option of a procedure with a potentially better long-term functional outcome. An example would
be in vascular surgery, where a high-risk patient eligible for a distal bypass would be
recommended an amputation instead.
146
program can use less effort. FTE calculators are available on-line to calculate the amount
required.
Table 2, Chapter 14. Comparison of American College of Surgeons national surgical quality
improvement program use options
Elements
Classic
Essential
Eligibility
Any hospital
Any hospital
Variables
Collected
69 clinical variables
46 clinical variables
(subset of Classic)
Surgical Specialty
Versions
General/Vascular
Multispecialty
General/Vascular
Multispecialty
Multispecialty
Case Volume
Requirements
General/Vascular
=1680 cases/yr (or all
if <1680)
Multispecialty=20%
case volume by
specialty
General/Vascular =40
cases/8 day cycle
Multispecialty=may be
>40 per 8 day cycle
General/Vascular
=1680 cases/yr (or all
if <1680)
Multispecialty=20%
case volume by
specialty
General/Vascular =40
cases/8 day cycle
Multispecialty=may be
>40 per 8 day cycle
Maximum=1680
cases/yr
Case Sampling
Procedure
Targeted
Any hospital
Core of 46 clinical
variables +
Procedural specific
variables (same as
Essential)
General/Vascular
Multispecialty
Minimum =1680
cases/yr (depends
on # and volume)
The minimum number of cases the surgical clinical reviewer will enter is 1,680 (however,
this number may be smaller or larger, depending on the program chosen). If data can be entered
automatically from the electronic medical record, then an estimated 2,000 to 2,300 cases can be
reviewed per year.
Currently, 431 sites are enrolled in ACS NSQIP, which represents roughly 10% of the almost
4,500 hospitals in the United States. A geographic map of 258 sites that had reported clinically
useful data (from the July 2011 semi-annual NSQIP report) shows the distribution within and
outside of the U.S. (Figure 2).
147
Figure 2, Chapter 14. Geographic distribution of American College of Surgeons national surgical
quality improvement program participating sites
Figure 2 provided from ACS NSQIP Web site, from July 2011 semi-annual NSQIP report, accessed December, 2011
Reprinted by permission from the American College of Surgeons NSQIP.
Of participating sites, 49 percent are teaching or academic centers. The majority of these
hospitals are high volume, as only 3 percent perform fewer than 100 cases per year, 7 percent
perform 100-299 cases per year, 43 percent perform 300-499 cases per year, and 47 percent
perform more than 500 cases per year. This skewed distribution of hospital size means that the
10 percent of hospitals participating in ACS NSQIP represent 32 percent of the procedures
performed (based on Medicare data from ACS NSQIP, personal communication with Clifford
Ko). Certain complex procedures are captured at an even higher rate, for example 57 percent of
esophagectomies and 53.4 percent of pancreatectomy cases billed to Medicare are performed at
ACS NSQIP sites (Table 3).
Table 3, Chapter 14. Percent of Medicare surgical cases covered by the national surgical quality
improvement program
Procedure
Esophagectomy
Cystectomy
Abdominal Aortic
Aneurysm Repair
Pancreatectomy
Colectomy
Proctectomy
AortoIliac bypass
LEB
Liver Resection
Hip Fracture Repair
Abdominoplasty
Lung Resection
1158
3346
MC cases
NOT in
NSQIP
875
4501
3762
3901
32444
6745
2255
12203
2465
40030
1058
16065
MC cases in
NSQIP
2033
7847
Percent
covered by
NSQIP
57.0
42.6
6448
10210
36.8
3399
103056
15767
4974
30100
2201
151140
1829
27391
7300
135500
22512
7229
42303
4666
191170
2887
43456
53.4
23.9
30.0
31.2
28.8
52.8
20.9
36.6
37.0
Total MC
cases
148
Table 3, Chapter 14. Percent of Medicare surgical cases covered by the national surgical quality
improvement program (continued)
Procedure
MC cases in
NSQIP
MC cases
NOT in
NSQIP
Total MC
cases
Percent
covered by
NSQIP
Endovascular
Abdominal Aortic
8944
17324
26268
34.0
Aneurysm Repair
Nephrectomy
9727
16375
26102
37.3
Hysterectomy
17954
45108
63062
28.5
Total Hip
56700
195528
252228
22.5
Arthroplasty
Laminectomy
60650
154858
215508
28.1
TURP
11345
42928
54273
20.9
Ventral Hernia
19360
57735
77095
25.1
Carotid
20588
59710
80298
25.6
Endarterectomy
Total Knee
72916
279642
352558
20.7
Arthroplasty
Prostatectomy
10677
18808
29485
36.2
Breast recon
455
700
1155
39.4
Appy
8802
31635
40437
21.8
Thyroid
5358
12598
17956
29.8
Gastrectomy
3782
7382
11164
33.9
Carotid stent
3648
7883
11531
31.6
Small Bowel
10784
30836
41620
25.9
Resection
Mastectomy
6417
21378
27795
23.1
Cholecystectomy
29386
117327
146713
20.0
Total
32.0
Notes: LEB=Lower extremity bypass; TURP= Transurethral resection of the prostate.
Collaboratives are a main feature of the ACS NSQIP. The collaboratives have taken many
different formsa handful of geographically close hospitals or all of the hospitals in a state
that work together as a team to implement the program and initiate quality improvement. They
also can represent a disease or patient population; thus a collaborative of hospitals need not be
geographically close. Collaboratives provide a collective voice for bargaining with potential
sources of funding. One reported approach is for the main insurer for the hospitals in the
collaborative to pay for 50 percent of the cost of the program over a set number of years.
Sometimes an option to renew the financial support is given if certain milestones are met or
improvements are shown. Some payors have judged there to be a business case for helping
support ACS NSQIP participation due to perceived cost-savings (detailed below). Table 4 shows
the current list of active collaboratives in ACS NSQIP.
Table 4, Chapter 14. List of American College of Surgeons national surgical quality improvement
program collaboratives including type, number of sites, and payor
Group
Canadian National Surgical Quality
Improvement Collaborative (CAN-NSQIP)
Connecticut Surgical Quality Coalition
(CTSQC)
Department of Defense/TRICARE
Florida Surgical Care Initiative (FSCI)
Type
# of Sites
Payor Involvement
Regional
Regional
16
Department of Defense/TRICARE
63
Systemwide
Regional
149
Table 4, Chapter 14. List of American College of Surgeons national surgical quality improvement
program collaboratives including type, number of sites, and payor (continued)
Group
Fraser Health Systems (Canada)
Illinois Surgical Quality Improvement
Collaborative (ISQIC)
Kaiser Permanente Northern California
Regional NSQIP Collaborative
(KPNCRNC)
Kaiser Permanente Southern California
Regional NSQIP Collaborative
(KPNCRNC)
MaineHealth Collaborative
Mayo Clinic Surgical Quality Consortium
(MCSQC)
Northern California Surgical Quality
Collaborative (NCSQC)
Nebraska Collaborative
Oregon NSQIP Consortia
Pennsylvania NSQIP Consortia
Partners HealthCare
Type
Systemwide
# of Sites
3
Regional
12
Systemwide
21
Systemwide
MaineHealth
Mayo Clinic
Regional
Regional
Regional
Regional
Systemwide
2
8
10
Systemwide
Systemwide
Payor Involvement
A pilot pediatric collaborative for ACS NSQIP collects data for patients under age 18.34,35
Variables have been modified to pediatric surgery practices and needs.
Henry Ford hospital recently reviewed their lessons learned after implementing ACS NSQIP
over 5 years ago.36 Their findings were summarized into 12 steps (Figure 3).
150
Figure 3, Chapter 14. The 12 steps to implement the national surgical quality improvement
program
Adapted from Velanovich V, Rubinfeld I, Patton JH Jr, Ritz J, Jordan J, Dulchavsky S. Am J Med Qual. 24(6):474-9. 2009 by
Sage Publications. Reprinted by Permission from the SAGE Publications.
hospital. Examples of such savings reported by NSQIP sites are shown in Table 5. Shown are
pre-post data without control groups; thus, inference of a causal relationship is limited by the
study design.
Table 5, Chapter 14. Example of reductions in complications and associated costs
Hospital
37
Surrey Memorial Hospital
Henry Ford Hospital
36
22
VA
Complication Reduction
Reduced SSI over four years: 13%,
10%, 7.5%, 7.2%
Reduced LOS by 1.54 days over 4
years for general surgery, vascular
and colorectal procedures
Surgical pneumonia alone
Savings
$2.54 million savings
$2 million annual savings (increased
billing by $2.25 million/yr as
underbillings were identified )
9.3 million in savings annually
Respiratory complication
152
around the same time. Then at the State Chapter ACS meeting, the idea was posed to develop a
statewide collaborative and use a funding mechanism modeled after that in Michigan, where
some funding support would be provided from major payors in the area (in this case, Blue
Cross/Blue Shield). Discussions with the payor were initiated, and the surgical leaders made a
site visit to Michigan to learn more about developing a collaborative. The collaborative included
the hospitals (and the chapter), Blue Cross/Blue Shield (BC/BS), and the Tennessee Hospital
Association. Ultimately, it was decided that control of the collaborative (meaning the data)
would remain with a leadership committee that comprised four surgeons who were appointed by
the Chapter, along with two hospital CEOs and one member of the Tennessee Hospital
Association. The proposal included funding for participation of eight hospitals (three hospitals
that were currently enrolled and five new ones) estimated to be $2,550,000 for 3 years. The
money would cover half the expense of the surgical clinical reviewer at each site, and provide for
some salary support for each site surgeon champion ($5,000 each per year) and the cost of the
administrative Tennessee Center for Patient Safety (TCPS), which would house the data. By
2008, three more hospitals wanted to join, and BC/BS increased their support to include them.
This example highlights many of the key components to building a successful programsurgical
leaders taking a role, supportive administration, collaborating with other hospitals.
A new and strikingly different collaborative is underway in Florida. The story of how this
collaborative started was outlined at the national meeting. In brief, the drive to participate in
NSQIP and improve care started with the hospitals. The Florida Hospital Association (FHA) was
aware of the high surgical mortality demonstrated by the Dartmouth Atlas project (which is
dedicated to identifying and showing disparities in access and utilization of health care) in their
State. The FHA, along with the payor, BC/BS, collaborated to generate a financial incentive for
hospital participation. According to the model, the first step was to identify the surgeon
champions. In order to make the program financially viable to many of the smaller hospitals and
to reach more hospitals, the ACS NSQIP along with the FHA devised a new version of the
program that would collect only four outcomes, thus lowering costs. Currently, 64 hospitals are
participating in the Florida Surgical Care Initiative (FSCI) and participation of 39 more is
pending. This example demonstrates additional features that help encourage participation:
individuals at the State level and hospital administration taking a lead, flexible program design to
fit the needs of the collaborative, and the role of the local payor to incentivize hospital
participation.
153
be effective across sites) are similar to those of other successful patient safety practices such as
the Michigan Keystone ICU Project to reduce catheter-related bloodstream infections.39 Despite
ACS NSQIP and the Keystone ICU Project having started with different original interventions
(the feedback of procedure-specific surgical outcome data to surgeons and a checklist of
processes to reduce infections) the observation that the current version of the interventions
include so many similar components probably suggests something generalizable about the
implementation of certain kinds of practices across hospitals.
ACS NSQIP provides hospitals and providers with usable clinical data that are otherwise not
available to them. Currently, all hospitals use administrative data to some degree to assess
quality through the Centers for Medicare and Medicaid Services Hospital Compare program or
the Surgical Care Improvement Project (SCIP). These data lack clinical information and are
limited by the variables reported for claims. More importantly, the correlation between
administrative data and actual complications or diagnoses is inadequate. For example, urinary
tract infections are poorly reported in administrative data. Furthermore, studies show that
adherence to SCIP measures do not correlate to better outcomes. ACS NSQIP has the power to
show providers the most problematic clinical data.
The greatest benefit has been seen in the larger hospitals in the procedures with the higher
complication rates. Whether the above improvements will translate to low risk but common
procedures, such as out-patient procedures is unclear. Also, most of the early adopters have been
large academically affiliated hospitals. How successfully and widely it can be implemented at
smaller hospitals remains to be seen. A summary table is located below (Table 6).
Table 6, Chapter 14. Summary table
Scope of the
Strength of
Problem Targeted by Evidence for
the PSP
Effectiveness
(Frequency/Severity) of the PSPs
Common/High
Moderate-tohigh
Evidence or
Potential for
Harmful
Unintended
Consequences
Low
Estimate of
Cost
Moderate
Implementation Issues:
How Much do We
Know?/How Hard Is it?
Moderate/Moderate
References
1.
2.
154
3.
4.
5.
6.
14.
7.
15.
16.
17.
18.
19.
20.
8.
9.
10.
11.
12.
13.
155
21.
22.
23.
24.
25.
156
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
157
38.
39.
Introduction
Leaving surgical items inside patients is a rare but potentially deadly mistake. The most
common such item is a surgical sponge. Researchers at the Mayo clinic found that during the
four-year period from 2003 to 2006, the rate of retained foreign objects was 1 in every 5,500
operations,1 and 68 percent of the retained objects were sponges. The greatest subsequent risk to
the patient is infection, which can be fatal. Other risks include perforations and granulomas.
Risk factors for such incidents were explored by Gawande and colleagues (2003),2 who
examined factors surrounding 42 retained sponges and 19 retained instruments. The majority
required reoperation, and one patient died. Compared with control incidents, item retention was
more likely to occur in the context of emergency surgery, an unexpected change in surgical
procedure, high body-mass index (an x-ray is recommended), and the lack of item counts.
A review of the literature on the topic of retained surgical sponges conducted for the original
report identified only one study (a case series) that attempted to assess the use of sponge and
instrument counts to prevent retention.3 The goals of the present review were to identify
interventions implemented since the previous review and to report on studies assessing their
effectiveness. We conducted a review of the literature from 2001 to 2011 and reviewed all
studies relevant to methods used to prevent surgical items from being left inside patients during
surgical procedures.
without unique item ID numbers (abbreviated RF), and (3) radiofrequency tagging with unique
item ID numbers (RFID). Bar coding is an established and low-cost technology, but a direct line
of sight between the bar code scanner and the items bar code label is needed in order to scan it,
and blood-soaked items may be difficult to scan accurately. The RF technologies (a penny-sized
or smaller chip implanted into the device) allow items to be detected by a specialized wand that
is waved over the patients body during and after the procedure. This scan can prevent the need
for a radiograph, which itself can increase surgical risk because of the added time on the
operating table and under anesthesia. RFID represent an advance from the simpler RF
technologies because if each item is assigned a unique ID number, then the manual count can be
checked against the RFID systems baseline count.
The FDA has cleared four products relevant to the above technologies for marketing in the
United States (U.S.):
1. Safety-Sponge System, (SurgiCount Medical, Temecula, California).6 This system
comprises bar-coded sponges.
2. RF Surgical Detection System (RF Surgical, Bellevue Washington). This technology
permits detection of devices but does not provide a count, because items do not receive
unique ID numbers. The detection wand is single use. The system can be used with
sponges, laparotomic pads, gauze, and towels, but not surgical instruments or sharps.
3. SmartSponge System (Clear Count Medical, Pittsburgh Pennsylvania). This system
assigns a unique ID for each device, so it is used for both detection and counting. As the
procedure progresses and staff remove sponges or other items, they put the items into a
specialized bucket fitted with an antenna that detects and counts the RFID items. If a
discrepancy occurs between the baseline counts and the final counts, it notifies the OR
team with auditory and visual warnings, thereby initiating a search for the lost item(s)
using a detection wand. This system also can be used with sponges, laparotomic pads,
gauze, and towels, but not surgical instruments or sharps.
4. ORLocate (Haldor, Boston, Massachusetts).7 This system also assigns a unique ID for
each device, so it is used for both detection and counting. Unlike the two systems
described above, it can be used for instruments and sharps as well as sponges and other
non-metallic items
listed additional factors that could help minimize this type of mistake: (1) Knowledge of risk
factors, (2) Use of modern technology, (3) Improved perioperative patient processing systems.
160
always X-ray before closure; 3. Count, and also X-ray before closure only for high-risk
operations).
Low
Evidence or
Potential for
Harmful
Unintended
Consequences
Negligible
Estimate of
Cost
Implementation Issues:
How Much do We
Know?/How Hard Is it?
References
1.
2.
3.
4.
161
5.
6.
7.
8.
9.
10.
11.
162
Introduction
Patient monitoring is one of the central tasks in operating rooms (ORs). Because of the
increasing use of advanced technologies in surgical procedures, todays ORs are commonly
crowded with freestanding devices, support systems, and monitors. In addition to the traditional
monitors that continuously present the patients hemodynamic, respiratory, and
electrophysiological signals, many innovative devices recently introduced into the OR feature
their own platforms for data display. These devices may fall into one of four categories:1
Surgical machine-controlled applications (e.g., robotics, minimally invasive surgery,
video-endoscopic surgery, master-slave systems)
Designated diagnostic and real-time navigation devices (e.g., magnetic resonance
imaging, computed tomography, three-dimensional ultrasound)
Information technology (IT) applications generating a real-time connection between the
OR and the hospital medical record archives (e.g., picture archiving and communication
systems [PACS], videos, electronic medical records (EMRs), hospital information
systems, and other laboratory data)
Telecommunication and teleconferencing systems connecting the OR in real time with
other medical centers
The increasing use of these devices leads to a congestion of data displays in the OR, compels
OR staff members to increase the time devoted to monitoring the displays, and may divide their
attention between monitoring and other tasks.1 Meanwhile, the proliferation of freestanding
devices and displays in the OR makes coordination difficult.2 Surgeons, nurses, and
anesthesiologists have their own perioperative devices or systems on which to focus. Human
coordination of multiple electromechanical devices may lead to misunderstandings and delay
action.
As some experts have commented, what is currently lacking in most ORs is a high-level
overview of all the information that is already available in the room.3-5 The lack of integration of
patient data makes ORs inefficient, overcrowded, and less safe. When patient data are not
displayed to caregivers in an integrated fashion, OR staff have to frequently with multiple
displays to obtain updates and exert control over the various devices at their disposal.4,5
OR integration is an emerging technology that has the potential to address the long-standing
problem of segregated data display in ORs. This technology organizes and consolidates patient
data for clinicians during a surgical procedure. This chapter focuses on the latest form of the
technology, which features a centralized data display platform.
163
also securely archive all case data, including video, using various device and information
protocols to permit case review after the surgery.7,8 With additional software provided by Global
Care Quest, the ICIS Dashboard provides secure access to the collected patient data in real time
through wireless mobile devices, including personal digital assistants and smart phones that
clinicians can use where a wireless connection is available.7
Contribution to operative and perioperative safety. OR integration technology with
centralized data display can potentially help improve operative and perioperative safety in
several ways. First, the technology allows easy, just-in-time access to patient information from
disparate devices or systems that is often unseen, unrealized, or unused.8,9 Increased access to
this information may improve team situation awareness (TSA, i.e., the task- and team-oriented
knowledge held by everyone in the team and the collective understanding of the unfolding
situation).8,10 TSA is one of the critical factors in OR teamwork that can affect patient safety and
quality of care. Augmented TSA can improve communication among clinical personnel and thus
help reduce the number of medical errors.8,11,12
Second, the integration of previously isolated information sources may open new
opportunities for decision support and augment vigilance.8 For example, allergy information
from the hospital information system may alert the team to not administer certain drugs to the
patient and, thus, prevent harmful drug-related adverse events. Information from the laparoscopic
insufflator can inform the team of impending asystole from insufflation. Information from the
location tracking system may help the surgical team check the accuracy of patient identity.
Integration with the order-entry system can help update the team on workflow and resource
acquisition such as pathology, radiology, and the blood bank. The order information can be
continuously displayed throughout the operative period to decrease uninformed or delayed
decisions.
Additionally, the OR integration technology provides the ability to flexibly change the source
or destination of an AV signal without requiring the cumbersome process of reconfiguring direct
links between sources and destinations each time such a change is needed.6 This ability might
decrease the risk of medical errors in the reconfiguration process. OR integration may also
generate other opportunities for improving patient safety. For example, the technology might
allow real-time, remote consultation from experts outside of the OR.6 The technology could
enhance patient data collection during the surgery and decrease stale or duplicate data. These
data can be analyzed later for the purpose of improving patient safety. The technology may also
have a positive psychological effect on clinical personnel, making them feel more comfortable
and more confident that things are going well.10
165
The second study discussed the technical issues that must be addressed when installing
integrated ORs.14 The issues included controlling the images, integrating team members, preconstruction planning, working with vendors, and managing the final project phases. It was not
feasible to provide additional details about these studies in this brief review. Readers can refer to
the original studies for detailed instructions on implementing integrated ORs.
Data about costs. According to an ECRI Institute study, as of October 3, 2007, a LiveData ORDashboard system costs about $150,000.7 Total system costs can vary widely depending on the
features that a hospital requests and the number of systems installed at a facility.6,18 Facilities
could also face significant additional costs to integrate new systems into their existing IT
infrastructure. Similar cost information for the ICIS Dashboard was not reported.7
Effect of context on effectiveness. We identified a survey of 17 surgeons and 9 scrub nurses
from a single hospital that evaluated their satisfaction after 2 years of use of integrated ORs.13
The surgeons and scrub nurses agreed that a great degree of education and a cultural change were
needed to use the system in a correct and complete way.13 However, we were not able to verify
whether the integrated ORs described in the study had the centralized data display feature. We
did not identify any other study that evaluated the effect of context on the effectiveness of an
integrated OR and centralized display systems in improving patient safety.
166
Low
Evidence or
Potential for
Harmful
Unintended
Consequences
Negligible
Estimate of
Cost
Moderate
Implementation Issues:
How Much do We
Know?/How Hard Is it?
Moderate/Moderate
References
1.
2.
3.
4.
167
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
168
Introduction
Myocardial infarction and cardiovascular death are the most common complications of major
non-cardiac surgery; thus, they have long been a focus of preoperative evaluations and a target of
perioperative management strategies. Based on strong evidence linking myocardial ischemia
with postoperative myocardial events and preliminary evidence that beta-blockade blunts
electrocardiographic signs of ischemia, clinical researchers in the late 1990s began examining
the effects of perioperative beta-blocker administration on patient outcomes. The 2001 report1
reviewed the evidence up to that point regarding the effectiveness, safety, and cost-effectiveness
of this intervention. Based on the results from several well-designed clinical trials, the authors
concluded that use of beta-blockers in the perioperative period was associated with significant
reductions in patient cardiac morbidity and mortality. However, as of publication of that report,
many questions remained regarding the optimal type of beta-blocker, the patients most likely to
benefit, and the safest and most effective dosing regimen.
myocardial infarction, and non-fatal cardiac arrest (in other words, the exact composite outcome
with the statistically significant effect in the earlier meta-analysis). Indeed, at 30 days, patients
receiving metoprolol had a hazard ratio (HR) for the primary outcome of 0.84 (95% CI 0.70 to
0.99), due primarily to fewer myocardial infarctions. However, patients treated with metoprolol
had a statistically significantly greater risk of stroke (HR 2.17, 95% CI 1.26 to 3.74), and, even
more alarmingly, a greater risk of all-cause death (HR 1.33, 95% CI1.03 to 1.74). The authors of
POISE concluded that the perioperative use of beta-blockers has both benefits and risks. For
example they calculated that for every 1,000 patients undergoing noncardiac surgery, the use of
extended-release metoprolol would prevent 15 patients from having a myocardial infarction and
three from undergoing cardiac revascularization, but that there would be eight extra deaths and
five extra strokes. Based on these differences in benefits and harms, and on the potential for
patients to place different values on these outcomes, the authors of POISE concluded that authors
of current guidelines advocating the use of beta blockers should reconsider their
recommendations.
The later systematic review/meta-analysis was published in 2008, and included the POISE
results.4 This review, which scored 11 out of 11 relevant domains in AMSTAR, included 33
trials, now encompassing 12,306 patients. Recalling that POISE contributed more than 8,000
patients alone, in most of the pooled analyses the POISE results contribute 75 percent or greater
weight to the pooled result. Unsurprisingly, the meta-analysis found statistically significant
benefits for treatment for the outcomes of non-fatal myocardial infarction and myocardial
ischemia, nonsignificant results for all other potential benefits, and statistically significant
adverse effects for nonfatal stroke (pooled odds ratio[OR] of 2.16), perioperative bradycardia
requiring treatment (pooled OR of 2.74), and perioperative hypotension requiring treatment
(pooled OR 1.62). The effect on mortality was adverse, but did not reach statistical significance
(pooled OR 1.20, 95% CI 0.95 to 1.51). The authors of this review concluded that evidence
does not support the use of beta blocker therapy for the prevention of perioperative clinical
outcomes in patients having non-cardiac surgery.
170
Evidence or
Potential for
Harmful
Unintended
Consequences
High evidence High (death,
we know
stroke,
harms may
hypotension and
equal or
bradycardia
exceed
benefits
Estimate of
Cost
Low
References
1.
2.
3.
4.
5.
171
Implementation Issues:
How Much do We
Know?/How Hard Is it?
N/A
Introduction
Central venous catheters (CVCs) have multiple indications, including parenteral nutrition,
treatment of intravascular depletion, access for vasoactive medications, hemodynamic
monitoring, intravenous access during cardiopulmonary arrest, difficult peripheral intravenous
(IV) access, and long-term IV access for medications, such as antibiotics.1,2 Although these
catheters can be life saving, they are also associated with significant risk.3 This risk is heightened
by a number of factors, including patient characteristics (e.g., morbid obesity, cachexia, or local
scarring from surgery or radiation treatment), patient setting (e.g., patients receiving mechanical
ventilation or during emergencies such as cardiac arrest), co-morbidities (e.g., bullous
emphysema or coagulopathy), the variable training and experience of the clinicians who perform
the procedure, and the method of insertion (e.g., percutaneous insertions are often performed
blind and rely on anatomic landmarks).3-5 However, protocols have been developed that use
portable ultrasound (US) devices to provide bedside imaging of the central veins during catheter
placement. The advantages associated with US-guided CVC placement include detection of
anatomic variations and exact vessel location (for example, the carotid artery is anterior to the
internal jugular vein in 3% to 9% of patients6), avoidance of central veins with pre-existing
thrombosis that may prevent successful CVC placement, and guidance of both guidewire and
catheter placement after initial needle insertion.
The original report included a review of the efficacy, safety, and cost-effectiveness of realtime US guidance on the safety of CVC insertions. This review found that, in general, US
improves the success rates and reduces the risks of CVC placement, particularly for
inexperienced clinicians and for patients in high-risk situations. The purpose of the present report
is to provide an update on the impact of US CVC insertion. We used the articles cited as
evidence in the 2001 report to create a list of search terms and then used these terms to conduct
an update search.
172
placement, failure on the first attempt, mean number of attempts to successful catheterization,
and seconds to successful catheterization), two-dimensional US had statistically significantly
better outcomes than the landmark method for internal jugular vein catheterization in adults.
More limited data in children and for subclavian and for femoral vein insertion favored the use of
two-dimensional US. Pooled results from studies of Doppler US also favored its use. No studies
directly compared two-dimensional and Doppler US. The authors made an indirect comparison
by assessing the size of the pooled effects for each compared with the landmark method. This
analysis favored the use of two-dimensional US. This review scored nine of 11 relevant
AMSTAR criteria. A companion cost-effectiveness analysis estimated the marginal cost (in
2002) for use of US in CVC to be about 10 pounds sterling (approximately $16) per procedure,
assuming the machine was used for 15 procedures each week. The base case scenario estimated
that for every 1000 patients, 90 complications would be avoided, with a net cost saving of about
2000 pounds sterling (approximately $3200).16
Since that time, randomized trials in adults have consistently supported the conclusions about
effectiveness, including patients treated in the Emergency Department,17 ventilated patients,18
critical care patients,19,20 and patients in other miscellaneous clinical settings.21,22 A new
outcomecentral venous catheter-associated bloodstream infectionhas been assessed and
found to be statistically significantly lower in one trial of US-guided catheter insertion compared
with landmark methods.19
A more recent meta-analysis included five studies that focused only on children, most of
whom were cardiac surgery patients. Although pooled point estimates favored the use of US, the
95% confidence intervals were wide and none of the results were statistically significant.23 Two
trials published since that meta-analysis, one of which compared real time to static US, both
found that two-dimensional real-time US improved some outcomes.24,25
Recent trials of US have focused less on its use in adult internal jugular vein catheterization
and more on its use in other locations and refinements of the technique, including the insertion of
hemodialysis catheters,26 the radial artery,27-29 the femoral artery,30 and even peripheral venous
catheters in difficult patients.31-36 In general, studies reported that US guidance improved
outcomes compared with techniques without US guidance. Systematic reviews of the use of US
guidance for hemodialysis catheter insertion37 and radial artery catheters38 each concluded that
the use of real-time two-dimensional US improved outcomes.
Clearly, USG central line placement education varies in undergraduate, graduate, and
continuing medical education. While educators at all levels are making inroads, greater
consistency is needed in curricula, evaluation of outcomes, and guideline development.
174
High
Evidence or
Potential for
Harmful
Unintended
Consequences
Negligible
Estimate of
Cost
Low-tomoderate
Implementation Issues:
How Much do We
Know?/How Hard Is it?
A lot/Moderate
References
1.
2.
3.
10.
11.
12.
13.
14.
15.
16.
4.
5.
6.
7.
8.
9.
175
17.
25.
18.
26.
19.
27.
28.
29.
30.
31.
32.
33.
20.
21.
22.
23.
24.
176
34.
35.
177
36.
37.
38.
178
Post fall review: to assess potential reasons for a specific instance of a fall and to
remediate possible contributing factors
Patient education
Staff education
Footwear advice
Scheduled and supervised toileting
Medication review: to assess for use of medication(s) that can affect mental alertness and
balance (see ProFANE taxonomy for further details, Appendix C).
The most recent Cochrane review notes a striking variability in type, targeting, intensity,
and duration within the fall prevention programs and does not attempt to draw conclusions
about which components might be most effective.8 Table 1 lists all the studies in the reviews by
Cochrane and by Oliver, as well as new studies from our update search, and the components
included in the intervention.
All multicomponent interventions also included an assessment of falls risk. In about 60
percent of studies this was a formal falls risk assessment tool such as the Morse Fall Scale or
STRATIFY, and the remainder used informal or idiosyncratic or unstated methods for assessing
patients at increased risk of falls.
Other single intervention components include use or removal of bedrails, use of physical
restraints, movement alarm devices, low-low beds (beds closer to the floor), exercise or
additional physical therapy, increased observation or assistance, calcium or vitamin D, hip
protectors, and prevention of delirium (this last topic is covered in Chapter 20). Since most
reviews conclude that multi-component interventions are more effective than single components,
in this chapter we will consider only multi-component interventions. Multicomponent
interventions are also referred to in the literature as multifaceted or multifactorial interventions.
Although some authors draw distinctions between these labels, we will not do so here, and refer
to all of them as multicomponent.
179
22%
()
()
?
Postfall
Review
Movement
Alarms
Medication
Review
Urine
Screening
Exercise
Toileting
Schedules
Footwear
Schwendimann et
al, 200622
Stenvall et al,
200723
Grenier-Sennelier
et al, 200216
Haines et al, 200417
Healey et al, 200418
Cumming et al,
200813
Dykes et al, 201014*
Fonda et al, 200615
Other Interventions
Staff
Education
Patient
Education
Bedrail
Review
Vest/Belt/Cuff
Restraint
Hip
Protectors
Alert
Wristband
Bedside Risk
Sign
References
Environment
Modified
Table 1, Chapter 19. Components of multi-factorial falls prevention trials in hospitals, 1999 to 2009
180
Low beds; interventions specific to each risk factor in model used. Used
Hendrich II Falls Risk Model
Low beds; Introduction of a computerized falls reporting and analysis
c
system Used STRATIFY falls risk assessment tool
Risk Factors assessed
Falls history and continence assessment added to standard admission
documentation / Unstated method of risk assessment
Modification of tool developed the Centre for Education and Research on
Ageing in Sydney, Australia
Tailored plan of care; computerized Fall Prevention Tool Kit (FPTK)
Used Morse Fall Scale
Low beds, volunteer observers
Used Falls Risk Assessment Scoring System
Improved assessment of mobility and self-efficacy
Unspecified method for assessing risk
Used the Peter James Centre falls risk assessment tool
Vision testing, lying and standing blood pressure
Brief falls risk factor screen
Stand by me notices to prompt staff to wait outside toilets ready to assist.
Mobility level signs at bedside
Unstated method of risk assessment
Used Morse Falls Scale
Nursing and medical checklist for remediable risk factors, content not
described and compliance poor
Used STRATIFY falls risk assessment tool
Briefly screened for falls risk using 3 items
Additional therapy and nurse staffing Routine dietary protein
supplementation Protocol driven delirium screening
No clear risk assessment instrument, but population can be assumed to all
be at elevated risk
Career education
A new formal risk assessment instrument created for the study
Identification of high risk patients on the basis of a recent fall or 4 other
criteria
Postfall
Review
Movement
Alarms
Medication
Review
Urine
Screening
Exercise
Toileting
Schedules
Other Interventions
Footwear
Staff
Education
Patient
Education
Bedrail
Review
Vest/Belt/Cuff
Restraint
Hip
Protectors
Alert
Wristband
Bedside Risk
Sign
References
Environment
Modified
Table 1, Chapter 19. Components of multi-factorial falls prevention trials in hospitals, 1999 to 2009 (continued)
Vassallo et al,
Bedside commodes
and Krause, 200727
Used STRATIFY falls risk assessment tool
Table adapted from Oliver 1
* New studies added from update search
yes = component included within the intervention; (yes) = component planned but not implemented; ? = component implied but not explicit; = intervention discouraged use of this component;
= intervention encouraged use of this component.
a
(yes) indicates intervention in design but not applied in practice (e.g., environmental hazards identified but not addressed). ? indicates that the article implies, but does not specify, that an intervention
was included. For bedrails and body restraints, indicates the intervention was to discourage their use, indicates the intervention aimed to encourage their use, while yes indicates either direction
not described or a neutral risk versus benefit review was required.
b
Where interventions are described that would be considered very standard practice for control as well as intervention (e.g., call bell left in reach, walking aids provided as appropriate), these are not
listed.
c
This potentially confounded the findings as this changed the method of collecting outcome data on falls at the same time as the intervention was introduced.
Reprinted from Clin Geriatr Med. 26(4), Oliver D, Healey F, Haines TP., Preventing falls and fall-related injuries in hospitals, 645-92, 2009 with permission from Elsevier
181
182
(a) Firm mattresses; low beds; appropriate chair heights and depths for easy transfer; chairs with arm rests; and secured handrails throughout the movement of a patient. (b) Nonslip surfaces in floors/bathtubs; shower seats; grab bars next to the toilet/bathtub; toilet seats that allow easy transfer; door magnets that hold doors in the open position; and arm
rests next to the toilet.
*An intervention or a factor whose efficacy was NOT tested as a single factor in any healthcare setting. **An intervention or a factor whose efficacy was tested as a single factor in
other healthcare settings but NOT specifically in a hospital setting. ***An intervention or factor whose efficacy was tested in a hospital setting.
Figure taken from Choi et al, 201128
Choi YS, Lawler E, Boenecke CA, et al. Developing a multi-systemic fall prevention model, incorporating the physical environment, the care process and technology: a systematic
review. J Adv Nurs. 2011. Permission granted by John Wiley & Sons, Inc.
183
184
Figure 2, Chapter 19. Meta-analysis from Oliver et al. 2006 for multifaceted interventions in
30
hospital falls (random effects model)
Reproduced from Strategies to prevent falls and fractures in hospitals and care homes and effect of cognitive impairment:
systematic review and meta-analyses. Oliver D, Connelly JB, Victor CR, et al. 334(7584):82. 2007 with permission from BMJ
Publishing Group Ltd.
The Cochrane and Oliver reviews were supplemented with an update search (described
below) and an additional search by Hempel and colleagues (discussed in more detail later),
which addressed the prevention of inpatient falls. After using 15 existing reviews and reports to
identify pertinent sources, which included the two reviews in this chapter, Hempel then searched
multiple databases for relevant literature. The search covered January 2005 to August 2011 and
included randomized controlled trials, non-randomized trials, and before-after studies in Englishlanguage publications that addressed falls in the hospital setting. Details of the search strategy
are in Appendix C.
In the update search, we focused on studies with large sample sizes (at least N=1,000), that
assessed multi-component interventions in acute-care hospitals, in the general population or
older adult population. We were looking for pivotal studies, as defined by Shojania and
colleagues (see Methods, Chapter 2 p.ES-4) that could provide a signal when an existing
systematic review is out of date.32 We identified two new relevant studies, both of which showed
statistically significant improvements in intervention groups when compared with controls, and
which we discuss briefly here. A third study is reviewed because of its unique design. Data for
all studies included in the Oliver review, the Cochrane review, and our update search are in an
evidence table in Appendix D. Table 3 provides an abbreviated description of each study.
185
Table 3, Chapter 19. Abridged evidence tables, adapted from Oliver and colleagues
Author, year
Ang et al, 2011
9
*
Barker et al,
200910
Barry et al,
200111
Brandis, 199912
Study
design
RCT
Before/After
Before/After
Before/After
Cumming et al,
200813
Dykes et al,
2010 14*
Cluster RCT
Fonda et al,
200615
GrenierSennelier et al,
200216
Haines et al,
200417
Before/After
Cluster RCT
Before/After
RCT
Healey et al,
200418
Koh et al,
200919
Krauss et al,
200820
Oliver et al,
200221
Schwendimann
et al, 200622
Cluster RCT
Stenvall et al,
200723
RCT
Uden et al,
199924
Van der Helm
et al, 200625
Before/After
Cluster RCT
Before/After
Before/After
Before/After
Before/After
Outcomes+
1822 patients.
Quality
Score**
25
271,095 patients
16
SFI
15
SFI
11
NFF
27
NFF
27
SFF
20
SFF
11
SFF
626 patients
26
SFF
3386 patients
26
NFF
14
NFF
18
NFF
NGF
15
NFF
199
25
SFF
379 patients
12
NGF
2670 patients
11
NGF
Setting
Participants
3 wards; subacute
rehab and elderly;
Australia
8 wards; acute and
rehab; 3 hospitals; UK
2 hospitals; acute;
Singapore
General medicine;
acute academic hospital
Elderly medical unit;
acute hospital; UK
300 bed; internal
medicine, geriatric and
surgical; Switzerland
3 wards; orthogeriatric,
geriatric, orthopedic;
Sweden
Geriatric dept; acute
hospital; Sweden
Internal med ward and
neurology ward; acute
hospital; Netherlands
3 wards; rehab; UK
SFF
Vassallo et al,
Cohort Study
825 patients
25
NFF
200426
Von RentelnBefore/After
Elderly acute and rehab 7254 patients
17
SFF
Kruse et al,
wards; Germany
200727
*New studies added from update search
** Downs and Black Quality Score,33 evaluated by the authors
+
SFF= significantly fewer falls; SFI=significantly fewer injuries; NFF= nonsignificantly fewer falls; NGF= nonsignificantly
greater falls
Reprinted from Clin Geriatr Med. 26(4), Oliver D, Healey F, Haines TP., Preventing falls and fall-related injuries in hospitals,
645-92, 2009 with permission from Elsevier.
186
Dykes and colleagues compared the fall rates of four intervention units to matched control
units in four urban United States hospitals over a 6-month period.14 Control units received usual
care, which included fall risk assessments, signage for high-risk patients, patient education as
needed, and manual documentation in patient records. The intervention group tested the Fall
Prevention Tool Kit (FPTK), which was developed by the study team. The FPTK is a health
information technology application that includes a risk assessment and tailored signage, patient
education, and plan of care components. The FPTK is integrated with, and seeks to enhance,
existing workflow and communication patterns. Adjusted fall rates in the intervention units (3.15
per 1,000 patient days [95% CI, 2.54 to 3.90]) were significantly lower than in control units
(4.18 per 1,000 patient days [95% CI, 3.45 to 5.06]), with a particularly strong impact among
patients aged 65 or older (rate difference of 2.08 per 1,000 patient days [95% CI: 0.61 to 3.56]).
This study was judged to have a low risk of bias using the criteria of the Effective Practice and
Organizational Organisation of Care (EPOC) Cochrane Group (score of 8 of 9 components). 34
In the second study, Ang and colleagues 9 randomized patients in eight medical wards of an
acute-care hospital in Singapore over a 9-month interval. They used an assessment tool to match
high-risk patients with appropriate interventions, in addition to a tailored educational session, in
the intervention group. Both the intervention and control groups in this study received usual care,
which included environmental modifications, review of medications and fall history, and
educational sessions. The proportion of patients with at least one fall in the intervention group
was 0.4 percent (95% CI, 0.2 to 1.1) while in the control group this was 1.5 percent (95% CI, 0.9
to 2.6) for a relative risk reduction of 0.29 (95% CI, 0.1 to 0.87). Using the EPOC criteria, this
study was judged to be at low risk of bias (score of 8 of 9 components).34
One additional study was identified and is noted here because of its unique design. The study
by van Gaal and colleagues evaluated a program that targeted three patient safety practices
(pressure ulcers, urinary tract infections, and falls prevention) simultaneously and found an
overall positive effect on the development of any adverse event, a composite measure of pressure
ulcers, urinary tract infections, and falls.35,36 The study was not powered to assess falls
separately, yet it is worth noting that the point estimate for the relative risk reduction in falls was
0.69, which is within the range of results reported in other studies and meta-analyses. The value
of this study is the demonstration of simultaneous improvements in several intervention targets.
Thus, new large controlled trials continue to support the conclusion of existing meta-analyses
that multifactorial falls prevention programs are effective in reducing inpatient fall rates.
187
Existing Infrastructure
Five studies reported on the existing quality and safety infrastructure. Here we describe this
infrastructure in terms of factors that may affect implementation of a patient safety practice,
which could include presence of electronic health records or prior experience with quality
improvement or patient safety practices. The five studies included text that captured this concept;
of these, four described their usual fall prevention care. The fifth study provided a more explicit
statement, namely, prior to this study none of the wards carried out specific fall assessments or
interventions, and investigations such as lying and standing blood pressure or ophthalmology
referral occurred on an ad hoc basis. There was no specialist falls clinic or other falls service
available at this hospital.18 Another explanation was less explicit, and was embedded in the
authors explanation of the intervention, which noted that the two control wards continued with
the regular fall prevention policy used at the hospital (i.e., daily assessment of fall risk, review of
fall prevention with the patient and/or their family, use of fall prevention signage, and
implementation of other prevention strategies as needed).20 Two other reports of randomized
controlled trials discussed usual care in a similar fashion when contrasting it with the
intervention.9,14 These descriptions illustrate the potential diversity that may exist in the
control sites in terms of usual care.
In addition to a description of the current fall prevention care, a second type of infrastructure
description addressed an inadequate information system, reporting that the existing information
system was not useful for producing data that we could use to analyze the causes of falls.16 A
further example of this type of explanation is presented by Dykes and colleagues, who suggest
that including hospitals with diverse clinical information and documentation systems enhanced
the [intervention] generalizability.14 The remaining studies do not mention existing quality and
safety infrastructure.
188
Consequently, a dearth of data exists regarding the infrastructure needed to support fall
prevention programs or how the effectiveness of implementation may vary as a result of
infrastructure differences.
External Factors
Although a few studies briefly mentioned patient safety culture, teamwork, or leadership,
only four studies presented expanded explanations that merited mention. Grenier-Sennelier16 use
a framework from Shortell and colleagues37,38 to analyze safety on the unit level, teamwork at
both the organizational and unit level, and leadership on the organizational and unit level.
Stenvall discusses teamwork at the unit level in Table 2 of their article (See Appendix D).23 Koh
discusses leadership on the organizational and unit level: Successful implementation is
mediated by strong leadership and environmental support, which are integral to building positive
attitudes among nurses, ensuring that the sociocultural environment is conducive to the process
of change. In our study, the multifaceted strategy targeting barriers to change exemplified the
commitment of the leadership and environmental support.19 (p. 429) Van der Helm made
multiple observations addressing leadership on both the organizational and unit level:
Although the clinical ward management underlined the importance of implementing the
guideline at the outset of the project, the actual support given was too weak to be
effective. Some managers expressed doubt about the projects chances for success to the
project leader, stating that implementation had already failed before. Ward staff often
regarded improvement activities as unwanted additional work that hindered daily
operations. The two senior nurses often displayed a delegating rather than a directive
management style, for example, in terms of ensuring that the risk assessment tool was
completed or all incidents reported. (p.157)
nurses told us that the medical center did not take the falls problem seriously, which
therefore undermined their own motivation to contribute to the projects success. (p.158)
A measure in the Questionnaire Regarding Knowledge of the Guideline and Attitude
Toward Implementation, There is enough support from the management for guideline
implementation scored 44% to 53%.25
Implementation
The most commonly reported implementation details were patient characteristics (17 studies)
and an initial plan, or what was going to be done in the intervention (17 studies). Slightly less
often (14 studies), studies reported the intended roles of project staff, or by whom the intended
plan components were to be completed. The majority of studies reported the recipients of any
training component (15 studies), with slightly fewer reporting the type of training or giving a
description of the training (12 studies), and even fewer studies reporting the length of training (5
studies).
Another characteristic that distinguished studies was who conducted the risk assessments and
performed the interventions. In the reviews by Oliver and colleagues and the Cochrane group,
among the 17 studies of inpatient fall prevention programs, the risk assessments were performed
by the existing ward staff in 15 and by research staff in two. In 15 studies, the intervention was
performed by the ward staff: seven involved the nursing staff only, seven were multiprofessional,
and two involved physical therapy. In both of the new studies, clinicians or nurses from the
wards performed the risk assessments. The study with nurse risk assessments had research team
189
nurses provide the intervention, whereas the other study relied mainly on ward nurses, although
reference was made to clinicians more generally.
Thirteen studies provided the tools or materials used in the program implementation.
Whereas eight reported on adherence or fidelity to the designed initiative, only five described
how and why the plan evolved. Adherence or fidelity was most often characterized in a
qualitative statement, as with Brandis: The strategies implemented had high acceptance by
staff it is suggested that the higher reductions occurred in areas where the multidisciplinary
team enthusiastically embraced the project.12 An example from a less positive characterization
comes from Cumming: The lack of effect was evident in both wards and occurred despite the
planned nursing and physiotherapy interventions being successfully implemented.13 Dykes and
colleagues provided a strong example of adherence reporting, where protocol adherence was
measured by the completion of components in both control (81%) and intervention wards (94%).
Measures of adoption and reach were usually provided in the form of a flow chart: Six studies
presented these data for providers, and eight presented the data for patients.
For additional information on implementation, we used our update search and sought
suggestions of additional studies from experts. All of these studies had pre-post designs or were a
time series. Six were post-study evaluations of of falls implementations that reported a great deal
of detail about the potential reasons for effectiveness or lack thereof. Nine of the eleven studies
assessed implementation at only one or two facilities. Four of the studies did not report beneficial
effects of the fall prevention program and the article highlighted potential implementation factors
that might account for the lack of success. One study explicitly assessed the effect of some
contextual factors on intervention success across 34 facilities.39 One study explicitly assessed
sustainability. Details of these studies are presented in Appendix D.
We used five of the implementation articles to develop themes regarding effective
implementation and then reviewed all articles for these themes. The following are the most
consistently supported themes:
Leadership support is critical, both at the facility level and at the unit level (e.g.
clinical champions).
Engagement of front line clinical staff in the design of the intervention helps ensure
that it will mesh with existing clinical procedures.
Multidisciplinary committees guided or oversaw most interventions
developed/guided/overseen by
Pilot testing the intervention helps identify potential problems with implementation
Informational technology systems capable of providing data about falls can facilitate
evaluations of the causes, compliance with the intervention components, and (in one
case) be a crucial facilitator of the intervention.
Changing the prevailing attitude that falls are inevitable and nothing can be done
about them is required to get buy-in to the goals of the intervention
Education and training of clinical staff is necessary to help ensure compliance does not
diminish.
Table 4, below, presents textual support from the implementation articles for five of the
seven themes (pilot testing and information technology systems are not presented due to space
limitations).
190
Leadership
Support
Frontline
Engagement
Multidisciplinary
Committees
Pilot Testing
Browne et al.,
40
2004
--
--
Falls Committee;
quarterly meetings
191
Information
Technology
Systems
the redesign of
an adult
inpatient falls
program using a
computerized
information
systemthe
tool provides an
accurate
assessment of
the fall risk of
each patient.
Indicators are
embedded into
routine
assessment
documentation,
eliminating
added chargting
time. The
program allows
tailored
interventions for
specific patient
risks.
Attitude Change
Education and
Training
--
Nurses were
taught about the
redesigned falls
program by fall
and restraint
fairs that
coincided with its
implementation.
Results of
Intervention and
Implementation
Successful
Leadership
Support
Frontline
Engagement
Multidisciplinary
Committees
Pilot Testing
Capan et al.,
41
2007
A unit
champion was
selected to act
as a staff
resource
who was
respected as a
mentor and
passionate
about patient
safety
Staff involved
in choosing
equipment
Dempsey,
42
2004
--
Raised
concern over
nurses power
to induce
change
the hospital
quality council
chartered a
multidisciplinary
falls prevention
task force. The
team included
nurses, nursing
management, a
physician/geriatrici
an, nursing
educators, a
psychiatric clinical
specialist , risk
management staff,
performance
improvement/mea
surement staff,
and
representatives
from physical
therapy and
pharmacy.
--
A tool was
developed and
tested for interrater reliability in
a pilot study
when five nurses
of different
experience levels
assessed the
same patient.
On the basis of
the results of the
research project,
the Falls
Prevention
Programme
became standard
practice for
medical
patients
192
Information
Technology
Systems
--
Attitude Change
Education and
Training
The research
team educated
the staff about
falls and the
importance of
fall prevention,
including
background
information on
falls and how the
new tool was to
be used. 95% of
staff completed
the education
prior to the
implementation
of the tool.
--
The Falls
Prevention
programme
consisted of an
assessment tool,
an alert graphic,
and education
(patient and
staff) Staff
education
commenced at
the introduction
of the study and
continued
intermittently
though formal
and informal
means.
Results of
Intervention and
Implementation
Successful
Mixed results,
initial success
followed by
deterioration over
five years.
Leadership
Support
Frontline
Engagement
Multidisciplinary
Committees
Pilot Testing
Gutierrez,
43
2008
Identify clinical
champions;
leadership on
unit agreed to
send a nurse to
the EvidenceBased Practice
Institute
Leadership
formed a team
to address falls
issue, team
was led by a
senior vice
president,
information
was presented
to leadership
throughout
project
project
design
included
soliciting staff
and physician
feedback
--
--
--
Multiple tools
were tested
before the
redesign team
developed their
own, which was
also tested.
Kolin et al.,
44
2010
193
Information
Technology
Systems
--
Attitude Change
Education and
Training
--
Currently, the
team is are
working on an
interface to
connect the
system
electronic
medical record
with the event
reporting
system. The
system had a
combination of
paper
documentation
and electronic
record sites,
which had
separate
program roll out.
Implementation
means changing the
way nurses think
about falls accepting
that all patients are at
risk.
Comprehensive
nursing
education was
conducted
Results of
Intervention and
Implementation
Successful
Successful
Leadership
Support
Frontline
Engagement
Multidisciplinary
Committees
Pilot Testing
McCollam,
45
1995
Nursing
Administration
involved in full
implementation
--
Research in
Practice
Committee
oversaw the
project
Senior
leadership
support helps
remove
organizational
barriers to
change and
provides
resources
needed to
implement
change The
four sites that
reported
spreading
changes to
other facilities
also indicated
that leadership
was a major
success
factor.
--
teamwork skills
are an important
component of
sustained
success
Interdisciplinary or
multidisciplinary
falls team was a
core component of
all four high
performing sites.
Problems
identified during
the pilot included
inconsistent and
incomplete
reassessment,
identification of
secondary
diagnoses, and
score
consistencies
between shifts.
Adjustments
were made for
full
implementation.
--
Neily, 2005
39
194
Information
Technology
Systems
--
Attitude Change
Education and
Training
Training
sessions were
conducted for
nursing; video
tape was shown
about tool;
understanding
checked using
evaluation
--
--
--
Results of
Intervention and
Implementation
Successful
Successful
Leadership
Support
Frontline
Engagement
Multidisciplinary
Committees
Pilot Testing
OConnell,
46
2001
--
--
Team of
researchers and
clinicians
Rauch et al.,
47
2009
Leadership
hired a
consulting
team. All levels
of leadership
were engaged
and accepted
ownership of
the project. A
champion was
identified in
each unit.
It is
imperative to
obtain
frontline staff
input and
feedback to
ensure that
successful
change
management
occurs in the
clinical arena
If there are
any words of
advice here,
they would
be: never
change a
program
without
directly
involving and
getting buy-in
from those it
immediately
affects.
195
Information
Technology
Systems
--
--
Attitude Change
Education and
Training
--
educational
needs were
identified and
sessions were
scheduled
[including] an
introduction of
the assessment
tool and proper
utilization
Results of
Intervention and
Implementation
Unsuccessful
Successful
Leadership
Support
Frontline
Engagement
Multidisciplinary
Committees
Pilot Testing
SeminGoossens,
48
2003
Attempt to
involve medical
chiefs and
nurse
managers
could have
promoted
implementation
In our case,
efforts to reach
and involve the
people higher
in the hierarchy
such as the
Medical Chiefs
and nursing
managers were
not successful.
A project team
was formed
consisting of 9
nurses in various
positions, a clinical
epidemiologist,
and a consultant
for quality
improvement
projects.
After a 3 month
pilot, the
guidelines were
finalized.
Weinberg et
49
al., 2011
Hospital
leadership
initiated effort
and prioritized
fall prevention
We did not
believe in a
top-down
strategy and
so we
involved the
nurses in
rewriting and
implementing
the guideline.
Authors would
have tried to
get more buyin from floor
nurses if
given another
try, but they
did receive
feedback and
modify the
intervention
accordingly.
--
Committee was
formed by
leadership and
attendance was
mandated;
monthly fall
reviews were
attended by unit
managers, staff
involved in patient
care, and fall
prevention
initiative co-chairs
The Fall
Prevention
Initiative was
rolled out
incrementally,
using continuous
quality
improvement
methods
196
Information
Technology
Systems
--
Attitude Change
Education and
Training
Nursesfrequently
stated that it was
simply impossible to
prevent patients from
falling. Falling was
recurrently considered
to be an inevitable part
of aging,
hospitalization, and
illness, and therefore
seen as an
unavoidable accident,
rather than something
predictable and often
preventable.
Dissemination of
the guideline,
including large
posters.
--
Transforming the
culture was integral to
implementation;
emphasis placed on
building a just culture
and having a
constructive,
nonpunitive forum for
discussion
Yes
Results of
Intervention and
Implementation
Unsuccessful
Successful
197
High
Evidence or
Potential for
Harmful
Unintended
Consequences
Moderate
(increased use of
restraints and/or
sedation)
Estimate of
Cost
Moderate
Implementation Issues:
How Much do We
Know?/How Hard Is it?
Moderate/Moderate
References
1.
2.
3.
4.
5.
6.
7.
198
8.
9.
10.
11.
12.
13.
14.
15.
16.
26.
17.
27.
18.
28.
29.
30.
31.
32.
33.
34.
19.
20.
21.
22.
23.
24.
25.
199
35.
36.
37.
38.
39.
40.
200
41.
42.
43.
44.
45.
46.
47.
48.
49.
201
202
and incidence of delirium. Depression was evaluated in 10 studies, but only four (40%) found a
significant association with delirium occurrence.
Other patient-specific risk factors that showed a significant association with delirium in more
than one study include male gender, multiple medications, comorbidities (e.g., diabetes),
pneumonia, various anesthetics, neuropsychiatric drugs (e.g., benzodiazepines), anticholinergics,
blood transfusions, abnormal serum chemistry (e.g., urea levels, creatinine levels),
apolipoprotein E4 (APOE4), atrial fibrillation, heavy alcohol intake, volume depletion
(dehydration), oxygen levels, complications, restraints (rendering patients immobile) and visual
impairment. Several studies evaluated patients undergoing specific surgical procedures (e.g., hip
repair or replacement, cardiac surgery); some of these studies focused on surgery-specific risk
factors (e.g., blood transfusions, intraoperative anesthesia) and evaluated few non-surgical
factors.
Given the multifactorial nature of delirium, a patient safety practice designed to assess and
remediate multiple factors is believed to be more likely to be effective. Indeed, the list of
components in successful delirium prevention bundles targets several factors identified in this list
of patient and environmental contributors to delirium. For example, the Hospital Elder Life
Program (HELP) specifically targets six risk factors for delirium: cognitive impairment, visual
impairment, hearing impairment, sleep deprivation, immobility, and dehydration.7 Of this list,
only hearing impairment was not identified as a risk factor by the studies in our evidence base,
but this may be because only one of those studies even evaluated it as a possible risk factor.
Multicomponent Interventions
Hospital Inpatient Care
Of the multicomponent intervention studies, two used HELP and a third used a modification
of HELP. One was a controlled before-and-after study with a concurrent control group consisting
of patients from usual care units7,10; this study had a moderate risk of bias. The remaining two
studies were before-and-after studies where the usual care group consisted of patients treated
prior to implementation of HELP (historical control)5,6; these studies had a high risk of bias. All
three of these studies found a significant reduction in incident delirium after implementation of
HELP compared with usual care. Although the findings of the studies were consistent, the
average risk of bias was high mainly due to lack of randomization and blinding.
203
Two studies used proactive geriatric consultation with targeted recommendations based on a
structured protocol for patients with hip fracture. One was a single-blind RCT with usual care
control,8 while the other was a before-after study with a historical usual care control.9 Both
studies reported a significant reduction of incident delirium for the geriatric consult group
compared with the usual care group; however, the RCT findings were no longer statistically
significant after adjustment for baseline imbalances. The risk of bias was high and moderate for
the respective studies.
Of the remaining multicomponent studies, all but one reported a significant reduction in
delirium by at least one measure in the intervention group versus the control group. The
exception was a study of a system-wide quality improvement project.11 A study of nursefacilitated family participation reported significantly fewer patients with a diagnosis of delirium
(defined by a score 4 on the Intensive Care Delirium Screening Checklist [ICDSC]) in the
intervention group, but also reported no significant between-group difference in mean scores;
this study placed more emphasis on the latter measure.12 Overall, the findings are consistent with
the findings from studies of the HELP intervention, although the risk of bias was high again due
to lack of randomization and blinding.
Long-Term Care
The single study set in a nursing home setting reported that homes randomized to use
pharmacist-led geriatric risk assessment medguide (GRAM) reports and automated medication
monitoring plans had a significant reduction in potential delirium onset among newly-admitted
residents compared with homes randomized to usual care.13 However, it is unclear how much of
this is due to delirium prevention or resolution of new-onset delirium.
The majority of the evidence suggests that multicomponent interventions are effective in
preventing onset of delirium in at-risk patients. However, these studies do not address the
question of which particular components within a program provide the most benefit.
Single Interventions
The majority of the single-intervention studies also found a significant reduction in delirium
incidence for the study interventions, but roughly one-third (five studies) did not find a
significant reduction. Unlike the multicomponent evidence base, almost all of the singleintervention studies were RCTs. However, few studies used the same medication or comparison
treatment in the same patient population, making it difficult to determine consistency of findings
for most of these interventions.
204
A study of light versus deep Propofol sedation during spinal anesthesia for hip repair found
that patients receiving light Propofol sedation (measured by the bispectral index [BIS]) had a
significantly lower rate of postoperative delirium.16 The risk of bias was moderate. It is unclear
how this compares to the amount of Propofol used in the study comparing Dexmedetomidine and
Propofol, which reported the amount in g/kg/minute.14
Patients undergoing cardiac surgery had significantly lower rates of postoperative delirium
after receiving Ketamine (an NMDA receptor antagonist) during anesthetic induction compared
with placebo.17 The risk of bias was moderate, and the findings should be confirmed by other
studies.
Patients undergoing joint surgery had significantly lower incidence of delirium after
receiving fascia iliac block prophylaxis via Bupivicaine (a local anesthetic) compared with
placebo.18 The risk of bias was moderate, and the findings should be confirmed by other studies.
Acetylcholinesterase inhibitors. One study of patients undergoing cardiac surgery did not find a
significant between-group difference for those receiving Rivastigmine compared with those
receiving placebo.19 This study was judged to have a low risk of bias. Patients undergoing joint
surgery who received a different acetylcholinesterase inhibitor (Donepezil) did not show a
significant reduction in postoperative delirium compared with those receiving placebo.20 This
study had a moderate risk of bias. Pooling of these two studies findings resulted in a relative risk
of 1.11 (95% CI 0.69 to 1.79); the confidence interval was too imprecise to rule out the
competing possibilities that acetylcholinesterase inhibitors are ineffective or might confer a
benefit.
Atypical antipsychotics. Patients undergoing cardiac surgery had significantly lower rates of
postoperative delirium after receiving Risperidone compared with placebo in one RCT with a
moderate risk of bias.21 Patients undergoing joint surgery who received Olanzapine had a
significant reduction in postoperative delirium compared with placebo in another RCT with low
risk of bias.22 Both studies showed a substantial reduction with almost identical risk ratios; the
combined summary relative risk is 0.35 (95% CI 0.25 to 0.50, P<0.0001). The findings are
therefore consistent and precise for this drug class.
Typical antipsychotics. Patients undergoing joint surgery who received Haloperidol did not
show a significant reduction in postoperative delirium compared with those receiving placebo.23
The risk of bias was moderate and the findings were imprecise, therefore requiring confirmation
from additional studies.
Melatonin. Patients undergoing joint surgery had significantly lower incidence of delirium after
receiving Melatonin compared with placebo.18,22,24 The risk of bias was moderate, and the
findings should be confirmed by other studies.
Benzodiazepines. One RCT found that postoperative delirium was significantly lower in
gastrointestinal surgery patients who received the benzodiazepines Diazepam + flunitrazepam as
a drip infusion in addition to a pethidine drip infusion for the first 3 days compared with those
who did not receive these infusions.25 The risk of bias was high.
205
Music therapy. In two RCTs conducted by the same authors at the same hospital, patients
undergoing hip or knee surgery had significantly lower rates of acute confusion after receiving
music therapy compared with those receiving usual care.26,27 Both studies had a high risk of bias,
in part because they employed an unvalidated delirium assessment method, and should be
repeated at other hospitals for confirmation of the results.
Long-Term Care
Hydration therapy. A quasi-randomized study comparing 8 weeks of hydration therapy to usual
care for delirium prevention among residents of four nursing homes (hydration or control was
randomized by nursing home) did not find a significant difference between intervention or
control homes in episodes of acute confusion.28 The risk of bias was high.
Acetylcholinesterase inhibitors. A study using Rivastigmine daily for two years in patients with
vascular dementia found that the Rivastigmine group had significantly fewer episodes of
delirium than those taking cardioaspirin (the control group).29 This was the only study in the
entire evidence base that exclusively enrolled patients with dementia, and it was judged to have a
high risk of bias. Although these were ambulatory outpatients, they were judged to be similar to
patient populations in long-term care settings.
206
Fourteen studies outlined the intended intervention and the general sequence in which the
components were implemented; only 11 studies included enough detail to determine the roles of
the various team members. However, this was generally a description of how the intervention
was supposed to be implemented; most studies did not describe any modifications or failures of
adherence that might have occurred during the actual implementation. Only one study actually
measured adherence to targeted recommendations, reporting an adherence rate of 77% regarding
implementation of a geriatric consultants recommendations for patients after hip fracture repair.8
Twelve studies reported patient characteristics.
Although implementation of multicomponent delirium prevention programs has not been
well-described in most studies, a few themes seem sufficiently constant to report here:
Engagement of front line clinical staff in the design of the intervention helps ensure that
it will mesh with existing clinical procedures.
A multidisciplinary team comprising clinical experts, nurses and additional staff is
helpful for implementation of a complex intervention
Education and training of clinical staff is necessary to help ensure compliance does not
fall.
208
factor in these studies. All but two studies were judged to have high risk of bias, and these
exceptions were judged to have a moderate risk of bias. The majority of the evidence suggests
that most multicomponent interventions are effective in preventing onset of delirium in at-risk
patients in a hospital setting (Strength of Evidence: Moderate). In general, successful delirium
prevention programs involved a multidisciplinary team of clinical experts, nurses, and other staff
(e.g., physical therapists, volunteers) and included protocols for early mobilization of patients,
volume repletion (for hydration and electrolyte balance), and addressing visual or hearing
deficits; a few programs included elimination of unnecessary medications.
Other components reported in more than one study included staff education, geriatric
consultation, therapeutic cognitive activities/orientation, extra nutrition, sleep enhancement, pain
management, anesthetic protocols, supplemental oxygen, assessment of bowel/bladder functions,
and prevention and treatment of medical complications. However, these studies do not address
the question of which particular components within a program provide the most benefit.
There was not enough evidence to evaluate the benefit of delirium prevention programs in
long-term care settings.
Although implementation of multicomponent delirium prevention programs has not been
well-described in most studies, a few themes seem sufficiently constant to report here:
Engagement of front line clinical staff in the design of the intervention helps ensure that
it will mesh with existing clinical procedures.
A multidisciplinary team comprising clinical experts, nurses and additional staff is
helpful for implementation of a complex intervention
Education and training of clinical staff is necessary to help ensure compliance does not
fall.
Although several of the single-intervention studies also found a significant reduction in
delirium incidence for the study interventions, few studies used the same medication or
comparison treatment in the same patient population; this makes it difficult to determine
consistency of findings for most of these interventions. For atypical antipsychotics, two RCTs
with a low to moderate risk of bias evaluating different drugs within this class showed consistent
and precise findings of reduction in postoperative delirium among surgical patients (Strength of
Evidence: Moderate). Although two RCTs reported a significant reduction in acute confusion for
patients receiving music therapy, these studies were conducted at the same institution by the
same authors and used an unvalidated delirium assessment method. Therefore, the evidence is
insufficient for a conclusion regarding music therapy. Two RCTs had inconclusive findings
(even when pooled) regarding the efficacy of acetylcholinesterase inhibitors, rendering the
strength of evidence insufficient.
Most of the remaining treatments (or treatment comparisons) were represented by only one
study with a moderate or high risk of bias; we judged the evidence about these treatments to be
insufficient.
Future comparative effectiveness studies with standardized protocols are needed, particularly
to identify which components in multicomponent interventions are most effective for delirium
prevention. Identification of the most effective bundle of components might encourage hospitals
to adopt a more standardized approach to delirium prevention. Additional RCTs are also needed
to determine which single-component medical therapies or drug classes are truly beneficial for
patients at risk of delirium. A summary table is located below (Table 1).
209
Moderate
Evidence or
Potential for
Harmful
Unintended
Consequences
Low
Estimate of
Cost
Moderate
Implementation Issues:
How Much do We
Know?/How Hard Is it?
Moderate/Moderate
References
1.
2.
3.
4.
9.
10.
11.
12.
13.
14.
15.
5.
6.
7.
8.
210
16.
17.
18.
19.
20.
21.
22.
23.
24.
211
25.
26.
27.
28.
29.
30.
31.
32.
33.
212
include scheduling regular repositioning and turning for bed- and chair-bound individuals and
using pressure redistribution surfaces in the operating room for high-risk individuals.
We identified two frameworks for Patient Safety Practices that appear to thoroughly embody
components used in PU prevention initiatives today (see Table 1 and Table 2 following). A
recent systematic review9 describes the Indiana State Department of Healths classification of PU
initiative components, as follows:
Organization components include team makeup, policies and procedures, ongoing quality
evaluation processes, educating staff, utilizing skin champions, and the development
and system-wide communication of the written care plan
PU prevention components include risk and skin assessment, moisture management,
nutrition and hydration optimization and pressure management
Care coordination components include the establishment of regular meetings to facilitate
communication, collegiality, and learning
The ABCDE of Pressure Ulcer Incidence Reduction Initiatives was outlined at the 12th NPUAP
Biennial Conference held on February 2011. The initiatives were described as: administrative
support backed by support at the patient care level; bundling care practices and having an
identifiable theme; creating a culture of change, commitment, and communication;
documentation of PU prevention practices must be visible; and education is essential.10
213
Table 1, Chapter 21. Components of pressure ulcer prevention trials in U.S. hospitals, 2000 to 2011
Study
Implement Review
Protocol
Wound
Care
Products
Upgrade
Automated
Systems
Integrate
New
Reporting
Education/ Risk
Skin
MultiAudit and
Training
Assessment Champion disciplinary Feedback
Tool
Team
Lynch and
Vickery
11
2010
Young et al.
12
2010
Bales et al.
b,d13
2009
Chicano and
Droishagen
b14
2009
Walsh et al.
15
2009
Dibsie L.
16
2008
McInerney J.
17
2008
Ballard et al.
18
2007
Catania et al.
19
2007
LeMaster K.
b,c20
2007
Courtney et
21
al. 2006
Gibbons et al.
d22
2006
Hiser et al.
23
2006
X
X
X
X
X
X
X
X
X
X
214
X
X
X
X
Lyder et al.
24
2004
Stier et al.
25
2004
X
X
a
Table 2, Chapter 21. Components of pressure ulcer prevention trials in long term care, 2000 to 2011
Study
Implement Upgrade
Integrate Education/ New
Use of
Skin
MultiProtocol
Training
Champion disciplinary
Automated New
Assessment Outside
Systems
Reporting
Tool
Team
Consultants
*
Horn et al.
26
2010
Rantz et al.
b27
2010
Milne et al.
d28
2009
Tippet A.
c,d29
2009
Rosen et
c30
al. 2006
Abel et al.
31
2005
Ryden et
32
al. 2000
Rantz et al.
a33
2001
Audit and
Feedback
X
X
X
X
X
X
215
216
appendix). Study designs were mostly time series assessments of changes before, during, and
after implementation of the intervention. Other study designs included randomized controlled
(k=2)27,33 and controlled before-and-after (k=1).32 A majority of the studies focused on universal
prevention (all risk levels); one focused on high-risk patients.26 Pressure ulcers were the primary
focus of 20 studies and part of a comprehensive approach in three.27,32,33
The review group agreed not to assess risk of bias or rate the strength of evidence for those
reviews primarily focused on implementation. Therefore, in this section, we briefly summarize
the primary results; subsequently, we provide detailed assessments of the implementation efforts.
Acute Care
Evidence presented below on PU prevention programs implemented in the acute care setting
is based primarily on one systematic review (Soban 2011) of nurse-focused quality improvement
(QI) initiatives.41 Of the 39 studies included in the review, 12 met our inclusion criteria.
Additionally, we discuss three other studies that were published since the Soban review was
completed.
The Soban review41 had three objectives: describe the intervention strategies used, describe
the process and outcome measures reported, and examine the interventions effects on outcomes.
Study findings were categorized as processes of care (e.g., staging of acquired stage 2 PUs)
and/or patient outcomes (e.g., PU incidence). Eleven studies reported patient outcomes; only one
study24 reported both. Because the review included limited data, we accessed information
directly from the studies.
First, we examine intervention effects on several processes of care reported in one
multihospital QI collaboration overseen by the Connecticut Quality Improvement Organization,
Qualidigm. In 2004, Lyder et al.24 reported a 2-year follow-up on 14 measures, seven of which
were process of care measures. Four plan-do-study-act (PDSA) improvement cycles
implemented at 17 hospitals resulted in significant increases in identifying high-risk patients
(20.3% vs. 35.3%, p<0.001); repositioning of bed-bound patients every 2 hours or every hour in
chair-bound patients (50.9 vs. 56.9, p = 0.01); use of nutritional consults in malnourished
patients (34.3% vs. 48.6%, p<0.001); and staging of acquired stage 2 or greater PUs (22.4% vs.
44.2%, p<0.01). No statistically significant findings were reported for the remaining processes of
care (including staging of acquired stage 1 PUs) or hospital-acquired incidence rate (baseline vs.
follow-up; 20.6 vs. 20.8, p=0.90).
Five of the 11 remaining studies reported in the Soban review conducted nurse-focused
initiatives facility- or system-wide. A majority of studies reported on prevalence or incidence
measures; both types of measures have been described as useful in assessing and improving
patient care Catania indicates that a declining incidence of PUs would indicate that a prevention
program is working to decrease new PU cases, while a declining prevalence indicates that the
treatment strategy was also having an impact on the duration of PUs.19 Several studies reported
that initiatives reduced prevalence or incidence to zero.
After a 10-month implementation of the SKIN (Surfaces, Keep the Patient Turning,
Incontinence Management, Nutrition) bundle, Gibbons et al.22 reported a 90% reduction (5.7% to
0.448%) in prevalence at the Nations largest Catholic and nonprofit health system.22 Similar
reductions were reported at a 548-bed, two-hospital system in Southwest Florida:17 a 5-year trend
analysis indicated a significant reduction in PU prevalence (overall [-81%] and ulcers located on
the heel [-90%]) after proactive assessment and management of at-risk patients. One study
reported zero PU prevalence and incidence after 1 year of a nurse-focused initiative at a 300-bed
217
community hospital.13 Other benefits listed were optimal patient care and avoiding the cost of
treating stage 3 or 4 ulcers.
In 2006, Courtney (SOS program/Six Sigma methodology) reported that one 710-licensed
bed, multisite, not-for-profit hospital reduced the PU incident rate from 9.4% to 1.8% over 3
years. Incidence was reduced by 6.3% after only 1 quarter.21 Two years earlier, Stier et al.25
reported reducing incidence by more than 50% at a 5,600 bed nonprofit health care system.
Significant improvements were also reported from initiatives implemented in patient care
units, with two studies reporting zero prevalence postimplementation. In 2006, a two-unit
intensive care unit (ICU) significantly reduced PU prevalence (34% to 8%), noting that National
Database of Nursing Quality Indicators (NDNQI) benchmark data were instrumental in helping
our unit focus on PU prevention, ultimately leading to improved patient outcomes.18 Catania et
al.19 reported reducing PU prevalence by more than 50% due to implementing the Pressure Ulcer
Prevention Protocol Interventions in five in-patient units at one cancer hospital. From September
2004 (baseline) to June 2006 (postimplementation), the percentage of patients with all types of
PUs and with hospital-acquired pressure ulcers (HAPUs) was reduced to 4% and 2%,
respectively. NDNQI benchmarks at the time were 12.65% for all ulcers and 6.84% for HAPUs.
In a 2007 study, preimplementation PU prevalence rates for a pulmonary and oncology unit
were 9% and 12%, respectively. LeMaster20 reported that nurse-focused QI initiatives reduced
prevalence in these two hospital units to zero. Rates were reduced from 9.2% to 6.6% in five
units in one Florida hospital.23 One medical ICU, which reportedly had the highest HAPU
prevalence (average of 29.6%) among the participating units, reduced prevalence to zero.
Dibsie et al.16 reported a facility-wide reduction in percentage of patients with stage 2 or
greater HAPUs (4.2% vs. 3.2%) in four adult critical care units (54 beds) at two U.S. hospitals.
Rates for the surgical intensive care unit, however, did not improve over time (6.1% pre- and
postimplementation). A 23-month initiative (Chicano 2007) at a 25-bed intermediate care unit
reduced incidence.14
Benefits described in three separate studies (Walsh et al., Young et al., and Lynch and
Vickery) included a reduction in prevalence and incidence and improvements in processes of
care. In 2009, Walsh et al. reported a reduced PU prevalence from 12.8% to 0.6% from an 18month initiative. Walsh also reported increased focused communication among patient
caregivers; once improvements were noticed, clinicians behavior and clinical processes
improved.15 Young et al. reported several successes, including streamlining online policies (from
7 to 1) and reduction in time spent documenting skin care. Young also reported clinically
relevant reductions in development of nosocomial PUs.12
In one year, PU rates were reduced by 82.8% (2.8% to 0.48%) at one rehabilitation hospital.
Lynch and Vickery reported that streamlining documentation increased timely and accurate
completion of documentation from 60% to 90% in 90 days. This facility also increased
patient/family involvement in patients care by providing an educational brochure and reviewing
interventions on admission.11
Long-Term Care
A total of eight studies met our inclusion criteria in their evaluation of multicomponent
programs to prevent PUs in long-term care facilities. Study duration ranged from 6 months to 6
years. Among these eight programs, four reported significant reduction in PU incidence or
prevalence rates.
218
In 2010, Horn et al. reported on three main outcome measures.26 First, they considered
prevalence of PUs using Centers for Medicare & Medicaid Services (CMS) quality measures
(QMs). Based on data from seven facilities, they report that the CMS high-risk PU QM
decreased from 13.0% (baseline) to 8.7% (12 months postimplementation). A second outcome of
interest was the number of in-house acquired PUs. The average number of in-house PUs (all
stages) per facility was reduced by 62% (12.1 [baseline] to 4.6 [postimplementation]). Lastly,
Horn et al. reported a 53.2% reduction in the average number of certified nursing assistant
(CNA) documentation forms and a mid-90% completeness rate of CNA documentation.
Tippet et al. reported that nosocomial PU ulcers were eliminated after 6 months.29 They also
reported an 86% reduction in average incidence: 5.19% (preinitiative) vs. 0.73% (postinitiative);
p<0.0001. By the end of the fourth year, both incidence and prevalence were reduced by 99%.
In 2006, Rosen et al. reported a significant reduction in PU incidence.30 The percentages of
patients identified as high-risk were 22.3% and 28.0% at the baseline and intervention periods,
respectively. Significant reductions in PU incidence were reported for stage 1 and beyond
(P<0.001) and stage 2 and beyond (p<0.05). However, these improvements were not sustained
during the postintervention periods when no weekly reports (indicating completion of training)
were provided; no targets or goals were established; and no financial incentives were offered to
staff.
One RCT was conducted in three privately owned facilities in the midwestern United
States.32 This 6-month study evaluated the effectiveness of advanced practice nurses (APNs) to
successfully implement scientifically based protocols for PUs and other clinical problems. The
APNs delivered treatment in two facilities; in the third facility, patients received usual care. At
6 months, the percent of APN-treated patients with PUs was significantly reduced (19.8% vs.
3.5%; p=0.04). The percentage of patients with PUs for the usual care group was also reduced,
although not significantly.
Four studies conducted in long-term care facilities reported no significant findings from
primary analysis. At 12 months, Rantz et al.27 reported relative improvements in high-risk PU
scores (negative indicating improvement) were -53%, -12%, -5%, and +435% for Group 1, 2, 3,
4, respectively. At 24 months, relative improvement was -3%, -8%, +59%, and +105% for these
same groups, respectively.
Abel et al. describe results from a 2-year study conducted in 20 sites.31 They report a
significant improvement for 8 of 12 PU-related quality indicators; however, only a trend toward
a lower incidence of facility-acquired PUs (x2 MH = 3.66, p = 0.06) was observed. Facilities also
fell short in two other key measurements: proportion of high-risk residents with facility-acquired
PUs whose care plan intervention reflects treatment orders and proportion of skilled nursing
facilities that have a wound protocol.
Milne et al.28 reported several successes. The facility-acquired PU prevalence rate at baseline
was 41%. PUs were reduced to <3% on two units due to increased monitoring of modified nasal
cannula (pulmonary unit) and increased attentiveness to heel offloading, support surfaces, and
proper positioning (SCI/trauma unit). Of the 396 charts reviewed, fewer than 1% had missing
data. A review of 45 patient charts showed that wound teams consistently determined staging
and wound etiology in more than 90% of cases. The facility-acquired PU rate was reduced by
37% within 1 year postimplementation.
In 2001, Rantz et al.33 randomly assigned 87 facilities to receive workshop plus feedback
reports (Group 1); workshop, feedback reports, and clinical consult (Group 2); and control
219
(Group 3). Primary analysis indicated no statistically significant findings for prevalence of stage
1-4 PUs or prevalence of stage 1-4 PUs (low-risk residents).
220
CMS will no longer allow higher DRG (diagnosis-related group) payments for patients with
stage 3 and 4 HAPUs. Catania reported that one dedicated cancer hospital was responding to the
identification of two stage-4 PUs and evidence from the NDNQI survey that the prevalence of
PUs in the hospital exceeded the national benchmark by close to 50%.19 A 25-bed intermediatecare unit indicated that identification of high prevalence rates, nursing peer reviews, chart audits,
and unit observations played an important role in the hospitals response.14 Lynch mentioned that
the facility experienced an upward trend in PUs on two units.11
One 528-bed nonprofit facility, at which prevalence of HAPUs was lower than national
norms, set out to eliminate HAPUs completely.22 In 2006, Courtney et al.21 described the
emergence of new guidelines from the American Nurses Association and the Agency for
Healthcare Research and Quality as showing a revitalized interest in preventing and treating
PUs. Additionally, studies using the Nursing Care Quality Initiative guidelines revealed high
prevalence of PUs (13%) and lack of documentation and management. Two critical incidents
(not specified), concerns within individual units, and inconsistent documentation were listed as
external motivators by Dibsie et al.16 Additionally, the frequency with which concerns and
incidents were discussed, but went unreported within the internal reporting system were of
concern. Young reports stakeholder commitment to improve patient outcomes and a goal to be
recognized as a quality provider of patient services.12
External factors influencing the staff at one 151-bed Midwest skilled facility were a G-level
citation (a deficiency judged to cause actual harm to residents) and State survey deficiencies.29
This facility recorded PU prevalence and average incidence rates as high as 25% and 23.9%,
respectively. One facility reported receiving multiple citations from the Department of Health.30
Abel et al.31 indicated that the 20 participating facilities were identified from 143 Medicarecertified skilled nursing facilities with high rates of PUs despite a high volume of residents
receiving preventive care.
Teamwork/Leadership
Although a majority of studies utilized a multidisciplinary team, skin champions were
described as key team members. Studies set in acute care settings described use of certified
221
wound ostomy continence nurses (CWOCNs),13,18,23 staff registered nurses or patient care
technicians,21 clinical nurse specialists (CNSs),19 and a collaboration between CNSs and wound
ostomy nurses20 in this role. Of eight studies set in long-term care settings, skin champions were
designated in five studies;26,28,29,31,32 in two studies, advanced practice nurses served in this
role.28,32
Three studies included lengthy descriptions of leadership within their facilities. Stier et al.
described leadership support to multidisciplinary teams at a 5,600-bed nonprofit New Yorkbased health care system.25 Teams consisting of clinical experts from 18 facilities convened to
openly discuss the various risk assessment tools and facility protocols in place. Multidisciplinary
teams agreed to develop a uniform policy, skin care formulary, and specialty bed contract.
System leadership (e.g., nurse executives, quality management directors, and senior physicians)
provided support to the team at both the system and facility level vis-a-vis resources, ensured
staff orientation and education, maintained quality control programs, and continually assessed
actions to improve performance through system-wide care committee meetings.
Dibsie16 described the importance of teamwork and leadership at a multisite academic
medical center. Discussions on serious skin-related issues were held with unit nursing
management, immediate senior nursing management, and selected peers. Discussions later
involved a larger group of managers and clinical specialists after it became evident that the
issues crossed many areas and could be better handled by the group together. When necessary,
senior management stepped in to stress the importance of resolving issues related to
preventionthroughout the organization.
Young et al. described a change in leadership at a 540-bed acute care facility in Indiana.
Clinicians were initially wary of managements intent for clinician involvement. Their
hesitation was attributed to past experiences when some clinicians joined the task force to attain
required activities relating to clinical advancement or in response to a managers request. As a
result of mandates that the new skin team be clinician-led, the majority of the original task force
members leftThe few remaining committee members were charged with selecting new task
force members who could serve as unit champions.12
222
knowledge deficit, and communication deficits were identified as root causes of reported high
incidence/prevalence at one teaching hospital. Incomplete initial and ongoing skin assessments,
inconsistent implementation of prevention interventions, and lack of coordination among staff
were cited as examples of preimplementation safety culture.21 Lastly, analysis of survey results
at one 25-bed intermediate care unit revealed that admission documentation did not identify
patients with an increased risk for developing PUs.14
In the long-term care setting, five studies reported minimal information on patient safety
culture at the organizational level.26,28-31 One study included information at the unit level.28
Abel31 reported that facilities were plagued by inadequate assessments and data omissions
associated with risk. Milne indicated that one facility had above-average PU prevalence, used a
faulty EMR that was inconsistently used by clinicians, and had deficient documentation of risk
assessment.28
Implementation Tools
Below, we describe examples of unique tools that were used for audit and feedback,
education and training, monitoring progress, identifying specific groups of patients at risk, and
streamlining products and processes in more than 20 PU prevention initiatives. For a complete
listing of implementation tools, see evidence tables in the appendix.
223
Guest speakers discussed new concepts and educated physicians about the CWOCNs
role, level of clinical expertise, and best-practice wound care interventions.23
During 30-minute mandatory continuing education sessions, participants sit on bedpans
that act as a reminder that PUs can occur within as brief a time as 1 hour.12 Compensation
was provided to staff participating on their own time. Educational content was tailored
for RNs, LPNs, secretaries, and CNAs. A post-test survey measured effectiveness of
presentation.
Staff education included critical thinking using case studies and role-playing exercises.19
Staff participated in skin-care training using an interactive video.30
McInerney et al.17 reported that using computerized charting and order entry helped
identify specific groups of patients at risk for developing PUs. A consult with a specially
trained nurse is automatically generated when patients are identified as high/very high
risk. Consults are also generated for two other patient groups considered high risk
(e.g., patients placed on a ventilator and patients receiving hemodialysis).
Monitoring Progress
To monitor progress at a large 528-bed hospital, a SKIN Bundle Compliance Tool was
developed. Nutrition-related items include noting completion of a nutrition consult,
orders written, and orders carried out.22
One rehabilitation hospital posted report cards on every unit allowing staff to track
progress against other units and unit goals.11
A large nonprofit health care system streamlined a skin product line, trimming it from
100 products to 24. Standardizing products controlled costs and reduced training.25
Four critical care units conducted extensive in-service education on a new wound care
product line and made vendor support available. Vendor clinical experts were available to
educate staff on new products. Dibsie recommends informing clinicians when
modifications are made to the skin protocol or product line.16
Nursing leadership, nursing staff and those from other disciplines (e.g., nutritionists,
respiratory therapists) compared current policies and procedures to clinical practice
guidelines. The Director of Informatics facilitated revisions12 of seven existing policies
into one.
Barriers
Reported barriers to implementing PU prevention programs included expansion of the
initiative to a larger scale,16 unmotivated staff,13,14,19 staff turnover,26,31-33 staff resistance,24,31
limited resources,12 inconsistent documentation,13,20,31 difficulties in exporting data,26and
miscommunication between electronic systems.20 One facility also faced increased PU rates.28
Examples of barriers described and ultimate solutions are as follows:
224
Acute Care
Dibsie16 reported that expanding initiatives from a 20-bed critical care unit to all nursing
units in two sites provided an extra challenge. Obstacles included the coordination of a
skin committee, difficulty in coordinating schedules, and keeping abreast of new
equipment that can contribute to PU development.
Chicano14 reported a challenge motivating staff who were relatively uninvolved in
planning and implementing initiatives. Perseverance of council members encouraged the
staff to finally accept the concept of shared governance and acknowledge that it can
positively influence patient care.
During the early stages of one initiative, staff members were not sufficiently committed.
Although a QI analysis indicated that two stage 4 PUs were the result of inconsistencies
in or lack of documentation, staff awareness and assessment, staff members denied that
the PUs were due to poor nursing care. Catania indicated that overcoming it took
additional education, mentoring, and support at the unit level.19
As reported by Lyder et al.,24 hospitals identified as a major barrier the view that PU
prevention was a nursing issue. The medical staff was reportedly the most resistive when
asked to play a role in PU prevention. Due to this mindset, considerable time was spent
re-educating various disciplines about their roles in PU prevention.
Young et al. reported that clinicians complained of time constraints, insufficient
computer resources, and competing goals. To address these concerns, clinicians were
allocated 4 paid hours to carry out responsibilities related to PU prevention, and web
access to library resources was added to the organizations intranet.12
Bales et al.13 reported unmotivated staff and lack of proper reporting and documentation.
Monthly to quarterly campaigns were launched to maintain staff motivation. Nursing
units that had zero HAPUs were recognized and awarded during campaigns.
LeMaster20 reported Braden scores (a scale for predicting PU risk) were not documented
at 100% according to policy. Patients were missed because of a communication failure
between two different electronic documentation systems. To address this, the facility
transitioned to a single, universal electronic record system.
Long-Term Care
225
Organization (TMF Health Quality Institute) and improving performance may have
helped resolve these issues.31
Barriers reported by Milne et al.28 included a climb in PU rates once strict monitoring of
processes was leveled off from weekly to monthly for 1 month. To overcome this barrier,
the wound team increased presence on the units, monitored charts more closely, provided
feedback to staff, and scheduled biweekly prevalence rounds.
Rantz 200133 indicated that staff turnover, especially the nurse coordinator, cancellations
of site visits at the last minute, and teams mired in the MDS [minimum data set]
assessment process and coding issues impeded initiatives. To address some of these
issues, multiple nurses were assigned responsibility for processes, accomplishments were
posted, and a quality manager was placed on staff to support care delivery.
Sustainability
Several acute and long-term care facilities reported sustainability of PU prevention programs.
Conducting quarterly prevalence studies and continually monitoring all HAPUs were key to
sustaining improvements at a large nonprofit health system. A focus on skin pigmentation and
the development of a skin fragility assessment tool were the most recent strategies introduced.22
McInerney17 indicated that publicizing improvements in PU rates will keep staff focused on
prevention efforts.
One 710-bed multisite facility reported that the overall culture change at the medical center
remains a work in progress. Therefore, PU incidence continues to be a measurement used in
organization-wide scorecards and staff bonus programs. Measuring performance routinely has
since become a priority facility-wide and at sister facilities.21 LeMaster20 reported a request by
staff members for items to provide both visual and auditory clues to sustain improvements at
their facility. Dibsie et al.16 reported staff members continued to meet on a monthly basis to
discuss skin issues, participate in quarterly prevalence data collection, and learn from the
medical centers expert WOCNs as well as from one another.
Walsh indicates the importance of maintaining gains, including keeping current regarding
initiatives for improved patient safety, changes in regulatory mandates, and changes in EBP
[evidence-based practices]. At a 540-bed acute care facility, RNs and LPNs must demonstrate
competency annually; monthly updates provided via intranet to staff include product changes.12
One rehabilitation hospital printed quarterly newsletters and attached them to paychecks. The
newsletters described findings, results, and new initiatives in PU management.11 According to
Bales et al.,13 sustainability requires awareness of key management skills and priorities, such as
strong leadership, involvement of staff in decision-making, and a desire to foster
interdisciplinary relationships.
Eleven long-term care facilities across seven states stated that managing the manual data
collection, faxing forms to the project office and creating clinical reports for distribution back to
the facilities on a weekly basis could not be maintained over the long term. By date of study
publication, more than 70 additional facilities were participating in the On-Time program.26 A
wound care coordinator position29 and a wound care committee32 were established to help sustain
program gains in two long-term care programs. Lastly, one 108-bed LTC facility noted that two
wound coordinators provided additional education. Monthly review of documentation and the
presence of multiple PU prevention interventions on units also helped to sustain improvements.28
226
Lessons Learned
Several programs implemented facility-wide reported key lessons learned. Recommendations
from one large nonprofit health system include the following: (1) use current leadership to
support staff; (2) disregard a spike in reported skin breakdown, which is probably due to staffs
increased awareness, education and reporting; (3) make staff accountable for success of
initiatives; (4) do not assume that the knowledge base between disciplines is equal; (5) stay on
task and celebrate successes.22
Seventeen hospitals reportedly liked the idea that PDSA cycles could be completed using a
small number of cases to identify improvement areas and implement potential interventions at a
reduced scale before implementing programs at full scale, saving time and human resources.24
Lessons learned in Courtney et al. include the following: (1) identify and involve the process
owner (unit manager) and S.O.S. (Safe our Skin) unit champion early in the project; (2) develop
a detailed training plan to include delegation and team work; (3) communicate the changes to
everyone; (4) streamline the process to make it as easy as possible; (5) celebrate success, give
recognition, spread the success; (6) define roles and responsibilities; (7) realize that you cannot
fix everything at one time; (8) support and commitment from senior leadership are critical to the
success and help to sustain the gains.21
Abel et al. reported lessons learned by 20 Medicare-certified skilled nursing homes while
working collaboratively with TMF, a State quality improvement organization.31 A TMF
representative provided the following recommendations:
Strive to transition from quality assurance to QI, moving from defect detection to defect
prevention, while making continuous improvement in the process of care delivery.
Incorporate interventions designed to address barriers to preventive care while sustaining
existing processes that have proven effective.
Publicly recognize front line staff successes.
Share data often.
Implement change on one unit/hall at a time to ensure staff buy-in.
Accept failure in the systems as an opportunity to improve.
Operationalize systems that use continued measurement to monitor and improve
performance that is reported to be below a designated threshold.
Allow staff to maintain a level of autonomy during design of the intervention.
Ensure accountability in following agreed-upon process changes.
Ensure consistency in a formalized staff training/orientation related to documentation
requirements.
Include staff that directly affect the process in the intervention design.44
227
A Midwest 710-bed teaching hospital estimated additional cost per case (in 2001) was
$3,037, and the additional overall cost related to PU development was $4,877,000. Based on
these figures, a reduction in the number of HAPUs by 50% to 5% would reduce overall costs by
$2,438,000.00. This amount would not preclude the anticipated significant improvements in
patient satisfaction, length of stay/quality, staff awareness of skin integrity issues, and risk
management issues.21
The most recently reported costs in a long-term care setting were from 2010 in a four-group
comparison study that assessed the value of a bedside electronic health record PU initiative.27
Total costs for the 3-year evaluation for the groups of facilities implementing technology
increased $15.11 (12.5%) for Group 1 and $16.89 (9.6%) for Group 2, while those for the
comparison groups did not. Rantz et al. attributed cost increases to the cost of technology,
including maintenance and support and ongoing staff training to use the EMR system.27
In 2009, a 151-bed Midwest skilled facility described cost savings 4 years after program
implementation. The authors estimated cost savings per PU/per month was $1,617; monthly
savings, $10,187; and yearly savings greater than $122,000.29 Lastly, in 2006, Rosen et al. stated
based on a mean cost of $2700 to treat a single stage II PU,45 reducing the incidence of ulcers
by approximately 15 over 12 weeks would yield savings of approximately $40,000. Less than
$10,000 was distributed as financial incentives.
228
Abel et al. reported that higher QI scoring and greater improvement in scores equaled lower
PU incidence rates. He indicated that 10 facilities with the highest quality indicator scores at remeasurement showed a trend toward lower PU incidence rates than the 10 facilities with the
lowest QI scores at re-measurement (-4.6% vs. -2.6%) (S = 125.5, p = 0.07). In addition, the 10
facilities with the greatest improvement in QI scores had significantly lower PU incidence rates
than the 10 facilities with the least improvement in QI scores (-6% vs. -1.2%) (S = 131.0,
P = 0.03). Later, Abel noted that study results demonstrated a relationship between PU incidence
rates and improvement in QI scores, suggesting a relationship between the process of care
measures and patient outcome (i.e., PU incidence). He adds that improved performance was
primarily noted in measures that were less dependent on staffing, i.e., use of support surfaces and
risk assessment tools rather than patient turning.
Milne et al.28 reported on revisions of care in two units with greatly improved PUs. On a
pulmonary focused unit, PUs were observed at the ear/scalp junction of 25% of patients. With
the adaption of the nasal cannula, PU rates on the ear/scalp junction were reduced within
2 months to less than 3%. On a separate unit, 33.8% of SCI and trauma patients were suffering
from sacral and heel sores due to immobility and sensory deficit. Due to several focused
initiatives, including pressure redistribution support surfaces, PU rates dropped to 2.9%, a 30.9%
reduction.
Rantz 200133 observed that intensive consultation contributed to a greater improvement in
MDS QI measures for several outcome measures, including two focused on PUs.
229
Moderate
Evidence or
Potential for
Harmful
Unintended
Consequences
Negligible
Estimate of
Cost
Moderate
Implementation Issues:
How Much do We
Know?/How Hard Is it?
Moderate/Moderate
References
1.
2.
3.
4.
5.
6.
7.
230
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
231
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
232
40.
41.
42.
43.
44.
45.
233
Table 1, Chapter 22. Large trials (n > 500) evaluating the health outcome effects of intensive insulin therapy
Patient Population;
Implementation Context Glucose Target, Inpatient BG Mortality T v C (RR, 95% CI)
Diabetes Mellitus (%);
T v C (mg/dL)
Achieved,
Single or Multicenter;
T v C (mg/dL)
Country;
Study Quality
SICU
Insulin protocol was
80-110
103 v 153 ICU mortality 4.6 v 8% (p=0.005
13
developed and use overseen
v 180-200
(p<0.001)
unadjusted)
by study investigators.
Single center
RR 0.42 (95% CI 0.22-0.62);
10
Hospital mortality: 7.2 v 10.9%
Belgium
Fair
(p=0.01)
RR 0.66; 95% CI 0.48-0.92
Neurosurgical ICU
Efforts made to limit nursing 80-110 v 180-200
92 v 143
6-month mortality: 74.0 v 72.0%
NR
turnover. New nursing staff
(p<0.001)
(p=0.82)
Single center
worked with experienced
58
Italy
staff.
Fair
MICU
Study conducted in a
80-110 v
111 v 153
ICU mortality: 24.2 v 26.8%
16
hospital that had already
180-200
(p<0.001)
(p=0.31)
conducted similar IIT study in
Hospital mortality: 37.3 v 40.0%
Single center
59
SICU patients. Authors note
(p=0.33)
Belgium
RR 0.93; 95% CI 0.81-1.08
Fair
the nurse:bed ratio of 2.5
90d mortality: 35.9 v 37.7%
was not changed for study.
(p=0.53)
MICU
No details provided
80-110
112 v 151
28d mortality: 24.7 v 26%
30
v 180-200
(p<0.001)
(p=0.74)
Multicenter
RR 0.95, 95% CI 0.70-1.28
21
90d mortality: 39.7 v 35.4%
Germany
Fair
(p=0.31)
MICU/SICU
Characteristics from each
80-110
117 v 144
ICU mortality: 17.2 v 15.3%
17 T, 22 C (p=0.031)
study site were reported.
v 140-180
(p<0.001)
(p=0.41)
Hospital mortality: 23.3 v 19.4%
Multicenter
Median nurse:bed ratio was
60
(p=0.11)
Europe
2. ICUs ranged widely in
Fair
28d mortality: 18.7 v 15.3%
size, patient volume, and
(p=0.14)
number of glucometers per
ICU.
234
Hypoglycemia
Definition
(mg/dL),
rate ,T v C,
RR (95%CI)
<40,
5 v 0.76%,
RR 6.65 (2.8315.62)
Other Reported
Outcomes*
TvC
Renal replacement
4.8 v 8.2% (p=0.007)
Sepsis
4.2 v 7.8% (p=0.0003)
<50, 93.8 v
62.8%, p<0.001
Renal replacement
(patient days) 519 v 523
(p=0.75)
Table 1, Chapter 22. Large trials (n > 500) evaluating the health outcome effects of intensive insulin therapy (continued)
Patient Population;
Implementation Context Glucose Target,
Diabetes Mellitus (%);
T v C (mg/dL)
Single or Multicenter;
Country;
Study Quality
MICU/SICU
24/7 ICU coverage by
80-110
32 T, 48 C (p<0.001)
intensivists. Protocol
v 180-200
designed by multidisciplinary
Single center
19
team at study site.
Saudi Arabia
Fair
Physicians and nurses
attended training sessions
before and during study.
MICU/SICU
Three month staff training 80-110 v 180-200
13 T, 12 C (p=NS)
period before study.
Single center
54
Colombia
Fair
MICU/SICU
Pre-trial pilot studies carried 80-108 v <180
20
out to test/improve insulin
13
Multicenter International protocol. Final computerized
Fair
insulin protocol algorithm
accessible to study sites
through a central Web site.
No clear explicit training prior
to study.
Acute MI
No details provided
126-198 v NR
39
Multicenter CCU
61
Sweden
Fair
Acute MI
No details provided
group 1 and 2:
77 established DM; 23
126-180
new DM
group 3: NR
of < 1y
62
Multicenter Europe
Poor
Inpatient BG
Achieved,
T v C (mg/dL)
115 v 171
(p<0.001)
120 v 149
(p,0.001)
115 v 144
(p<0.001)
24 hours:
T: 172.8 (59.4)
C: 210.6 (73.8)
p < .001
Hypoglycemia
Other Reported
Definition
Outcomes*
(mg/dL),
TvC
rate ,T v C,
RR (95%CI)
ICU mortality: 13.5 v 17.1%
< 40, 28.6 v
Renal replacement
(p=0.70)
3.1%, p < 0.001 11.7 v 12.1% (p=0.89)
RR 1.09 (0.70 -1.72)
Sepsis 36.9 v 40.9%
Hospital mortality: 27.1 v 32.3%
(p=0.35)
(p=0.19)
RR 0.84 (0.64 -1.09)
ICU mortality: 33.1 v 31.2%; RR <40, 8.3 v 0.8%
1.06 (0.82-1.37)
28d mortality: 36.6 v 32.4%;
RR 1.1 (0.85-1.42)
28d mortality: 22.3 v 20.8%
(p=0.17)
RR 1.07 (0.97-1.18)
90d mortality: 27.5 v 24.9%
(p=0.02)
RR 1.14 (1.02-1.28)
<54, 15.0 v 0%
3 month mortality:
(p < .001)
12.4% v 15.6%, p = NS
1 year mortality:
18.6% v 26.1 %,
RR 0.69; 95% CI 0.49-0.96
24 hours:
Adjusted 2-year mortality:
< 54, Gr 1 v Gr2
group 1: 163.8 Group 1 v 3 = 1.19 (0.86 - 1.64) v Gr3: 12.7 v 9.6
v 1.0
(54.0), group 2: Group 2 v 3 = 1.23 (0.89 - 1.69)
163.8 (50.4),
group 3: 180.0
(64.8)
p = .0001
235
Table 1, Chapter 22. Large trials (n > 500) evaluating the health outcome effects of intensive insulin therapy (continued)
Patient Population;
Implementation Context Glucose Target,
Diabetes Mellitus (%);
T v C (mg/dL)
Single or Multicenter;
Country;
Study Quality
Stroke
Conducted as a pragmatic 72-126 v <306
17
trial as part of routine clinical
63
care. No clear explicit
Mutlicenter Britain
Poor
training prior to study.
Inpatient BG
Achieved,
T v C (mg/dL)
Hypoglycemia
Other Reported
Definition
Outcomes*
(mg/dL),
TvC
rate ,T v C,
RR (95%CI)
24 hour mean
90-day mortality:
< 72 for > 30
difference I v C
30.0% v 27.3%,
mins, 15.7,
(95% CI): 10.3 OR (95% CI) = 1.14 (0.86-1.51) control group rate
90-day severe disability:
NR
(4.9 - 15.5), p <
35.1% v 36.0%,
.0001
OR (95% CI) = 0.96 (0.70-1.32)
Notes: Abbreviations: BG = Blood glucose; d = day; CCU = coronary care unit ; ICU = intensive care unit; MICU = medical intensive care unit; SICU = surgical intensive care
unit; C = comparator; DM = diabetes mellitus; NR = not reported; NS = not statistically significant; RR = relative risk; T = treatment
Other reported outcomes include renal replacement, infection, cardiovascular events, and long-term disability.
Quality was assessed using criteria from the U.S. Preventive Services Task Force.
SI unit conversion for glucose: 1 mg/dL x 0.0555 = 1 mmol/L.
* Infection includes wound infection, urinary tract infection, or pneumonia; or a combination of these.
Morning blood glucose.
Average of blood glucose measurements, not otherwise specified.
236
However, over the last decade, IIT trials have failed to replicate these results consistently.
Our recent meta-analysis of 21 randomized controlled trials (RCTs), including a total of 14,768
inpatients, found no effect of IIT on short-term mortality (RR 1.00, 95% CI 0.94 to 1.07) (see
Figure 1), with remarkably little heterogeneity among studies (I2 = 0.0%, p=0.463).12 The body
of evidence is strongest in ICU settings.
Figure 1, Chapter 22. Short-term mortality in studies of intensive insulin therapy, by inpatient
setting and condition
Table 2 summarizes the main characteristics and results from the largest trials (n > 500
patients). The NICE-SUGAR trial, with 6,104 medical intensive care unit (MICU) and SICU
patients, was by far the largest and likely provides the most generalizable results, given its size,
multicenter design, and the broad ICU population included.13 Patients randomized to a lower
blood glucose target (80-110 mg/dL) had higher 90-day mortality than those assigned a higher
blood glucose target (140-180 mg/dL), with approximately one excess death for every 39
patients treated with the more intensive protocol (n = 6,022 with 90-day outcomes reported; RR
1.14; 95% CI 1.02 to 1.28).
237
Morbidity outcomes have also been assessed. Trials of IIT failed to demonstrate consistent
benefits in reducing renal failure or length of stay. The effects of IIT on infection are mixed. The
2001 Van den Berghe SICU trial did find IIT reduced the incidence of sepsis,10 whereas the
NICE-SUGAR trial did not.13 Trials also reported a variety of other infection-related outcomes,
including wound infections, pneumonia, and urinary tract infection. A pooled analysis of these
trials found a nonsignificant reduction in infection, though the results were quite heterogenous
(RR 0.68; 95% CI 0.36 to1.30, I2 = 56.3%, P=0.033).12
Several trials have reported results since January 2010 (the end of the search period of our
systematic review). Two trials found no benefit of IIT on neurologic or mortality outcomes in
patients with traumatic brain injury or stroke.14,15 One trial did find postoperative IIT reduced
wound infections in diabetic patients who had undergone partial gastrectomy (7.6% v 18.4%, p =
0.03), but the trial had several methodologic flaws.16 Finally, the recent RABBIT 2 trial was
among the first to compare the effects of a subcutaneous basal-bolus insulin regimen with
sliding-scale insulin on health outcomes.17 In this trial, 211 noncritically ill general surgery
patients were randomized to either a basal-bolus insulin regimen using insulin glargine and mealtime glulisine or a sliding-scale insulin regimen (whereby fixed doses of insulin are given in
response to specific glucose readings), with a goal glucose target in both groups of 100-140
mg/dL. The basal-bolus group achieved significantly better glycemic control and a lower
incidence of wound infections (2.9% vs. 10.3%, p = 0.05), but several methodologic issues,
including poor outcome ascertainment methods and no blinding of outcome assessors, threaten
the validity of results.
238
Figure 2, Chapter 22. Risk for hypoglycemia in studies of intensive insulin therapy in various
inpatient settings
239
Glucose Targets
Not surprisingly, we found that protocols targeting higher blood glucose ranges were
generally associated with lower rates of hypoglycemia. This observation echoes findings from
the trial literature in which insulin infusions were used to target strict (80-110 mg/dL) or
moderate (140-200 mg/dL) glucose goals. The rates of severe hypoglycemia were substantially
higher in the strict glucose target groups.
We also found protocols that achieved mean blood glucose <120 mg/dL were not
consistently safe, even when clinicians used relatively sophisticated computerized
algorithms.23,24 Two observational studies evaluated the safety of phasing in progressively
stricter glucose targets over time.25,26 One of these was a large single-center retrospective study
evaluating the effects of an increasingly aggressive IIT policy in the ICU.25 The authors found a
nearly four-fold increase in the incidence of hypoglycemia as the institution moved from no
insulin protocol to IIT with a target of 80-130 mg/dL and finally to a target of 80-110 mg/dL.25
The infusion protocol details were not available. A second study of a relatively simple infusion
protocol reported a doubling of the rate of severe hypoglycemia as the glucose target moved
from 120-150 mg/dL to 80-110 mg/dL, although the overall rate of severel hypoglycemia
remained less than 5 percent.26
Computerized Protocols
Most protocols studied have used a paper-based algorithm to guide nurses in making insulin
dose adjustments and timing glucose measurements. In recent years, computer-based algorithms
have become available and, in 2010, the first RCT comparing these algorithms to a paper-based
algorithm was reported.31 This multicenter trial of 153 MICU patients compared a paper-based
algorithm to the computerized Glucommander system, which directed insulin dosing and glucose
monitoring timing using glucose measurements at the patients bedside. The glucose target was
80-120 mg/dL. Patients in the computer-algorithm group had fewer instances of marked
hyperglycemia (glucose > 200 mg/dL, 11.7% vs. 25%, p = 0.05, but a similar rate of severe
hypoglycemia (< 40 mg/dL:, 3.9% vs. 5.6%, p = NS). An older observational study of the same
computerized system had reported similar results.32 However, a recent small observational study
found only minor improvements in glucose control using a computerized protocol and no change
in hypoglycemia rates.33 This study also found that using the computerized protocol led to more
240
frequent glucose testing and insulin dose adjustments. Finally, another recent observational study
found that fewer dosing errors occurred with a computer-based protocol than with a paper-based
protocol.34
Nutrition
Most insulin protocols neither coordinate insulin dosing with patient nutrition nor provide
detailed nutritional guidance. In one RCT, 337 critically ill patients were randomized to either a
carbohydrate-restrictive strategy or an insulin infusion regimen targeted to blood glucose levels
of 80-120 mg/dL.44 Although the glucose level achieved in the carbohydrate-restrictive group
was higher than in the infusion group (144 vs. 134 mg/dL, p = 0.03), the difference was modest
and the rate of hypoglycemia was substantially lower in the carbohydrate-restrictive group (3.5%
vs. 16%, p < 0.01). The results suggest that more-intensive nutritional strategies may be a
promising adjunct, or alternative to, IIT. An observational study of the SPRINT protocol, which
directly prescribes both insulin dosing and dietary intake, found that it improved glycemic
control and reduced the risk of severe hypoglycemia.28
241
RABBIT 2 surgery trial17 and several small, single-center trials in general medical47-49 and
gastric bypass populations50 found that basal-bolus regimens were more effective in lowering
blood glucose than SSI, although both strategies had similar rates of hypoglycemia. The
RABBIT 2 surgery trial is the only one to have reported the effects of basal-bolus insulin on
health outcomes.17
with use of IIT, although theirs was an uncontrolled study, and they did not report overall
hypoglycemia rates. In contrast, another institution went from no glycemic control policy to a
normal glucose target over 5 years in a large population of critically ill patients (n = 10,456);
these investigators reported markedly increased rates of hypoglycemia and a trend to increased
mortality.25 These results may well reflect the difficulty with broad, rapid implementation of
aggressive glucose control practices.
comparative effects of different subcutaneous insulin regimens has not been well studied
in non-surgical populations.
IIT is a complex endeavor requiring buy-in from nurses and physicians. Implementation
of IIT in a given setting is likely best done iteratively, with multidisciplinary involvement
and training, and using real-time data to inform continuous quality improvement of the
process.
Moderate-tohigh evidence
it doesnt help
Evidence or
Potential for
Harmful
Unintended
Consequences
High
(hypoglycemia)
Estimate of
Cost
Low-tomoderate
Implementation Issues:
How Much do We
Know?/How Hard Is it?
N/A
References
1.
2.
3.
4.
5.
244
6.
7.
8.
9.
10.
11.
20.
12.
21.
13.
22.
23.
14.
15.
24.
25.
26.
27.
28.
16.
17.
18.
19.
245
29.
38.
30.
39.
31.
40.
41.
32.
33.
42.
34.
43.
44.
35.
45.
36.
46.
47.
37.
246
48.
56.
49.
57.
50.
58.
59.
60.
61.
62.
63.
51.
52.
53.
54.
55.
247
248
Published guidelines from the American College of Radiology, the European Society of
Radiology,12 and the Canadian Association of Radiology13 all recommend the above measures.
The original review of this topic for Making Health Care Safer (2001) also recommended
volume expansion with normal saline and avoidance of high-osmolar contrast. The 2009
American College of Cardiology/American Heart Associated guidelines for percutaneous
coronary interventions also recommend avoidance of high-osmolar contrast media.2
In addition to standard care, several interventions have been widely studied to prevent CIAKI. These practices are the focus of this review:
Volume expansion with intravenous sodium bicarbonate
Administration of n-acetylcysteine
Use of iso-osmolar (instead of low- or high-osmolar) contrast media
Prophylactic renal replacement therapy (dialysis)
Administration of HMG CoA-reductase inhibitors (statins)
249
generally not stable by 5 years after publication.15 The revised search identified 32 studies, of
which 20 were confirmed to be meta-analyses after full-text review (the others were largely
narrative reviews). These 20 meta-analyses evaluated the effectiveness of 5 distinct interventions
for preventing CI-AKI:
Hydration with intravenous sodium bicarbonate (N=11)16-26
Administration of oral N-acetylcysteine (NAC, N=3)27-29
Use of iso-osmolar radiocontrast media (N=3)30-32
Prophylactic renal replacement therapy (RRT, N=1)33
Administration of HMG CoA-reductase inhibitors (statins, N=1)34
In addition, one study35 evaluated the combination of NAC and bicarbonate in preventing CIAKI compared with NAC alone.
We followed the methodology previously outlined by Whitlock36 for incorporating
previously published systematic reviews into a new review. Each identified review was
evaluated for quality using the AMSTAR checklist37, and information was extracted on the
interventions and outcomes assessed, the study populations (including the types of radiologic
studies for which contrast media was used) and sample size, the definition of CI-AKI used, and
the overall conclusions of the review (Table 1).
250
heterogeneity was found in all 11 meta-analyses, and all of the meta-analyses that included
unpublished studies found evidence of publication bias.
Therefore, we conclude that sodium bicarbonate therapy appears to offer only marginal
benefit at best over routine saline volume expansion, and the primary literature suffers from
significant limitations. Routine bicarbonate administration cannot be recommended to prevent
CI-AKI.
251
contrast media (LOCM, of which there are several agents). The most recent and largest metaanalysis32 identified 36 randomized controlled trials published before December 2009. This
meta-analysis was high quality, scoring 11 on the AMSTAR scale. It did not find a statistically
significant reduction in biochemical CI-AKI for iso-osmolar contrast compared with all LOCM
agents (pooled OR 0.77, 95% CI 0.56 to 1.06). However, a subgroup analysis did find that isoosmolar contrast was associated with a reduction in CI-AKI in studies comparing iodixanol to
one specific low-osmolar agent, iohexol (pooled OR 0.25, 95% CI 0.11-0.55, N=10 trials). This
finding was also noted in the other two meta-analyses of this question.30,31 None of the metaanalyses evaluated the effect of iso-osmolar contrast media on clinical outcomes.
Other than this advantage of iodixanol over the specific agent iohexol, there is therefore no
convincing evidence supporting the routine use of iso-osmolar contrast. The 2009 ACC/AHA
guidelines for percutaneous coronary intervention2 recommend use of iso-osmolar contrast or use
of LOCM other than iohexol. This is a change from the 2007 guidelines, which specifically
recommended use of iso-osmolar agents.
Administration of Statins
One recent meta-analysis34 identified 6 small RCTs evaluating the effect of statins on
biochemical CI-AKI. There was no overall beneficial effect of statins on prevention of CI-AKI.
252
One potential harm is that administration of intravenous fluids may increase the risk of
clinically significant congestive heart failure (CHF) in patients with a known diagnosis of CHF.
However, the largest meta-analysis of intravenous bicarbonate administration did not find an
increased incidence of symptomatic CHF.25
Are There Any Data About Costs of the Patient Safety Practice?
We did not identify any formal cost-effectiveness analyses of the various modalities
proposed to prevent CI-AKI published since 2007. Interventions such as bicarbonate and NAC
are low cost, whereas iso-osmolar contrast media (IOCM) is more costly than standard LOCM.
One cost-effectiveness analysis demonstrated that IOCM is cost-effective compared with
LOCM,41 but this analysis was based on earlier, more favorable estimates of the benefits of
IOCM that have not been borne out in subsequent trials or meta-analyses. We also identified one
cost-effectiveness analysis of prophylactic RRT published in 2006,42 which found that
prophylactic RRT might be cost-effective only in a subset of patients with stage 4 chronic kidney
disease. This analysis was also based on favorable treatment estimates that have not been
confirmed in formal systematic reviews.
253
Low
Evidence or
Potential for
Harmful
Unintended
Consequences
Negligible
Estimate of
Cost
Low
Implementation Issues:
How Much do We
Know?/How Hard Is it?
Little/Not difficult
References
1.
2.
3.
4.
254
5.
Weisbord SD, Palevsky PM. Contrastinduced acute kidney injury: short- and
long-term implications. Semin
Nephrol;31:300-9.
6.
7.
8.
9.
10.
11.
20.
12.
21.
13.
22.
23.
14.
15.
24.
16.
25.
26.
17.
27.
18.
28.
19.
29.
255
30.
39.
31.
40.
41.
32.
42.
33.
34.
35.
36.
37.
38.
256
257
contrast, RRSs are non-specific and address a panoply of conditions. Therefore, we do not
include the disease-specific systems (BATs and HATs etc.) in this review.
A Rapid Response System generally has four components:
Criteria for notifying the response team and a system for activating it (the Afferent
Limb). The criteria usually include vital signs (single trigger criteria or more complicated
algorithms including aggregate and weighted early warning scores). However, in some
cases a clinician or family member might initiate activation, based on clinical judgment
and concern even though specific activation criteria are not met (e.g., heart rate >130).
The response team the Efferent Limb. Refers to personnel and equipment (can be led by
a critical-care physician, other physician, or by an nurse or respiratory therapist). Team
composition varies based on local needs and human resources.
Feedback loop to collect and analyze event data and quality improvement.
Administrative component, coordinating resources, staff, equipment, and education.
Jones et al2 also cites importance of support of leadership and administration, use of criteria
that are not too complicated (argues for simple vital signs triggers as opposed to complicated
early warning scores), education of the personnel on the team regarding the criteria (including
possibly simulation training), and involvement of physicians who can facilitate ICU transfers and
end-of-life planning. In a narrative review of data from the MERIT trial (the only multi-center
cluster randomized trial of RRS) and subsequent data, Jones et al. also note that RRSs exhibit a
dose response curve, where utilization rates (number of RRS activations) positively correlate
with reduction in the incidence of cardiac arrest. The authors found that a utilization rate of
approximately 17 RRS calls/1000 patient admissions is required to reduce the incidence of
cardio-respiratory arrest by 1/1000 admissions. Given this relationship, many hospitals have
sought to increase utilization of their RRSs to realize improvements in outcome.
that can be directly triggered should also help to circumvent the delays that typically occur in
summoning a physician or higher level expertise. Together, these two elements (afferent and
efferent limbs) should catch treatable problems early before they are life-threatening. Finally, the
feedback component should help make clinicians aware of the need and benefits of using the
RRS, the quality improvement component should ensure improvement or maintenance over time,
and the administrative component should ensure that adequate resources are available to respond
to patient rescue needs.
259
260
cardio-respiratory arrest (codes) rates outside of the ICU and Emergency Departments since
these patient populations are not part of the exposure group (RRSs do not respond to these
locations), yet often hospital-wide rates were reported. One study9 included ICU arrests in their
analysis, concluding there was no effect, though data presented on their non-ICU code rate
showed a statistically significant difference. Cardiac arrest rates are also affected by changes in
patient casemix over time and the frequency of do-not-resuscitate orders and terminal illness,
which most studies did not account for.
Most studies reported in-hospital mortality. Only one reported longer-term mortality (such as
180-day mortality) reflecting patient survival more accurately. Most other outcomes reported,
such as the cardiac arrest rate, unanticipated intensive care unit admissions, or other health care
utilization measures are also indirect outcomes. In terms of precision, we did not identify any
additional studies that would have been assessed as high-quality in the 2009 meta-analysis7 all
would have been fair or poor quality. Evidence for association of RRSs with lower in-hospital
mortality was not strong. A summary table is located below (Table 1).
Table 1, Chapter 24. RRS summary table: effectiveness
Author, Year Description of PSP
Study Design Outcomes: Benefits*
Anwar ul,
PICU physicians (Pediatric MET)
Pre-post
Cardiac arrest: Y
10
2010
11
Bader, 2009 Nurse led, with Critical care outreach component
Pre-post
Mortality (non ICU): NR
(proactive rounding on ICU-discharged patients
Cardiac arrest: Y
Benson,
Advanced practice nurses (APN) with intensivists
Pre-post
Mortality: Y
12
2008
and other disciplines involved as needed
Cardiac arrest: NR
Campello,
ICU physician and ICU nurse
Pre-post
Mortality: Y
13
2009
Cardiac arrest: Y
9
Chan, 2008 Respiratory therapist and 2 ICU nurses (RRT model) Pre-post
Mortality: N
Cardiac arrest (hospital-wide): N
Cardiac arrest (non-ICU): Y
Gerdik,
RRT (specifics not described) including option for
Pre-post
Mortality: N
14
2010
patient and family activation
Cardiac arrest: NR
Hanson,
PICU fellow, resident, nurse and respiratory therapy Pre-post
Mortality: N
15
2009
Cardiac arrest (ward): N
16
Hatler, 2009 ICU nurse and respiratory therapy (RRT model)
Pre-post
Cardiac arrest: NR
Konrad,
17
2010
Kotsakis,
18
2011
Laurens,
19
2011
Lighthall,
20
2010
Pre-post
Pre-post
261
Pre-post
Pre-post
Pre-post
Pre-post
Pre-post
Pre-post
Pre-post
Pre-post
Description of PSP
MET: critical care physician and nurses
262
different settings (intensive care unit, emergency room), as well as different management,
administrative staff, or quality oversight involvement. The majority of studies utilized
interdisciplinary teams comprised of at least one physician and one nurse. However, several
studies examined alternative RRS configurations. For example, two studies examined systems
that leveraged nurse leaders or nurse liaisons as primary first responders.12,30 Implementation
processes varied widely, often guided by the Institute on Healthcare Improvement (IHI)
suggestions or using IHI materials. Education and promotion of the new service was often a
factor, although actual staff training (such as simulation training) was uncommon. A variety of
different objective criteria were used for calling the team, and some interventions depended on
nurses clinical judgment; a few studies also developed and promoted a system for family or
patient initiation of the team.
Fifteen studies met our inclusion criteria for studies of the implementation processes
surrounding Rapid Response Systems. Eleven of these studies used quantitative methods,
primarily for evaluating the impact of a change in the implementation process for an RRS
program, and four used primarily qualitative methods such as interviews or focus groups of staff
regarding RSS implementation issues.31-33 The majority of implementation studies were
conducted in academic hospitals; however two studies specifically detailed implementation
efforts within community hospitals.31,34 Another study also examined the effects of separating
the overall emergency response system into two teams with different activation criteria and
processes in order to increase utilization.35 Results indicated significant increases in utilization
(15.7calls/1000 admissions vs. 24.7 admissions/1000admissions, p < .0001) after changes were
implemented.
Activation criteria and reasons for activation were focal study topics related to RRS
implementation. Several studies included subjective activation criteria (e.g., staff were worried
that a patient was at risk for an adverse event) in addition to traditional activation triggers based
on vital sign abnormalities.8,35 For example, one study that examined data from the MERIT trial
found that MET hospitals were 35 times more likely to activate their emergency response team
based upon this worried criteria compared with control hospitals (14.1% of activations vs.
0.4% of activations, p < .001).8
Descriptions of themes in the implementation processes included the categories of
technology and tools, staff and training, and barriers and facilitators. In terms of technology and
tools, no studies reported use of technology (such as computerized alerts) in RRS
implementation. Tools mentioned included changing activation criteria, triggers, or activation
methods, including one study changing to mandatory activation based on alert criteria; and
review of events, feedback, and rewards. In terms of staff and education, several implementation
studies brought on new staff, such as a nurse educator or liaison. Most studies indicated that
implementation processes explicitly included educational activities; however, these varied in the
degree to which they were strictly information-based (e.g., emails, meetings) or included
dedicated training and practice opportunities for either RRS members or staff. The majority of
studies also explicitly noted that on-the-job cognitive aids such as posters with activation criteria
or badge cards listing activation criteria were included. Finally, barriers and facilitators
mentioned included knowledge of activation criteria and other knowledge and attitudes about the
RRS; communication, teamwork, and lack of criticism for calling the team; perceptions about the
teams helpfulness to nurses and patients; and the importance and role of RRS champions.
One study specifically examined MET processes over time with the maturation of the MET
(and therefore potentially higher team skill level and more acceptance from ward staff). The
263
study found that the proportion of patients with delayed MET activation was significantly lower
(40.3% vs. 22%, p <.001) and that the proportion of patients with unplanned ICU admissions
was lower in a later cohort (31.3% vs. 17.5%, p < .001). A summary table is located below
(Table 2).
Table 2, Chapter 24. RRS summary table: implementation studies
Author, year
Adelstein,
36
2011
Buist, 2007
30
Calzavacca,
37
2010
Chen, 2010
Cretikos,
38
2007
Donaldson,
33
2009
Foraida,
39
2003 ,
DeVita,
3
2004
Genardi,
34
2008
Jones,
35
2006
Jones,
40
2006
Jones,
41
2010
Shapiro,
32
2010
264
Implementation Themes
Tools: centralized activation system, review of all
events, automatic escalation to code team if MET did
not respond within 30 min
Staff/training: nurse educator for training and
compliance
Staff/training: nurse liaison, development and
education
Barriers/facilitators: Maturation of system over time
Williams,
31
2011
Implementation Themes
Barriers/facilitators: Both executive and managerial
champions were important; were skilled
communicators, well-respected and familiar with
institutional culture. Champions were educators,
advocated for RRT, built relationships, and navigated
boundaries between professions/units.
Barriers/facilitators: advantages of RRT to nurses
(develops skills, autonomy, resource and way to
circumvent unit problems), perceived benefits for
patients; degree of teamwork with RRT; RRT skills;
concerns about activating an RRT
265
Despite their strong face validity, RRSs have exhibited mixed results in the literature. There
are several potential explanations for this--none mutually exclusive. The afferent limb can
provide clear definitions to identify which patients are likely deteriorating and can educate staff
on those definitions. However, activation triggers were originally developed through clinical
chart review of patients who had arrested or been transferred to the ICU, and subsequent
attempts to improve upon this model have not generated a better approach. Studies of aggregate
scores, weighted scores, and single parameter triggers have not demonstrated clear superiority of
one over another.2 Confounding this approach is the way that vital signs, which constitute most
of the data for afferent limb systems, are collected. On general wards, vital signs are, at best,
collected every 4 hours and more typically every 6 or 8 hours, leaving ample time for
deterioration to develop unrecognized. The fidelity with which vital signs are collected and
recorded is also known to be poor,1 amplifying the problem. Finally, vital signs are not the only
variable predicting risk of deterioration. Weighted and aggregated scores try to address this
issue, but the interconnectedness of these changes is complex and varies with specific
populations.
There are also a number of issues with the implementation of the efferent limb (the RRS
team). Optimal team composition is unknown, including the structure (including a physician or
not, level or training, and overall team composition), and whether the RRS should be unified
with the code team or be separate not only in function but in personnel. Hospitals are reluctant to
fund free standing RRSs whose only responsibility is to attend to deteriorating patients and/or
arrests. As a result, RRS team members need to leave other duties (often caring for critically ill
patients in the ICU) to respond. This may limit the available resources they can bring to the ward
patient and risks harm to the patients they have stepped away from. Restricted financial
resources may also impact the RRSs ability to selfaudit and evaluate code events and
unanticipated ICU transfers that occur outside an RRS intervention. As a consequence, the RRS
cannot make appropriate assessments in order to improve systematically. Efforts to improve
utilization may likewise suffer, especially given evidence that utilization (dose) matters, that
utilization can be improved with changes in implementation strategies, and that many programs
have low utilization rates. Utilization of RRSs is reported to be low often because of issues with
the culture of safety, including reluctance on the part of the ward staff to activate the team.
Finally, there are a number of issues regarding how outcomes in RRS studies are measured.
Cardiac arrests and hospital mortality can be affected by many other factors such as patient
characteristics and other aspects of care, including trends over time in reducing hospital mortality
and length of stay and in caring for more terminally ill patients outside the hospital setting.
Additionally, several metrics commonly used to evaluate RRSs count patients who are not
exposed to the intervention (i.e., total hospital mortality), potentially affecting the results.
Unfortunately, using metrics such as preventable general ward-only mortality is more difficult
and potentially introduces bias (chart review to determine preventability of a death).
In summary, RRSs are clearly associated with decreased rates of cardiopulmonary arrest, but
the question of whether RRSs as currently defined and implemented affects mortality is unclear.
Insufficient evidence exists on the impact of context, different implementation strategies, or RRS
structure. RRSs are not likely to realize their full potential for improving outcomes without
accurate, more frequent (possibly even continuous) and integrated patient specific data to inform
the afferent limb, an understanding of what team structure and training works best, greater
commitment to fully support RRSs so they can carry out all necessary functions unencumbered,
266
a greater focus on patient-centered care and patient safety, and improved measurement and
reporting. A summary table is located below (Table 3).
Table 3, Chapter 24. Summary table
Scope of the
Strength of
Problem Targeted by Evidence for
the PSP
Effectiveness
(Frequency/Severity) of the PSPs
Common/High
Moderate
Evidence or
Potential for
Harmful
Unintended
Consequences
Low
Estimate of
Cost
Moderate
Implementation Issues:
How Much do We
Know?/How Hard Is it?
Moderate/Moderate
References
1.
2.
Jones DA, DeVita MA, Bellomo R. Rapidresponse teams. N Engl J Med 2011;
365(2):139-46.
3.
4.
5.
6.
7.
8.
267
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
Medina-Rivera B, Campos-Santiago Z,
Palacios AT, Rodriguez-Cintron W. Critical
Care and Shock: The effect of the medical
emergency team on unexpected cardiac
arrest and death at the VA Caribbean
Healthcare System: A retrospective study.
2010; 13:98-105.
22.
23.
24.
25.
268
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
269
270
different professions (physicians, nurses, pharmacists) work together and with patients (and their
families) to ensure the accurate and consistent communication of medication information across
transitions in care.
Figure 1, Chapter 25. Overview of medication reconciliation
The World Health Organization (WHO) has prioritized medication reconciliation as one
of five top patient safety strategies, within the Action on Patient Safety: High 5s.7 National
campaigns targeting the reduction of preventable patient adverse events such as the Institute for
Healthcare Improvements 100,000 Lives Campaign in the United States (U.S.) as well as the
Canadian Patient Safety Institutes Safer Healthcare Now! have championed medication
reconciliation as one of a few core interventions. Furthermore, accreditation authorities such as
The Joint Commission in the U.S. and Accreditation Canada made medication reconciliation best
practices a mandatory requirement for various health care settings. However, of note, The Joint
Commission no longer formally scores medication reconciliation during accreditation surveys,
although the latter loosely remains part of the National Patient Safety Goal to Maintain and
communicate accurate patient medication information.8
The Best Possible Medication History (BPMH) constitutes the cornerstone for medication
reconciliation. The BPMH is more comprehensive than a routine primary medication history, as
it involves (1) a systematic process for interviewing the patient/family; and (2) a review of at
least one other reliable source of information (e.g., review of a central medication database,
inspection of medication vials, or contact with the community pharmacy) to obtain and verify
patient medications (prescribed and non-prescribed).9
271
Some may argue that ambulatory patients face greater risks from medication problems than
do patients in a protected hospital setting. Studies of ambulatory reconciliation have begun to
appear.10,11 However, most studies of medication reconciliation still focus on the hospital setting,
which remains the focus of this review.
Setting
Study Design
Transition
Targeted
Coffey,
200918
Pediatric ward in
academic medical
center in Canada
Admission to
hospital
Cornish,
20052
Medical ward in
academic medical
center in Canada
Prospective
post-intervention
study (272
patients)
Prospective
post-intervention
study (151
patients)
Admission to
hospital
272
Additional
Interventions
Beyond Medication
Reconciliation
None
Outcome
None
Clinically significant
unintentional
discrepancies
Clinically significant
unintentional
discrepancies
Table 1, Chapter 25. Studies of medication reconciliation that include assessment of clinically
significant unintended discrepancies and emergency department visits and hospitalizations within
30 days of discharge (continued)
Study
Setting
Study Design
Transition
Targeted
Gleason,
200419
Surgical and
medical wards in
U.S. academic
medical center
Admission to
hospital
Gleason,
201020
Medical ward in
U.S. academic
medical center
Kripalani,
201221
Medical and
cardiology wards
in two U.S.
academic medical
centers
Medical and
cardiology wards
in two U.S.
academic medical
centers
Post-intervention
study (204
patients, 12 adult
medical-surgical
units)
Prospective
post-intervention
study (651
patients)
RCT (428
patients)
Kripalani,
201221
Lee, 20105
Inpatient wards
and critical care
units in two
academic medical
centers in Canada
Pippins,
200822
Medical wards in
two U.S. academic
medical centers
Stone,
201023
Pediatric ward in
U.S. academic
medical center
Vira,
200624
Wong,
20084
Medical ward in
academic medical
center in Canada
RCT (423
patients)
Prospective
post-intervention
study (129
patients, 10
patient care
units)
Prospective
post-intervention
study (180
patients, 7
medical teams)
Prospective
post-intervention
study (23
patients on 2
medical teams)
Retrospective
post-intervention
study (60
patients)
Prospective
post-intervention
study (150
patients)
Additional
Interventions
Beyond Medication
Reconciliation
None
Outcome
Clinically significant
unintentional
discrepancies
Admission to
hospital
None
Clinically significant
unintentional
discrepancies
At time of
enrollment in
study, discharge
home, and inhospital transfer
At time of
enrollment in
study, discharge
home, and inhospital transfer
Discharge counseling
Clinically significant
unintentional
discrepancies
Pharmacist
intervention including
in-patient pharmacist
counseling, lowliteracy adherence
aids, and postdischarge phone call
None
Clinically significant
unintentional
discrepancies
Discharge home
None
Clinically significant
unintentional
discrepancies
Admission to
hospital
None
Clinically significant
unintentional
discrepancies
Admission to
hospital;
discharge home
None
Clinically significant
unintentional
discrepancies
Discharge home
None
Clinically significant
unintentional
discrepancies
In-hospital
transfer
273
Clinically significant
unintentional
discrepancies
Table 1, Chapter 25. Studies of medication reconciliation that include assessment of clinically
significant unintended discrepancies and emergency department visits and hospitalizations within
30 days of discharge (continued)
Study
Setting
Study Design
Transition
Targeted
Schnipper,
200925
Medical wards in
two U.S. academic
medical centers
RCT (162
patients, 7
medical teams)
Admission to
hospital;
discharge home
Dedhia,
200926
Additional
Interventions
Beyond Medication
Reconciliation
None
Medical wards in
U.S. academic
medical center,
community
teaching hospital,
and urban
community
hospital
Medical ward in
U.S. academic
medical center
Prospective
before-after
study (185
patients)
Discharge home
Safe STEPS
intervention including
admission
assessment,
communication with
PCP, multidisciplinary
discharge meeting
RCT (373
patients)
Discharge home
Nurse discharge
advocates created
after-hospital care
plan, and postdischarge phone call
Koehler,
200928
Medical ward in
U.S. academic
medical center
RCT (21
Admission to
patients, 2
hospital,
hospitaldischarge home
medicine groups)
Counseling by
registered nurse
Koehler,
200928
Medical ward in
U.S. academic
medical center
RCT (20
Admission to
patients, 2
hospital,
hospitaldischarge home
medicine groups)
Kramer,
200729
Medical ward in
U.S. community
teaching hospital
Prospective
before-after
study (136
patients)
Admission to
hospital;
discharge home
Supplemental elderly
care bundle:
counseling by
pharmacist, postdischarge phone call,
discharge letter to
PCP
None
Schnipper,
200630
Medical ward in
U.S. academic
medical center
RCT (92
patients, 4
medical teams)
Discharge home
Jack,
200927
274
None
Outcome
Clinically significant
unintentional
discrepancies
Emergency
department visits
and
hospitalizations
within 30 days of
discharge
Emergency
department visits
and
hospitalizations
within 30 days of
discharge
Emergency
department visits
and
hospitalizations
within 30 days of
discharge
Emergency
department visits
and
hospitalizations
within 30 days of
discharge
Emergency
department visits
and
hospitalizations
within 30 days of
discharge
Emergency
department visits
and
hospitalizations
within 30 days of
discharge
Emergency
department visits
and
hospitalizations
within 30 days of
discharge
Table 1, Chapter 25. Studies of medication reconciliation that include assessment of clinically
significant unintended discrepancies and emergency department visits and hospitalizations within
30 days of discharge (continued)
Study
Setting
Study Design
Transition
Targeted
Additional
Outcome
Interventions
Beyond Medication
Reconciliation
Showalter, All admitted
Emergency
Retrospective
Discharge home Standardized
201131
mandatory electronic
department visits
patients through
before-after
discharge instructions and
emergency
study (17,516
department in U.S. patients)
document with
hospitalizations
academic medical
embedded
within 30 days of
computerized
discharge
center
medication
reconciliation
Walker,
Medical ward in
Prospective
Discharge home Pharmacist-facilitated Emergency
200932
U.S. academic
quasidischarge program
department visits
center
experimental
including counseling,
and
study (358
provision of
hospitalizations
patients, 2
medication
within 30 days of
medical teams
reconciliation list to
discharge
and 1 hospitalist
PCP, and postservice)
discharge phone call
Abbreviations: RCT, randomized control trial; PCP, primary care physician; Safe STEPS, Safe and Successful Transition of
Elderly Patients Study.
275
Table 2, Chapter 25. Key features of the 12 included medication reconciliation interventions
Study
Intervention
Coffey,
200918
Medication
reconciliation
by pharmacy
student
Medication
reconciliation
by pharmacist,
pharmacy
student, or
medical
student
Medication
reconciliation
by pharmacist
Cornish,
20052
Gleason,
200419
Selection for
More
Complex
Patients
No
4
medications
No
Medication
History
BPMH
performed by
pharmacy
student
BPMH
performed by
pharmacist,
pharmacy
student, or
medical
student
Structured
history
performed by
pharmacist or
PharmD
student
Structured
history
performed by
pharmacist
History
performed by
pharmacist
Electronic- or
Paper-Based
Medication
Reconciliation
Paper*
Institutional
Informatics
Functionality
CPOE*
Medication
Reconciliation
Became Order
Process
No*
Paper*
Limited*
No*
Paper*
Limited*
No*
Electronic*
EMR, CPOE*
No*
Electronic (at
one site)
EMR, CPOE,
Preadmission
Medication List
Builder
(embedded at
one site)
EMR, CPOE,
Preadmission
Medication List
Builder
(embedded at
one site)
EMR, CPOE
(partial)*
Not reported
Not reported
Not reported
Gleason,
201020
Medication
reconciliation
by pharmacist
No
Kripalani,
201221
Medication
reconciliation
by physician
and nurse
No
Kripalani,
201221
Medication
reconciliation
by pharmacist
with
pharmacist
intervention
Medication
reconciliation
by pharmacist
Medication
reconciliation
by pharmacist
Medication
reconciliation
by pharmacist
No
History
performed by
pharmacist
Electronic (at
one site)
No
BPMH
performed by
pharmacist
BPMH
performed by
pharmacist
BPMH
performed by
pharmacist
Both*
Paper*
EMR*
No*
BPMH
performed by
pharmacist
BPMH
performed by
pharmacist
Paper*
Limited*
No*
Paper*
EMR, CPOE*
No*
Lee, 20105
Pippins,
200822
Stone,
201023
Vira,
200624
Wong,
20084
Medication
reconciliation
by pharmacist
Medication
reconciliation
by pharmacist
or pharmacy
resident
No
Identification
of medically
complex
conditions
based on
published
guidelines
No
No
276
Not reported
Table 2, Chapter 25. Key features of the 12 included medication reconciliation interventions
(continued)
Study
Intervention
Schnipper,
200925
Medication
reconciliation
by physician
and confirmed
by pharmacist
or nurse
Medication
reconciliation
by physician
and reviewed
by pharmacist
Medication
reconciliation
by nurse
Medication
reconciliation
by nurse and
reviewed by
pharmacist
Dedhia,
200926
Jack,
200927
Koehler,
200928
Koehler,
200928
Medication
reconciliation
by pharmacist
with
supplementary
elderly care
bundle
Kramer,
200729
Medication
reconciliation
by pharmacist
and physician
Schnipper,
200630
Medication
reconciliation
by pharmacist
Medication
reconciliation
by physician
Showalter,
201131
Selection for
More
Complex
Patients
No
Age 65
None
Age 70, 5
medications,
3 chronic
comorbid
conditions,
requirement
for assistance
with 1 ADL
Age 70, 5
medications,
3 chronic
comorbid
conditions,
requirement
for assistance
with 1 ADL
One or more
of: 7
medications,
significant
comorbid
condition,
previous
admission for
ADR, 4 drug
allergies
None
None
Medication
History
Electronic- or
Paper-Based
Medication
Reconciliation
Electronic*
Institutional
Informatics
Functionality
Paper*
EMR, CPOE*
No*
Electronic*
EMR, CPOE*
No*
Not reported
Paper*
Limited*
No*
Not reported
Paper*
Limited*
No*
Structured
history
performed by
pharmacist
Electronic*
Limited
(pharmacist
electronic
order entry)*
Yes*
History
performed by
pharmacist
Not reported
Paper*
EMR, CPOE*
No*
Electronic*
EMR, CPOE,
electronic
discharge
program
(embedded as
force function)*
Not reported
BPMH
performed by
physician and
verified by
nurse and
pharmacist
History
performed by
physician and
reviewed by
pharmacist
Not reported
277
EMR, CPOE,
linkage to
Preadmission
Medication List
Builder*
Medication
Reconciliation
Became Order
Process
Partial (not at
time of study)*
Table 2, Chapter 25. Key features of the 12 included medication reconciliation interventions
(continued)
Study
Intervention
Selection for
Medication
Electronic- or
Institutional
Medication
More
History
Paper-Based
Informatics
Reconciliation
Complex
Medication
Functionality
Became Order
Patients
Reconciliation
Process
Walker,
Medication
One or more
History
Electronic*
EMR, CPOE,
No*
200932
reconciliation
of: 5
performed by
internal
by pharmacist
medications, pharmacist
electronic
1 targeted
pharmacy
medications**,
database*
medication
requiring
monitoring, 2
changes to
regimen,
dementia or
confusion, or
inability to
manage
medications
Abbreviations: BPMH, best possible medication history; CPOE, computerized physician order entry; EMR, electronic medical
record; ADL, activity of daily living; ADR, adverse drug reaction
*Information obtained by communication with authors.
**Targeted medications included digoxin, diuretics, anticoagulants, sedatives, opioids, asthma and/or chronic obstructive
pulmonary disease medications, angiotensin converting enzyme inhibitor and/or angiotensin receptor blocker.
278
Figure 2, Chapter 25. Median and interquartile range for the number of clinically significant
unintentional discrepancies per patient for the 13 included interventions
279
20%-32%). The median rate of hospitalization or emergency department visits across seven
studies with control data was 30% (IQR, 22%-31%), a difference that was not statistically
significant.
Across three randomized controlled trials, readmissions and emergency department visits
were significantly reduced by 23% (95% CI, 5%-37%; I2 24%) (Figure 3). However, this pooled
result was driven by the statistically significant reduction achieved by an intensive intervention33
that included several interventions beyond medication reconciliation that were specifically aimed
at reducing readmissions.
Figure 3, Chapter 25. Emergency department visits and hospitalizations within 30 days of
discharge in three randomized controlled trials
Additional
interventions
No
Yes
No
Favors Intervention
Favors Control
With respect to the appropriate time period for observing an impact on post-discharge
utilization, it is particularly noteworthy that the two randomized controlled trials25,30 that
included no additional interventions beyond medication reconciliation did not reduce hospital
utilization within 30 days. However, one additional randomized controlled trial34 met all of our
inclusion criteria but was excluded because it measured hospital utilization at 12 months, rather
than 30 days, following discharge. This trial reported a statistically significant 16% reduction in
all visits to the hospital. The intervention consisted of intensive medication reconciliation in
which pharmacists identified drug related problems beyond unintended discrepancies, delivered
counseling to patients at admission and discharge, and telephoned patients 2 months after
discharge to ensure adequate home management of medications.
The lack of impact of medication reconciliation by itself on hospitalization utilization within
30 days of discharge may reflect the need to consider a longer window of observation to
demonstrate benefits from resolving unintended medication discrepancies. For instance,
inadvertent discontinuation of cholesterol lowering medications, antiplatelet or anticoagulant
agents, thyroid hormone replacement, anti-resorptive therapy for osteoporosis, and gastric acid
suppression agents all may produce adverse clinical effects requiring hospital utilization in the
long term, but not necessarily within 30 days of discharge.
280
by pharmacists, which is not the intervention implemented in routine practice and required by
accreditation bodies in the U.S. and Canada.
In two RCTs that evaluated medication reconciliation using ADEs as the outcome, one30
reported a reduction in preventable ADEs, but the other21 (a comparably rigorous RCT from the
same research group) found only a reduction in potential ADEs at one of the two sites. The
remaining included studies evaluated the outcomes that have been judgments about the potential
clinical importance of detected medication discrepancies. These judgments are far from
straightforward. First, there is the usual problem with inter-rater reliability seen in studies of
adverse events and ADEs. Second, assessing the impact of unintended medication discrepancies
involves speculation about a number of factors, including not just the potential risk to a given
patient associated with the discrepancy, but also the likelihood that the discrepancy will persist
and for how long before it is eventually detected by the patient, an outpatient physician, or some
later health care encounter.
In the studies that reported particularly serious (e.g., potentially life-threatening)
discrepancies, few events were judged to be serious. Moreover, in the widely quoted study of
post-discharge adverse events,35 even though one of the examples of post-discharge adverse
events involved a medication discrepancy, the subsequent analysis of ADEs highlighted
problems with drug monitoring as the most common cause, not problems with medication
reconciliation.36
281
Table 3, Chapter 25. Medication reconciliation in varying levels of intensity as seen in published
studies
Level
Bronze
Key Components
BPMH with admission reconciliation
Silver
Gold
Platinum
Diamond
Published Examples
Cornish et al. 20052;
Kwan et al. 20073
Schnipper et al.
200925; Wong et al.
20084
Cesta et al. 200640;
Dedhia et al. 200926;
Schnipper et al. 200925
Dedhia et al. 200926;
Murphy et al. 200941;
Nazareth et al. 200142;
Al-Rashed et al. 200243
Gillespie et al. 200934;
Jack et al 200927;
Karapinar-arkit et al.
200944; Schnipper et
al. 200630; Walker et al.
2009)32
282
not reduce hospital utilization within 30-days of discharge, but may do so when bundled with
other interventions aimed at improving transitions in care. It may by itself reduce hospital
utilization over timelines longer than 30 days.
Given limited resources, the paramount issue becomes how to target medication
reconciliation in order to direct resources most efficiently. This is especially important given that
all but three of the included interventions involved the use of pharmacists to conduct medication
reconciliation. Disappointingly, the studies that selected high-risk patients did not consistently
report higher rates of clinically significant unintentional discrepancies or show larger effects on
readmissions.
This null result could reflect the limited number of studies. But, the high risk criteria used
also have plausible limitations. For instance, elderly patients and patients with multiple chronic
conditions may take large numbers of medications. However, their medication regimens may
remain stable for some time and/or are well known to the patient or their caregivers. These risk
factors for unintended medication discrepancies do not account for such nuances. A more direct
risk factor is probably frequent or recent changes to medication regimens. Unfortunately, this
risk factor cannot be ascertained reliably without conducting a thorough medication history, not
unlike the BPMH required for medication reconciliation. A summary table is located below
(Table 4).
Table 4, Chapter 25. Summary table
Scope of the
Strength of
Problem Targeted by Evidence for
the PSP
Effectiveness
(Frequency/Severity) of the PSPs
Common/Low
Moderate
Evidence or
Potential for
Harmful
Unintended
Consequences
Low
Estimate of
Cost
Moderate
Implementation Issues
How Much do We
Know?/How Hard Is it?
Moderate/Moderate
References
1.
5.
2.
6.
3.
7.
8.
4.
283
9.
10.
11.
12.
13.
14.
15.
16.
17.
284
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
36.
37.
38.
39.
40.
41.
30.
31.
32.
33.
42.
34.
43.
44.
35.
285
45.
286
Introduction
Patients are often hospitalized after suicide attempts or because of suicidal ideation.
However, hospitalization is not fully protective and the inpatient population remains at risk.
Many risk factors are associated with inpatient suicide, but as detailed below reported rates
vary widely, and the importance of this topic derives from the fatality of the outcome in close
proximity to care, not primarily from its frequency. Suicide has been frequently associated with
certain mental health diagnoses, especially depression and schizophrenia, but the risk of suicide
is not limited to patients psychiatrically hospitalized: medical and surgical patients have
profound risk factors, including severe pain, altered mental status, and progressive or terminal
diagnoses. For all patients, these risks persist, even if patients are placed on special observation
status with nursing personnel directly monitoring them.1
Assessing and reducing the suicide risk for inpatients has become a component of national
patient safety efforts. In 1998, The Joint Commission released a Sentinel Event Alert about
inpatient suicides based on a review of 65 cases, making brief recommendations about suicide
risk assessment, policy and procedures, staff training, and modification of the hospital to reduce
environmental risks.2 Although the 1998 Alert was not specific to behavioral health units, in
2010 the Joint Commission added a Sentinel Event Alert for inpatient suicide on
medical/surgical units and in emergency departments.3 The current Joint Commission (2011)
National Hospital Safety Goals include the goal of identifying patients at risk for suicide
(NPSG.15.01.0), with three elements of performance (perform risk assessment, identify
appropriate treatment environment and safety needs, and provide patient and their family with
suicide prevention information at discharge).4 National Quality Forums Serious Reportable
Events (2011) lists suicide, suicide attempts, and self-harm that results in serious injury.5
Medicare has placed inpatient suicide on the never events list. The Centers for Medicare and
Medicaid Services (CMS) announced in August 2007 that Medicare would no longer pay for
additional costs associated with many preventable errors, including those considered Never
Events. Since then, many states and private insurers have adopted similar policies.6
The Agency for Healthcare Research and Qualitys evidence report, Making Health Care
Safer: A Critical Analysis of Patient Safety Practices (2001)7, focuses on general safety
practices that would extend to psychiatry and other areas of medical practice, and on the relative
lack of evidence for behavioral health interventions within the patient safety remit.
Consequently, the authors did not specifically address inpatient suicide. The purpose of this
narrative literature review is to identify new developments and trends starting from the date of
the AHRQ report up to the present.
This review addresses three important questions related to the safety of medical, surgical, and
psychiatric inpatients at risk for suicide.
What is the evidence that clinical, organizational, or environmental programs work to
reduce attempts or completions for hospitalized patients?
What is the state of programs in use at this time?
287
To conduct the review, we searched PubMed in October 2011 using major heading search
terms Suicide, and Hospital or Inpatient or Safety Management, for English language articles
published starting in the year 2000. We expanded the search using the PubMed related
citations feature, and Google Scholar to search for citing articles of those retained for review;
we identified additional relevant articles by reference mining. Clinical trials, large observational
studies, reviews, and reports on implementations were given priority. Systematic reviews were
scored for methodologic quality using the 11-point AMSTAR scale;8 items rated Not Applicable
were not counted towards either the score or the total.
disease. The mental status of patients was infrequently and inconsistently reported. Jumping
from a building was the leading mechanism, unlike the pattern seen in psychiatric inpatients and
in the general population. Bostwick24 in an informal review of the same area based on a case
series of 50 psychiatric consultations from general medical/surgical wards concluded that
medical and surgical patients have different risk factors, and a different profile from psychiatric
patients, typically by lacking a strong personal history of suicide attempts, psychiatric diagnoses,
and substance abuse.
Risk factors, and the difficulties of risk prediction. Suicide is relatively rare, making it
difficult to predict even in populations with multiple risk factors and high relative risk. This
conclusion, long established for outpatients and the general population, holds true for inpatients
as well. Large25 (AMSTAR score 9/11) in a systematic review and meta-analysis of 29 studies
concluded that some specific risk factors are associated with inpatient suicide, but using the
presence of multiple risk factors to identify high-risk patients produces many false positives, and
misses some who will go on to commit suicide in the hospital. They concluded that reducing
environmental risks and improving systems of clinical care are likely to have greater effects on
suicide reduction than reliance on suicide prediction methods. The difficulties of accurate
prediction for inpatients are consistent with conclusions reached by others, including Busch,26
Cassells,27 Paterson,28 Bisconer,29 and the American Psychiatric Association Practice Guideline
for the Assessment and Treatment of Patients With Suicidal Behaviors.30
Environmental risk reduction factors. The removal of physical or structural risk factors from
the hospital environment has been frequently proposed. Lieberman31 and Cardell32 both report
expert opinions of this topic, and make specific suggestions for environmental modifications.
The modifications follow from the frequency with which hanging is used in inpatient suicide by
removing both materials that could form a noose and anchor points for the noose. Most of these
recommendations target inpatient psychiatric wards. Bostwick24 notes the difficulties of applying
these same recommendations to typically open general medical wards, which are more difficult
to secure; they recommended use of nursing observation for those areas.
289
decrease was described as associated with the component involving the formal assessment of
suicide risk; unfortunately, the timing of the other components was not clearly described making
it difficult to assess their role in any reduction in suicides or attempts, and in the assignment of
causality to their intervention.
Other program experiences are described here more briefly. Temkin34 proposed a precaution
monitoring sheet to improve the consistency of documentation and communication within
treatment team, but did not report of evaluation of it. McAuliffe35 described the implementation
of a program at Trillium Health Centre, Ontario, Canada, reporting on their experiences with risk
assessment, staff surveys and focus groups, and training workshops; no outcomes data of
inpatient suicides were reported. Ellis36 reported on a program, called the Collaborative
Assessment and Management of Suicidality, underway at the Menninger Clinic in Houston. The
program began with the elaboration of suicide risk assessment into a comprehensive
collaborative framework for patient treatment and risk reduction. The framework does not appear
to be limited to inpatients. They noted the need for rigorous evaluation and planned to conduct a
randomized controlled trial of their program. Ballard37 proposed a framework for organizing the
response of a hospital to an inpatient suicide. No evaluation of this framework was reported.
290
next report.41 No outcomes measures were reported. They also noted there was no evidence that
the checklist was being used correctly. The target location was inpatient psychiatric units; they
recommended using one-to-one observation for general medical units.
Janofsky42 reported on the use of Failure Mode Effects Analysis (FMEA), a structured,
systematic, prospective methodology from systems engineering, to identify possible system
failures, and used this analysis to redesign the communication flow related to observation of
psychiatric patients. No outcome results were reported.
Wu43 examined the use of RCAs in medicine generally, and noted a very wide range of skill
in performing RCAs accurately, a lack of best practices in reporting and followup, and the
absence of peer-reviewed evidence of the effectiveness of RCAs or their cost-benefits tradeoffs.
291
usage across time, wards, and hospitals, perhaps driven by idiosyncratic differences in staff
preferences for or against the use of CSO.
Bowers et al. in their literature review15 noted that suicide rates showed a mixed association
with the presence of nursing observation (at different levels of intensity) in force at the time of
the suicide. The cautions about inference from observational studies apply here.
Because the observer cannot be simultaneously engaged in other activities, use of nursing
observation can be expensive. More details about cost and the implementation of observation
programs at one Massachusetts hospital are reported by Harding.49
Most other articles note the lack of evidence that constant observation is efficacious. Issues
such as the quality or therapeutic effect of the observer-patient relationship have not been
addressed here, but common sense suggests they might vary widely, and have therapeutic or
counter-therapeutic effects, depending on the kind of interpersonal relationship between the
observer and the patient. Cutcliffe1 noted that suicides have occurred while the patient was on
observation status.
Alternatives to constant observation were explored by Cox,50 who proposed an alternative
nurse-team framework, with greater nurse autonomy and greater engagement with the patient,
along with the use of intermittent observation. These proposals have not been formally
empirically tested.
Jayaram51 reported an informal survey of the use of 15-minute checks (observation of the
patient at least once every 15 minutes), which showed considerable variation in the use of this
practice. No outcomes data were reported.
uncertain if nursing observation raises, lowers, or has no effect on the rates of suicide and selfharm for any given level of suicide risk. The psychological effects of nursing observation on
both staff and patients are not the focus of this review; however, these might be expected to have
second-order effects, including forging risk-lowering relationships between the at-risk patient
and a staff person or, conversely, raising risk by interfering with patient privacy and autonomy,
and increasing patient confinement and alienation.
Low
Evidence or
Potential for
Harmful
Unintended
Consequences
Low
Estimate of
Cost
Moderate
293
Implementation Issues:
How Much do We
Know?/How Hard Is it?
Little/Moderate
References
1.
2.
3.
4.
5.
6.
7.
8.
294
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
295
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
48.
42.
49.
50.
43.
44.
51.
52.
53.
45.
46.
47.
296
Introduction
Stress-related gastrointestinal ulceration is a known complication of critical illness.
Disruption of mucosal barriers and gastric acid hypersecretion lead to diffuse shallow mucosal
injury and discrete ulcerations in the proximal stomach and duodenum, which in turn can lead to
gastrointestinal (GI) bleeding and perforation.1-3 The prevalence of clinically significant bleeding
in patients with documented stress ulcers varies from 0.6-15%, and mortality associated with the
complication of GI bleeding can be nearly 50%.4-7
Independent risk factors for bleeding include respiratory failure requiring mechanical
ventilation for longer than 48 hours and coagulopathy.4,8 Other associated risk factors for
mechanically ventilated patients include shock of any cause, renal failure, and burns.8
Several pharmacologic therapies have been studied for the prevention of stress-induced
gastrointestinal bleeding, including proton pump inhibitors (PPIs), histamine-2 receptor (H-2)
antagonists, sulcrafate, and enteral nutrition. Despite decades of research, significant controversy
continues to surround standardization of prophylactic therapy, particularly because of evidence
that prophylaxis is associated with pneumonia, inappropriate use, and cost. Independent of
prophylactic therapy, rates of clinically significant bleeding have actually declined, likely related
to other patient safety practices around management of sepsis and enteral nutrition.9
Multicomponent or bundled interventions are becoming increasingly common as a method
of improving outcomes by preventing complications in ICU patients. Examples of these
approaches include the Surviving Sepsis Campaign,10 which includes stress ulcer prophylaxis
and deep venous thrombosis prophylaxis along with evidence-based clinical strategies to
improve sepsis outcomes, and the Institute for Healthcare Improvements Ventilator Bundle,
one of the key components of the 100,000 Lives campaign11 and the Keystone ICU project.12
Given the wide implementation of these bundles, the key issues around stress ulcer prophylaxis
involve not only standardization of therapy with the most efficacious agents but also
appropriateness of therapy based on risk assessment, and discontinuation of therapy when
appropriate.
The 2001 Making Health Care Safer report reviewed evidence on the epidemiology of stressrelated GI bleeding, and included an evaluation of two meta-analyses and one large randomized
controlled trial (RCT) on the effectiveness of pharmacologic therapies, including H2-antagonists
and sucralfate.13-15 Both H2-antagonists and sucralfate were found to be effective at preventing
clinically significant GI bleeding in ICU patients, but the overall magnitude of benefit was small.
The review found a relatively low incidence of clinically significant stress ulcer-related GI
bleeding and a higher cost-to-benefit ratio for low-risk patients. Concern was also raised
regarding a possible associated risk of hospital-acquired pneumonia with acid suppression.
Therefore, the review concluded that no evidence supported the institution of universal stress
ulcer prophylaxis in the ICU. The report recommended considering stress ulcer prophylaxis with
either an H2-antagonist or sucralfate for the prevention of GI bleeding in certain high risk ICU
297
patient populations, including patients with respiratory failure, coagulopathy, renal failure,
and/or burns, and considering enteral nutrition for other populations.
reduced the incidence of clinically significant GI bleeding, but only in patients not receiving
enteral nutrition. In patients receiving enteral nutrition, H2-antagonists did not affect the risk of
GI bleeding; however, this finding is based on only three trials enrolling a total of 262 patients.
The possibility that enteral nutrition may have a protective effect on patients baseline risk of
stress ulceration implies that routine acid suppressive therapy may not be necessary even in
patients with traditional risk factors. This finding, while exploratory, is certainly worthy of
further study.
These systematic reviews suggest that acid suppressive therapy, while effective in preventing
stress-related mucosal bleeding, is also associated with significant risks, including pneumonia.
PPIs, though widely used, do not appear to be superior to H2-antagonists in preventing clinically
significant GI bleeding.
299
improve safety of mechanically ventilated patients, including stress ulcer prophylaxis.32 This
project was remarkably successful at preventing hospital-acquired infections and improving
other safety outcomes in the ICU, and also successfully increased stress ulcer prophylaxis rates.
Another successful approach to increasing prophylaxis was described by Krimsky et al, who
implemented a similar bundle approach incorporating several ICU prophylactic measures,
including stress ulcer prophylaxis. The implementation method emphasized team
communication, used prompts to providers to address the evidence-based measures on a daily
basis, and used a data wall to provide real-time feedback.33 This approach resulted in nearly
100% adherence to bundle use.
Evidence on efforts to control inappropriate prophylaxis use is limited. Coursol and Sanzari
described the implementation of an ICU algorithm with specific indications according to
guidelines on appropriate use, length of therapy, and cost.34 The algorithm was associated with a
reduction in inappropriate use of prophylaxis and costs.
Evidence on the cost of prophylaxis as it relates to implementation is also lacking. The cost
of acid suppressive therapy varies, with H2-receptor antagonists being less expensive than PPIs.
Decreasing inappropriate PPI use could likely be cost-saving for hospitals.
Moderate
Evidence or
Potential for
Harmful
Unintended
Consequences
Moderate
(pneumonia)
Estimate of
Cost
Moderate
Implementation Issues
How Much do We
Know?/How Hard Is it?
Little/Not difficult
References
1.
2.
300
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
301
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
302
28.
29.
Herzig SJ, Howell MD, Ngo LH, et al. Acidsuppressive medication use and the risk for
hospital-acquired pneumonia. Jama. 2009
May 27;301(20):2120-8. PMID: 19470989.
30.
31.
32.
33.
34.
Introduction
Deep venous thrombosis (DVT) refers to occlusion within the venous system, most
commonly of the lower extremities, which can lead to pulmonary embolism (PE), or embolism to
the pulmonary vasculature. Venous thromboembolism (VTE), comprising PE and DVT, is
estimated to account for 5 to 10 percent of all deaths among hospitalized patients,1,2 and also is
associated with significant morbidities. In 2008, the United States Surgeon General issued a Call
to Action to Prevent DVT and PE. The report brings to light the huge numbers of patients
afflicted by DVT (350,000-600,000) and killed by PE (>100,000) every year in the United
States.3 Even though high quality evidence exists for safe and effective strategies to reduce the
risk of VTE, studies continue to show that many hospitalized patients are not given riskappropriate VTE prophylaxis. One recent study across 32 countries found that only 59 percent of
at-risk surgical and 40 percent of at-risk medical patients received guideline-recommended VTE
prophylaxis4 and a United States registry study found that only 42 percent of patients diagnosed
with DVT during a hospitalization had received prophylaxis.5
The Agency for Healthcare Research and Quality (AHRQ) has indicated that delivery of
appropriate VTE prophylaxis is an essential patient safety practice and one that can prevent inhospital death.6 As of 2011, the National Quality Forum (NQF) has 10 VTE-related standards
and endorsed outcomes measures.7 Evidence-based best practice prophylaxis varies by primary
service (e.g. medicine, surgery, trauma, orthopedics) and patient risk factors. Risk of VTE
among hospitalized patients varies based on several risk factors including medical condition,
type of surgery, trauma, cancer, age, immobility, hypercoagulable state, and previous history of
VTE. Most hospitalized patients have one or more VTE risk factors, and well-developed
guidelines are available that specify which types of patients should receive prophylaxis
measures, and which specific measures are most appropriate.1
The original report, Making Health Care Safer, reviewed the effectiveness, safety, costeffectiveness, and indications for VTE prophylaxis. This review concluded that whereas VTE
prophylaxis shows clear benefits for a number of conditions and minimal concerns regarding
adverse events, the practice remains underused. A small number of interventions aimed at
improving use of prophylaxis were reviewed. The current review provides an update on the most
effective VTE prophylaxis regimens as well as on interventions aimed at improving adherence to
guidelines on the use of these preventive strategies. A MEDLINE search was conducted from
2001 to 2011 to identify studies that assessed the effectiveness and safety of VTE preventive
measures as well as those aimed at improving their use.
303
enoxaparin, dalteparin, and fondaparinux; warfarin; and aspirin, along with newer classes of antithrombotic agents. Mechanical prophylaxis includes anti-embolic stockings and intermittent
pneumatic compression devices. Because the underlying approach of all prophylaxis medications
is to decrease clotting, they may increase the risk of bleeding. The balance between bleeding and
clotting must be considered in every patient, and the benefits and harms must be weighed before
administering these drugs. For this reason, patient risk stratification is paramount to ensure that
only at-risk patients are treated and that they receive the right prophylaxis. Ongoing clinical
research and evidence-based medicine reviews suggest that blanket approaches that give the
same medication to all patients without risk stratification may not be beneficial and may even
cause more harm than benefit.9-11
304
305
306
practice, not all VTE events can be prevented;26,27 it has been estimated that best practice
prophylaxis may reduce incidence of DVT by up to 70 percent.1 Another potential limitation to
the use of DVT/PE rates alone to measure quality is the significant issue of surveillance bias
because many DVTs are clinically silent and therefore go undetected without routine screening.28
For example, in the field of trauma surgery, clinical ambiguity persists regarding the clinical and
cost effectiveness of the screening of high-risk asymptomatic trauma patients for DVT with
duplex ultrasound.28 As a result, certain providers and hospitals report higher DVT rates due
entirely to higher rates of diagnostic testing- a classic example of surveillance bias.29-31
Because of these issuesand variation in patient riskunadjusted VTE rates are likely not
appropriate for public reporting. A better definition of preventable harm may be obtained by
combining an outcome and process measure rather than relying on an outcome alone. For
example, it has been suggested that only VTE events occurring in patients who did not receive
adequate prophylaxis should be labeled a preventable VTE.28 This approach and specific
definition has been incorporated as one of the six Meaningful Use Quality Measures related to
VTE,22 although this measure has not yet been evaluated for its impact on VTE prophylaxis
compliance.
High
Evidence or
Potential for
Harmful
Unintended
Consequences
Moderate
(bleeding)
Estimate of
Cost
Low
307
Implementation Issues:
How Much do We
Know?/How Hard Is it?
Little/Moderate
References
1.
2.
3.
4.
5.
6.
7.
8.
308
9.
10.
11.
12.
Gmez-Outes A, Terleira-Fernndez A,
Surez-Gea M, Vargas-Castrilln E.
Dabigatran, rivaroxaban or apixaban versus
enoxaparin for thromboprophylaxis after
total hip or knee replacement: Systematic
review, meta-analysis and indirect treatment
comparisons. BMJ 2012;344:e3675.
13.
14.
15.
16.
17.
25.
18.
26.
19.
27.
28.
29.
30.
31.
20.
21.
Haut ER, Streiff MB. Mandatory ComputerBased Decision Support System [computer
program]. North American Thrombosis
Forum (NATF); 2010.
Haut ER, Lau BD, Kraenzlin FS, Hobson
DB, Kraus PS, Carolan HT, Haider AH,
Holzmueller CG, Efron DT, Pronovost PJ, et
al. Improved prophylaxis and decreased
preventable harm with a mandatory
computerized clinical decision support tool
for venous thromboembolism (VTE)
prophylaxis in trauma patients. Archives of
Surgery in press.
22.
23.
24.
309
Introduction
Fluoroscopically- and computed tomography (CT)-guided diagnostic and interventional
procedures are being performed with increasing frequency worldwide. From 1980 to 2007,
annual performance of CT in the U.S. increased from 3 million1 to 80 million.2 With this rapid
increase in the use of imaging techniques, there has been a concurrent increase in patient
exposure to ionizing radiation.1
Effects associated with radiation can be categorized as either deterministic or stochastic.
Deterministic effects manifest themselves in a relatively short time after a high-intensity
exposure to radiation (e.g., 1 or more sieverts).3 In 1994, approximately 50 radiation-induced
burns were reported to the U.S. Food and Drug Administration (FDA). In 2000, a review of 73
reports of radiation-induced skin injuries4 identified fluoroscopically-guided procedures as the
cause of 38 severe skin injuries (e.g., chronic ulceration); 18 requiring skin grafts.5,6 Radiationinduced burns have also been reported after extended radiation exposure during CT brain
perfusion scans.7
Stochastic effects are increased risks of various conditions (e.g., cancer, heart disease) that
manifest themselves over a longer time period. Recent estimates indicate that CT scans
performed in the U.S. in 2007 will be related to approximately 29,000 future cancers; killing
nearly 15,000. Almost one half of the projected cancers will be due to scans of the abdomen and
pelvis.8 Experts indicate that more than 400 patients (across eight U.S. hospitals) who recently
received higher-than-expected radiation doses while undergoing CT brain perfusion scans may
now face long-term risks of cancer and brain damage.9
310
311
312
by phone or email to discuss the request. Implementing this patient safety practice has
encouraged the ED at this institution to implement additional radiation reduction measures for
other diagnoses (e.g., renal colic).
Use of prospective gating was a core element of radiation reduction measures in two
studies.30,31 Other measures used in one study30 included limiting scan length, minimizing tube
current or voltage according to body physique, use of small bowtie filters, and tube current
modulation during cardiac cycle. A collaborative effort amongst three sites was involved in
protocol development in one study.31 LaBounty states that two measures (prospective ECG
gating and 100-kV tube voltage imaging) were only used in 92% and 67% of patients, suggesting
that additional radiation reductions would have been possible if protocol compliance had been
higher. Lack of awareness, uncertainty regarding appropriate implementation, and concern about
the quality of studies that assessed reduction techniques were also described as barriers to
implementing multiple radiation reduction techniques in everyday practice. The generalizability
of implementing a similar initiative at less experienced sites may be limited because the patient
population involved in this study underwent cardiac computed tomography angiography (CCTA)
at three large-volume, experienced centers.
One large urban medical center benefitted from adding a decision support (DS) system to its
existing radiology order entry (ROE) system.32 Before DS integration, referring physicians
completed a ROE form to initiate a CT exam. After introduction of the DS component, a second
form was populated providing physician feedback on appropriateness of the exam (1-9
appropriateness score), alternate procedures to consider, and options to proceed or cancel the
request. Appropriateness scores, based on ACR Appropriateness Criteria scores and locally
developed indication and procedure pairs, are continuously reviewed and modified.
Locally derived evidence-based imaging guidelines were the basis for a DS tool at another
multispecialty integrated health care network.33 Rapid implementation of the DS tool was
attributed to pressure from local commercial payers and an institutional culture already vested in
evidence-based medicine (including evidence-based imaging protocols) and lean health care
management methodology.34 An audit of imaging requests to determine outcome for orders
initially denied by the DS system was described as a potential screening method to determine
whether providers had gamed (developed ways to order inappropriate studies) the system.
Lastly, Raff et al. described implementation efforts at 15 hospital imaging centers
participating in the Advanced Cardiovascular Imaging Consortium in Michigan. Hospital
imaging centers were located in both small community hospitals and large academic medical
centers (1,000+ beds). Best practice recommendations were developed based on data (including
radiation dose and image quality metrics) from CCTA scanning of 620 patients acquired during a
13-month control period. During an 8-month intervention period, recommendations created by a
team consisting of a physician program director, a consulting radiologic technician, and a
licensed medical physicist were distributed to participating sites at scheduled consortium
meetings, during on-site visits by coordinating center staff and through personal communication.
This Best-Practice Model for Scan Acquisition includes directives on topics such as medical
history, administration of beta blockers and nitroglycerin, and protocol parameters (e.g., field of
view, tube current modulation). Scanner manufacturers were involved in training on scannerspecific techniques. Responsibility for on-site implementation was designated to a physician and
radiology technologist. Raff et al. reported that the greatest reduction in dose occurred at lowvolume sites (30 scans per month).
313
314
in reductions in mean radiation exposure by 70% (Strategy 1; 95% confidence interval [CI] 60%
to 78%) and 38% (Strategy 2: 95% CI 29% to 48%). The primary endpoint was the frequency of
complete inclusion of the acute pathologic region (detected on the complete CT scan) within the
scope of the two hypothetical z axis-restricted CT scans. Current standard practice indicates a CT
scan of the entire abdomen and pelvis. Abdominal pathology was completely included in limited
CTs in 17% to 36% of patients; completely or partially included in 84% to 92% of patients.
However, in 12 cases (eight from Strategy 1), the pathology detected at CT lay completely
outside the marked region of tenderness (see harms below).
Two studies examined use of algorithms to reduce radiation exposure from CT. Loftus et al.
examined use of an imaging algorithm to reduce radiation exposure in 60 patients with
aneurysmal subarachnoid hemorrhage (435 CT examinations).1 This imaging algorithm describes
the most appropriate time points at which to detect vasospasm with CTA and CT perfusion
imaging. Post-implementation results included a 12.1% decrease in cumulative radiation
exposure (p>0.05), a 25.6% reduction in mean number of CT examinations performed per
patient, and a 32.1% decrease in the number of CT perfusion examinations per patient. Stein et
al.29 implemented an imaging algorithm in which stable ED patients with a clinical suspicion of
pulmonary embolism underwent chest radiography followed by V/Q scanning (negative chest
radiograph) or CTPA (positive). Data indicates that when comparing CTPA to V/Q scanning, the
total effective dose from CTPA is almost five times greater; the dose to the female breast 20 to
40 times greater.35,36 After one year, results included a statistically significant 20% reduction in
mean effective dose (8.0 mSv to 6.4 mSv; p<0.0001); a 32% reduction in mean effective dose in
women younger than 40 years. From 2006 to 2007, no significant difference in the false-negative
rate (range, 0.8-1.2%) between CTPA and V/Q scanning occurred and CTPA usage in ED
patients with suspected PE declined from 64.6% to 39.4%.
Two studies evaluated the effectiveness of clinical DS systems to reduce unnecessary CT
imaging.32,33 Sistrom et al.32 reported results after integrating a new DS component to a
computerized ROE system at a large, integrated, multispecialty group practice. Significant
decreases were demonstrated in absolute growth (311 vs. 37; p<0.001) and growth rate (3% vs.
0.25%; p<0.001) of CT exams per quarter from 2004 to 2007. The authors reported that the
number of CT exams was essentially flat despite an increase in outpatient visits by almost
70,000 over the same period. One retrospective cohort study evaluated use of an evidence-based
clinical DS tool to reduce outpatient imaging use rates for several high-volume imaging
procedures.33 Two years after implementation, Blackmore et al. reported data from a single
commercial payer indicating a clinically and statistically significant decrease (-26%) in use of
sinus CT for suspected sinusitis (relative risk [RR], 0.73; 95% CI 0.65 to 0.82; p<0.001).
Secondary analysis indicated that use of the DS tool was also associated with a decrease in
overall volume of sinus CT studies, regardless of diagnosis.
The three remaining studies implemented several radiation reduction measures in cardiac
computed tomography angiography (CCTA);30,31,37 prospective gating was implemented in two
studies. Prospective gating was a core element in initiatives implemented in one study (n = 623)
by Choi et al.30 Results included a statistically significant difference in radiation dose between
the prospective (n = 384) and retrospective (n = 239) gating groups (2.0 vs. 9.6 mSv; p<0.0001).
In addition, median radiation doses per month decreased from 6.2 to 2.1 mSv over time due to
increased usage of prospective gating. One multisite study (n = 449) examined effectiveness of a
standardized BMI [body mass index]-based and heart rate-based protocol. Post-implementation,
LaBounty (2010) reported median radiation dose had decreased from 2.6 mSv (interquartile
315
range 2.0 to 4.2) to 1.3 mSv (interquartile range 0.8 to 1.9) due to use of the standardized
protocol (p<0.001). Statistically significant reductions (p<0.001 level) were also reported for
prospective (versus retrospective) electrocardiographic gating (-82%), reducing tube voltage
(-41% for 100 vs. 120 kV [kilovolts]), lowering tube current (-25% per -100 mA), and reducing
overall scan length (-6% per -1 cm). LaBounty also reported no differences between groups in
the frequency of interpretable studies on a per patient (96.4% vs. 95.5%; p = 0.66) or per artery
(99.1% vs. 98.5%; p = 0.26) basis.31
Lastly, Raff et al. (2009)38 reported improvements from dose reduction strategies from a
consortium of 15 imaging centers (n = 4,862). Radiation reduction measures involved
implementation of a best-practice model including techniques to minimize scan range, heart rate
reduction, electrocardiographic-gated tube current modulation, and reducing tube voltage in
suitable patients. Compared with the control period, patients estimated median radiation dose in
the follow-up period was reduced by 53.3% (dose-length product decreased from 1493 mGy x
cm [IQR 855-1823 mGy x cm] to 697 mGy x cm [IQR, 407-1163 mGy x cm]; p<0.001. A
statistically significant reduction in effective dose was also reported (21 mSv (IQR, 12-26 mSv)
to 10 mSv (IQR, 6-16 mSv) (P<0.001)). No significant changes were reported in median image
quality assessment (control vs. follow-up period) or proportion of diagnostic-quality scans.
Harms. Harms from a PSP were reported in one study when implementation of CT-reduced
strategies resulted in erroneous findings of no pathology in 12 patients.28 Three patients required
emergency treatment resulting in a laparoscopic appendectomy, stent placement, and admittance
for pyelonephritis.
316
Moderate
Evidence or
Potential for
Harmful
Unintended
Consequences
Negligible
Estimate of
Cost
Low
Implementation Issues:
How Much do We
Know?/How Hard Is it?
Moderate/Not difficult
References
1.
2.
Boone JM, Strauss KJ, Cody DD, et al. Sizespecific dose estimates (SSDE) in pediatric
and adult body CT examinations. College
Park (MD): American Association of
Physicists in Medicine (AAPM); 2011. 28 p.
(AAPM report; no. 204).
www.aapm.org/pubs/reports/RPT_204.pdf.
3.
4.
5.
6.
317
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
318
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
319
37.
38.
Introduction
Numerous studies have documented that the care patients receive at the end of life is often
not consistent with their preferences. In addition, communication about end-of-life issues is
suboptimal, and advance directive completion and documentation of health care proxies and
preferences for life-sustaining treatment often does not occur. For example, a study of advanced
cancer patients who died in one hospital and cancer center1 found that only 19 percent of patients
had documentation of an advance directive or a surrogate decisionmaker in the medical record.
Furthermore, only one of the 17 patients who received mechanical ventilation had documentation
of preferences regarding mechanical ventilation or documentation of why this information was
unavailable. Other research has found that a majority of physicians whose patients had advance
directives were not aware of them, that having an advance directive did not increase medical
record documentation of patient preferences, and that advance directives were often not used in
medical care.2
These gaps in quality can have several consequences. They can lead to patients receiving
care that is not consistent with their preferences: lack of appropriate documentation or other
miscommunication about the appropriate application of Do Not Resuscitate (DNR) orders or a
patients desire to not receive aggressive care at the end of life can lead to errors of providing
invasive treatments, such as intubation and resuscitation, that are inconsistent with patient
preferences and can lead to significant patient and family suffering. Poor documentation or
communication about these preferences can also lead to confusion among staff,
miscommunication with families, and errors in code situations. Significant harm may result if a
patients health care proxy is not documented or not followed, or if end-of-life decisions are
made with a surrogate who was not the patients surrogate of choice.
The original Making Health Care Safer report reviewed interventions aimed at increasing
individuals communication of their preferences for end-of-life care, through completion of
either advance directives or health care powers of attorney. This review found that although
policies to facilitate and increase rates of completion of such instruments completed are
widespread, evidence was lacking that these instruments actually improve compliance with
patients end-of-life care wishes. Based on a recent systematic review that we conducted on the
impact of quality improvement interventions on end-of-life care, this chapter updates the review
of interventions aimed at improving completion of advance directives, as well as interventions to
increase general communication about preferences and care at the end of life, and examines
evidence that communication or advance directives increase the likelihood that patients end-oflife care preferences are followed.3
320
generally, primarily in the intensive care unit setting. These studies included quality
improvement interventions to increase the frequency and/or structure of family meetings to
address these issues, and the use of palliative care and ethics consultants. Fifteen studies (6
RCTs, 9 non-RCTs) evaluated health care utilization, such as intensive care unit length of stay,
as an outcome. We found moderate strength of evidence to support the impact of interventions
on this outcome: 73% of studies found a statistically significant improvement in the intervention
compared with the control group.
Observational studies have shown that patients with advance directives or related documents
are more likely to receive care consistent with their preferences. A retrospective evaluation of the
care provided for residents with POLST orders found that care provided was consistent with
residents preferences 98% of the time for resuscitation and 94% overall.13 Another retrospective
analysis of survey data found that patients who died with advance directives received care
strongly aligned with their preferences (97% of those who requested comfort care received it).
Subjects with living wills or who had a health care proxy were statistically significantly less
likely to receive aggressive care than those without advance directives.14 Patients who had
discussions about prognosis and end-of-life care with their providers also had care at the end of
life that was less invasive and more consistent with their preferences.
Moderate
Evidence or
Potential for
Harmful
Unintended
Consequences
Low
Estimate of
Cost
Low
323
Implementation Issues:
How Much do We
Know?/How Hard Is it?
Moderate/Moderate
References
1.
8.
2.
9.
3.
10.
11.
12.
13.
14.
4.
5.
6.
7.
324
1. patient risk factors (e.g., increased body mass index, presence of comorbidity)
2. surgical skills (e.g., technical skills), and
3. operation profile (or system factors).
The operation profile includes a range of HFE-related system factors, such as operative
environment, team performance and communication, and decisionmaking processes. System
characteristics are factors that, in addition to patient characteristics and the skills of the surgery
team, can contribute to complications and adverse events7. A range of system factors can
influence the safety of surgery and can be addressed by using concepts, models, theories and
methods from HFE. Vincents approach can be extended to patient safety incidents in other care
settings besides surgery.8 Patient characteristics and clinician skills and knowledge are important
for patient safety; but poor system design can also contribute to patient safety incidents. HFE
helps to identify system design deficiencies and hazards that affect patient safety and provides
the concepts and methods to improve system design and, therefore, patient safety.
326
be considered well designed from the HFE perspective. See Table 1 for a summary of the key
characteristics of HFE.
Table 1, Chapter 31. Key characteristics of HFE and its application to patient safety
Definition of HFE by the
International Ergonomics
Association (www.iea.cc)
Name of the discipline
Range of HFE issues
Goal of HFE
Objectives of HFE
Ergonomics (or human factors) is the scientific discipline concerned with the
understanding of the interactions among humans and other elements of a system,
and the profession that applies theoretical principles, data and methods to design in
order to optimize human well-being and overall system performance.
Human Factors and Ergonomics are two different names for the same discipline.
Physical, cognitive and organizational (macroergonomic or sociotechnical) issues of
HFE are all relevant to patient safety.
The goal of HFE is to fit the system to the people instead of fitting people to the
system.
The objective of HFE-based system design is to improve both well-being and
system performance. Patient safety is one component of system performance.
327
result from decisions made by system designers and organizational decision-makers that lead to
unsafe conditions. Active failures are errors made by the operators of the system.
Vincent and colleagues8 have adapted Reasons Swiss Cheese model to patient safety. They
describe management decisions and latent failures that can influence error and create conditions
that produce safety violations. In turn, these conditions create problems for care delivery and
may lead to unsafe acts (i.e., errors and violations), which may then produce an incident if the
defenses and barriers are not appropriate. Reasons Swiss Cheese model and its patient safety
version by Vincent and colleagues include both errors and violations as active failures. Recent
HFE research has broadened the focus on human error and developed knowledge about the
contribution of violations to patient safety.31,32
Bogner33 has proposed another HFE model of human error and patient safety: the Artichoke
model defines layers of system factors that influence provider-patient interactions, such as legalregulatory-reimbursement, national culture, organization, physical environment, social
environment, and ambient environment. The frameworks proposed by Vincent and colleagues8
and Bogner33 can be used by health care organizations to investigate specific patient safety
incidents.
System Resilience
Recently, HFE research in patient safety has focused on system resilience.46 Resilience has
been defined as the ability of systems to anticipate and adapt to the potential for surprise and
failure.47 Because not all errors may be prevented, HFE researchers have developed models to
understand how errors can be detected, corrected, mitigated, and dealt with by operators.48,49 The
WHO model of patient safety incorporates the concepts of error detection and mitigation.50,51
Strategies for error detection and recovery have been explored among nurses,52 in particular
critical care nurses,53,54 and among pharmacists.49,55 This line of HFE research can produce
information about mechanisms for achieving resilience, such as cross-checking.56,57 Resilience
engineering builds on and extends the work done by High Reliability Organization (HRO)
researchers. A key characteristic of HROs is mindfulness, i.e., the ability to prepare for the
unexpected and to be vigilant about hazards, and one aspect of mindfulness is organizational
commitment to resilience.58
328
329
62
Figure 1, Chapter 31. SEIPS model of work system and patient safety
Reproduced from Work system design for patient safety: the SEIPS model. Carayon P, Schoofs Hundt A, Karsh BT, et al.
Quality & Safety in Health Care. 15(Supplement I):i50-i58. 2006 with permission from BMJ Publishing Group Ltd.
With recent emphasis on the role of health IT in patient safety, sociotechnical systems
approaches have been proposed, for instance, by the IOM report on Health IT and Patient
Safety.4 The work system of the SEIPS framework is a representation of a sociotechnical system
(the technology is part of a larger system and interacts with various system elements). The
sociotechnical system model proposed by the IOM4 includes all elements of the work system
model, except for the physical environment.
health IT staff, as well as leaders of health care organizations, policymakers, and vendors and
manufacturers of medical devices and health IT applications.
Health care organizations may hire HFE engineers in order to accelerate adoption and
dissemination of HFE. Integration of HFE engineers in health care organizations may enhance
the impact of HFE in a wide range of relevant departments and functions of health care
organizations,67,68 such as patient safety, risk management, quality improvement, employee
health, and health IT implementation and optimization.
2.
3.
4.
HFE Mechanisms
A work system that is not designed according to
HFE design principles can create opportunities for
errors and hazards (see Table 3 for examples of
design principles).
Performance obstacles that exist in the work system
can hinder clinicians ability to perform their work
and deliver safe care.
331
of the work system should minimize perception time, decision time, manipulation time, and the
need for excessive physical exertion, and optimize opportunities for physical movement.69,82,83
From an organizational HFE viewpoint, work systems should be designed so that tasks are
reasonably demanding physically and cognitively. Workers should have opportunities to learn,
adaptive levels of control over their work system, and access to social and instrumental support
(e.g., support from co-workers in case of emergency) within the work environment.84,85 Table 3
provides some examples of HFE design principles; additional information on HFE design for
specific work system elements can be found in the Handbook of Human Factors and
Ergonomics.86
Table 3, Chapter 31. Examples of HFE design principles
Focus of HFE
Physical HFE
Cognitive HFE
Organizational HFE
Given the systems focus of HFE, it is important not only that each component of the system
be designed appropriately, but also that system components be aligned75 and that system
interactions be optimized.64 For example, when a new BCMA system is introduced, it is
important to ensure that the technology is designed according to HFE principles (e.g., usability
heuristics). However, it is also important that the technology fits with the rest of the work
system. If there is not sufficient space in which to use the BCMA (interaction between the
technology and the physical environment) or if users are not provided with adequate training
(interaction between the technology and the organization), then BCMA may contribute to
diminished rather than improved clinician performance and patient safety.
The goal of HFE-informed design is work system that supports the work of individual and
teams.75,81 This is the essence of the user-centered design approach.87
332
refine and develop HFE implementation principles and methods for facilitating user participation
in designing work systems for patient safety.
334
of the sociotechnical system. The HFE framework was used to improve the design and
implementation of CPOE.100
335
modeling of the work systems. The HFE framework of Beuscart-Zphir and colleagues is now
routinely integrated into the IT project management of the Centre Hospitalier Universitaire of
Lille, France.
336
administration process conducted to assess the potential HFE and safety issues of a new IV
pump.110
337
338
ED
HFE Issues
Impact of fatigue and sleep deprivation on psychomotor performance and mood of
113
anesthesiology residents
114-116
Workload, production pressure and burnout of anesthesiologists
Poor display and control design of medical devices: auditory and visual alarms affect
117,118
vigilance and situation awareness of anesthesiologists
Working in a multidisciplinary team: anesthesiologists working with a new surgeon or
nurse may need extra effort to communicate effectively, in particular during stressful
113
conditions
Limited availability of information: patient history and other information are often not
117
easily accessible by ED clinicians who had no prior contact with patient
119,120
Design of ED physical environment: overcrowding, noise
Usability and workflow issues of ED status boards
Home Care
ICUs
x
x
113,119
339
x
x
x
x
x
x
x
x
x
x
x
Design and implementation of guidelines and best practices, e.g., for infection
114
control
37,115,116
Workload, stress and burnout of ICU physicians and nurses
Information flow and decisionmaking in handoffs of patients: across units, across
113
services; patients discharged; shift changes
25,117
Design of alerts/alarms in medical devices and health IT
118
Design and implementation technology for remote monitoring of ICU patients
T/T
119-121
x
x
x
x
x
x
x
x
x
x
Pediatrics
Primary
Care
Surgery
HFE Issues
Poor working conditions and job stressors: understaffing, training, feeling unable to
meet resident needs, overtime, heavy workload, mostly standing and walking, risk of
124
back injuries due to moving patients
Physical environment: doors cannot accommodate a wheelchair, layout of facility
124
does not allow nursing station in a convenient place
Cognitive, communication, and speech limitations of children and their dependency
on adults result in communication challenges, and risk of delayed diagnosis or
125,126
misdiagnosis
Equipment not designed for children: CT scan with adjustable exposure for children,
125,126
cribs with adjustable height
Design problems of CPOE technology: weight-based dosing, age-dependent lab
125
normal values
Design problems of BCMA technology: barcodes with different sizes, packaging of
125
pediatric medications
Reliance on memory: missing diagnostic testing results, lack of tracking system;
127,128
physician needs to remember ordered tests
Multi-modal communication between patient and clinicians: retrieving and recording
127
information, information loss
Memory and information processing: patients with multiple problems, incomplete
127
patient charts
Workload and time pressure of clinicians: addressing several patient problems in
127
limited time
Operating room environment: clutter, noise, lighting, temperature, motion/vibration;
129
impact on surgical performance
Teamwork: miscommunication, lack of coordination, and lack of team familiarity and
129,130
stability contribute to errors during surgery
Poor design and implementation of technology affect acceptance and use: e.g.
integration of information across displays, unreliable audible alarms, shape of input
129,130
controls, and lack of proper training for surgeons
Impact of physical and mental workload on performance: task duration, strength
requirement, mental demands, and time pressure increase stress and fatigue, and
129
may affect cognitive processing
131
Design and implementation of surgery checklist
T/T
x
x
x
x
x
x
x
x
x
x
x
x
x
x
Poor safety culture: lack of a culture to take responsibility for patient safety, report
errors, learn from mistakes, and adapt individual and organizational behavior based
x
x
x
129
on lessons learned from mistakes
* NOTE: Elements of the work system include the individual (I), his/her tasks (T), tools and technologies (T/T), the physical environment (E) and the organization (O) (see Figure
1).
340
Are There Any Data About Costs of the Patient Safety Practice?
The integration of HFE in health care and patient safety requires leadership and commitment
as well as resources (i.e., money, time, effort, knowledge, expertise, skills, methods and
structures).132-135 Health care organizations that invest in HFE typically engage in one or several
of the following activities: using HFE tools and methods, increasing HFE knowledge among
their staff, and recruiting human factors engineers.67 However, there is no information available
about the costs of these different HFE approaches.
It is important to recognize the key role of HFE in the early phase of system design.136 When
HFE is used early in the design process, system issues can be identified and solved more
efficiently and effectively, and with less risk that the fix to the system design will itself create
other hazards. This implies that designers, manufacturers, and vendors of health IT applications,
medical devices and other technologies must have in-house HFE expertise.
A case study of a medical device manufacturer demonstrates the challenges of implementing
HFE.137 Patient Controlled Analgesia (PCA) pumps that were introduced in 1988 were intended
to allow patients to administer small and frequent dosages of analgesia, and reduce nurse
workload. However, the poor HFE design of the device increased the likelihood of dosage
programming errors, which in some cases led to death.137 It took 6 years between the first
reported incident of patient death related to PCA pump programming error and the hiring of a
HFE engineer by the device manufacturer in 2001. Significant efforts may be required to speed
up the dissemination of HFE to improve patient safety across the health care industry.12,67
Fostering communication and collaboration between HFE and the health sciences and
professions is critical achieving significant improvements in patient safety. Clinicians and HFE
engineers need to learn to understand each others perspectives.132 Because the HFE knowledge
domain is broad and deep (see description above of the physical, cognitive and organizational
aspects of HFE), learning HFE can be a significant investment. It is not sufficient to have
physicians or nurses who have read a book or taken a seminar on HFE; this will not make them
HFE experts.132 On the other hand, HFE engineers need to understand health care before they
can have a significant impact on patient safety.67 The training of biculturals in both medicine
(or nursing or pharmacy or other health science) and HFE can accelerate the application of HFE
to improve patient safety.138 Because biculturals have deep knowledge of and training in both
HFE and a health science, they can help to translate and disseminate HFE knowledge and
tools.138
management and implementation tools (e.g., top management commitment, human resource
issues, funding, and communications), and wider environmental factors (e.g., legal and
regulatory requirements, efforts by national and international HFE organizations, and
collaboration between health care organizations).66,140 Cultural conflicts between the HFE
systems approach and health care can also impede the adoption and dissemination of HFE in
health care organizations (e.g., physician autonomy may hinder the team collaboration and
communication stressed by HFE).139,141
The case study described by Vicente138 shows that HFE is more likely to be integrated in the
organization of a medical device manufacturer if the manufacturer (1) has leaders who support
adoption of HFE, (2) experiences a profound performance crisis related to poor HFE
performances, and (3) operates in an environment in which advocacy for HFE can be found at all
levels of the complex sociotechnical system. Further research is needed to identify the key
contextual factors that can facilitate adoption and dissemination of HFE. Specifically, studies are
recommended for developing a theoretical framework to describe and evaluate contextual
elements and generating empirical evidence on how different contextual elements can influence
the success of HFE interventions.140
342
Not assessed
systematically
but Moderateto-High
evidence for
some specific
applications
Evidence or
Potential for
Harmful
Unintended
Consequences
Negligible
Estimate of
Cost
Moderate
Implementation Issues:
How Much do We
Know?/How Hard Is it?
A lot/Moderate
References
1.
2.
3.
4.
5.
6.
7.
8.
343
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
20.
21.
22.
23.
24.
25.
26.
27.
28.
344
39.
49.
40.
50.
41.
51.
52.
53.
54.
55.
56.
57.
42.
43.
44.
45.
46.
47.
48.
345
58.
69.
59.
70.
71.
60.
72.
73.
74.
75.
76.
77.
78.
79.
80.
81.
61.
62.
63.
64.
65.
66.
67.
68.
346
82.
83.
84.
85.
86.
87.
88.
89.
90.
91.
92.
93.
347
94.
95.
96.
97.
98.
99.
100.
101.
102.
103.
104.
105.
106.
107.
108.
109.
110.
111.
348
112.
113.
114.
115.
116.
117.
118.
119.
120.
129.
121.
130.
131.
122.
132.
133.
134.
135.
136.
137.
138.
139.
123.
124.
125.
126.
127.
128.
349
140.
141.
142.
143.
350
144.
145.
146.
147.
148.
149.
351
352
model, a 6-week study with a 3-month followup was conducted. During the intervention, patients
agreed (and were encouraged) to ask all health care workers with direct contact if they had
washed/sanitized their hands. Use of soap and sanitizer per resident-day was measured before,
during, and after the intervention. While the intervention itself was not multifaceted, multiple
methods were used to encourage patients to ask the handwashing question of their providers: a
visit by a premedical student with the patient to discuss hand hygiene (HH); an education
brochure; and multiple prompts for patients to ask their providers, including videos and visual
aids. Hand washing or sanitizing increased from five HH uses per resident-day during the
intervention to 9.7 HH per resident-day during the intervention (p<0.001), 6.7 HH per residentday post-intervention (at 6 weeks) (p<0.001) and 7.0 HH per resident day at 3 months (p<0.001).
Patients asked their physicians about hand hygiene 40 percent of the time, and their nurses 95
percent of the time.
Stone and colleagues15 carried out a pre-post study among 187 acute-care hospitals in the
National Health System of the United Kingdom that included patient engagement as part of a
multifaceted HH intervention. A HH campaign was introduced to these hospitals over a period of
7 months, including alcohol hand rub near the patients bedside, regularly changed wall posters
in the wards regarding HH, and materials telling patients to ask their providers to clean their
hands. Median use of alcohol hand rub increased statistically significantly in the participating
hospitals during the intervention period, from seven to 13 ml per bed-day of alcohol hand rub
over 6 months (p<0.001); the change in soap use was not statistically significant (p=0.06). Rate
of hospital-acquired infections did not change.
Because of the small number of studies meeting our criteria, and their heterogeneity, we
could not perform evidence grading.
Table 1, Chapter 32. Patient engagement in safety: effectiveness studies
Description of PSP
Outcomes: Benefits*
Multiple Interventions or
Multifaceted
Author, Year
Interventions
13
Weingart 2004
Proving patients with
personalized medication
list to help prevent
medication errors
14
McGuckin 2004
Asking all health care
workers who had direct
contact with them, Did you
wash/sanitize your hands?
15
Stone 2007
Instructing patients to ask
health care workers to
clean their hands.
*Statistical significance, Y (Yes) or N (No)
Study Design
Sample Size
RCT
209
Close-call rate: N
Pre-post
35
Pre-post
187
353
Implementation Themes
Tools: educational tools, motivation
and reminder tools, and role
modeling
Facilitators/barriers: Social barrier of
patient to confront health care
workers; Lack of evidence of
effectiveness
families could reach from any room in the hospital. In addition, families were educated via
posters in patients rooms and flyers. In-person family awareness surveys assessed families
awareness of the family activation approach. Nurses were trained in how to explain the RRT
activation to families. They were also given reminders in the electronic medical record and given
feedback on levels of family awareness from the surveys. After implementation of family
activation, the number of RRT calls per 1,000 discharges increased from 16 to 24, though no
statistical tests were employed to assess significance.
Dean and colleagues19 described a similar early warning system that empowers patients and
families to serve as an additional line of patient-safety defense by integrating them into the RRT
system of a major childrens hospital. This study specified the conditions under which patients or
families were encouraged to call the RRT, including a noticeable medical change in a patient that
had been unaddressed; a breakdown in care or uncertainty regarding treatment; the
administration of a medication that causes an adverse effect or that the patient/family believed
had not been sufficiently explained; or the provision of treatment that the patient or family
believed was meant for another patient or contravened their doctors wishes. Apart from the
criteria under which the alert system was to be activated, this study did not detail how patients or
families were to be empowered or educated to overcome any barriers to using the system. From
September 2005 through August 2007, the early warning system responded to 42 calls from
patients and parents; the authors state that the root cause for all calls was miscommunication
between patient and provider.
Gerdik and colleagues20 studied the implementation of a family- and patient-activated rapid
response team in an acute care hospital. Pickers Eight Dimensions of Patient-Centered Care
provided the conceptual framework, including the involvement of family of friends, which
includes involving family in decision-making. However, no explicit attention to patient or
family engagement was otherwise given. Implementation of the family-/patient-activated RRT
involved written educational materials, informational signs, instructional labels for telephones,
and scripted education and training on the part of staff. After RRT activation, patient and family
satisfaction were assessed. Following implementation of the RRT, codes decreased significantly
outside the ICU, from 25/month to 17/month. Patients and families alike were found to be
satisfied with the RRT.
These studies of family and patient engagement in RRT evaluated implementation and did
not explicitly set out to evaluate effectiveness. However, in the study by Gerdik and colleagues20,
after implementation of the RRT, codes outside the ICU decreased, potentially representing a
measure of effectiveness.
355
Dean, 2008
19
Gerdik, 2010
20
Implementation Themes
Tools: direct telephone number to
reach the RRT which families could
reach from any room in the hospital,
posters, flyers
Staff/education: mock script to help
medical team discuss RRT activation
with patients/families
Facilitators/barriers: physicians
concerned that their role would be
undermined; providers
understanding of RRT as extension
of care they already provide
Tools: telephone number to activate
RRT available to patients/families 24
hours, 7 days a week
Staff/education: explanation by
admitting units nurse to patient and
family, reinforced by video and
brochure
Facilitators/barriers: leadership,
provider involvement
Tools: dedicated phone line
Staff/education: patient and family
education
Facilitators/barriers: concern that
resources would be overwhelmed;
endorsement of hospital
administration, physicians, and staff
for patients at risk for any one of the adverse events addressed. Fewer falls per patient week were
found the participating hospital wards and nursing homes, but the study was not powered to
determine the statistical significance of the decreased incidence of particular adverse events.
Table 4, Chapter 32. Falls prevention intervention studies
Author, Year
21
Krauss, 2008
22
Implementation Themes
Tools: educating all patients and
families in fall prevention, patient
pamphlets
Staff/education: Nurses,
patient care technicians, and unit
secretaries all took part in education
modules
Facilitators/barriers: staff
turnover; high patient-to-nurse ratios;
high patient turnover
or high patient volume; competing
demands on nursing staff; lack of
buy-in from staff
Tools: education, patient involvement
and feedback on process and
outcome
Staff/education: Key nurses on each
unit implemented small-scale
educational program, two case
discussions on every ward, and
distributed a CD-ROM with
educational material
Facilitators/barriers: complexity of
intervention
357
DiGiovanni, 2003
26
Implementation Themes
Tools: verbal and written instructions
to mark surgical site, marking pen
provided; assessment for compliance
on day of surgery
Barriers/facilitators: patients primary
language, cultural tendency to rely
on physicians, younger patient age,
time between enrollment and surgery
Tools: written instructions to mark
limb NOT to be operated on;
assessment for compliance on day of
surgery
minor plans, suggesting that greater engagement in the program at the site level could affect
engagement by patients receiving the intervention.32 Qualitative data indicate that patient
perceptions of a caring relationship with transition coaches foster greater patient engagement in
the program, with implications for staff training.33
Emerging
practice (few
studies
available)
Evidence or
Potential for
Harmful
Unintended
Consequences
Uncertain
Estimate of
Cost
Low
Implementation Issues:
How Much do We
Know?/How Hard Is it?
Little/Moderate
References
1.
2.
359
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
360
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
29.
24.
30.
25.
31.
32.
33.
26.
27.
28.
361
362
type of intervention with the aim of identifying the full breadth of different practices being
described as interventions to promote safety culture.
363
procedures that are shared among members of a group (care team, profession, unit, service,
department, organization, system). The difference between culture and climate is often reduced
to a difference in methodology, with studies involving surveys being categorized as measures of
climate and ethnographic studies that involve detailed, longitudinal observations being
categorized as studies of culture.
Practically, the differentiation between culture and climate is often viewed as a primarily
academic exercise. However, the dichotomy raises an important and very practical point. If you
are measuring safety culture using a survey, than you are measuring clinician and staff
perceptions of culture (i.e., safety climate) and therefore in order to change culture you have to
change perceptions of the relative priority of patient safety compared with other unit or
organizational goals, and it must be salient to providers that their actions and attitudes supporting
patient safety are actively reinforced by their peers and leaders. For example, it must be explicit
that patient safety comes first relative to other unit or organizational goals, such as efficiency,
and there must be visible recognition and positive outcomes related to engaging in safe
behaviors.
For the purposes of this review we included studies of both patient safety culture and patient
safety climate. In our discussion, we use the term patient safety culture to simplify the reporting
of results.
364
We did not identify any prior systematic reviews specifically dedicated to interventions designed
to promote a culture of safety in health care.
Of the 18 studies reviewed, 12 were pre-post studies, 2 were cluster randomized control
trials, 2 were concurrent control studies, 1 was a pre-post with concurrent control, and 1 was a
quasi stepped-wedge design. Eleven studies measured patient safety culture/climate using the
Safety Attitudes Questionnaire (SAQ),41 4 studies used the AHRQ Hospital Survey on Patient
Safety (HSOPS),31 2 studies used the Patient Safety Climate in Healthcare Organizations survey
(PSCHO),42 and 1 study utilized the Safety Climate Scale (SCS).43 Though the primary method
of measuring patient safety culture/climate uses individual-level survey responses, culture is
considered a group-level phenomenon. Thus, the majority of studies operationalized culture at
the unit level of analysis. Three studies, however, operationalized culture at the hospital level of
analysis.15,24,44 Sample sizes returned ranged from 5461 individual responses nested within 144
units within a single hospital system to 28 individuals nested within a single unit. Response rates
ranged from 35% to 100% (see Summary Table and Evidence Tables).
The majority of interventions were multi-component interventions that combined several
improvement strategies under a single overarching initiative to promote a culture of safety. For
example, Belgen et al. 201019 utilized a three component approach that included team-training,
unit-based safety teams, and strategies for engaging patients in daily goal setting. Overall, 6
studies explicitly included teamwork and communication training and tools (e.g., structured
briefings or debriefings), 4 explicitly included some form of executive walk rounds, and 4
explicitly used a multi-component approach known as the Comprehensive Unit Based Safety
(CUSP) program (see Summary Table and Evidence Tables).
In terms of effectiveness, 9 of 18 (50%) reviewed studies reported a statistically significant
impact of the intervention on the overall culture score, the safety climate score, or on at least half
of reported survey items if analyzed at the item level. Several studies reported significant
improvements in teamwork climate, but did not find similar improvements in safety climate (see
Summary Table and Evidence Tables).20 None of the studies examining multi-component or
bundled interventions examined the relative effectiveness of individual intervention components.
Only one study directly compared the effectiveness of different interventions by comparing a
simulation-based team training intervention to didactic-only team-training intervention18. Results
indicated no change in safety culture survey scores for the didactic-only and control groups. An
increase in teamwork climate was reported for the simulation-training group; however, this
finding did not remain statistically significant after Bonferroni adjustment.
Seven studies also reported the impact of interventions on other outcomes, along with patient
safety culture. In terms of patient outcomes, one study that found significant improvements in
teamwork climate20 also found a significant decrease (0.56 vs. 0.15, p < .01) in the rate of
reported errors that resulted in patient harm after implementation of a multi-faceted suite of
interventions that included both cultural (e.g., feedback on errors in the form of posters and
emails, education and training) and system-focused changes (e.g., CPOE, medication
management protocols, changes to safety reports) implemented over 2.5 years. Another study
found that the number of rapid response system activations that led to code events decreased
from 29% to 22% following an intervention to promote safety culture in which paraprofessional
care providers learned how to utilize structured communication methods to communicate
changes in patient status. However, this difference was not tested statistically. Another study that
reported a marginal increase in teamwork climate18 also found that the experimental units
weighted adverse outcome score decreased by 37% after implementation of a team training
365
program designed to promote patient safety culture, compared with a 43% increase in a control
unit (p < .05). One study also provided descriptive information on reductions in nurse turnover
from 27% to 0% for two years following implementation of the CUSP program, but no statistical
analysis was reported (see Summary Table and Evidence Tables).26
The evidence to date suggests that several practices may help to promote a culture of safety;
however, methodologic issues related to variation in the practices studied and outcomes reported,
extremely small sample sizes, and lack of cluster randomized trials constrain the evidence for
intervention effectiveness available to date. Robust evaluations are needed that assess the impact
of practices to promote patient safety culture across multiple outcomes, based on theoretically
sound evaluation models. One previous review of interventions to promote safety culture also
noted methodological constraints in primary studies45 and one previous review dedicated to
strategies to improve culture concluded that there was no rigorous evidence available in the
current literature to demonstrate their effectiveness.46 While the criteria for the previous work
finding no studies that met inclusion criteria46 were more stringent than those utilized in the
present review, this conclusion supports our finding that the robustness with which interventions
to promote culture are studied, evaluated, and reported is in need of improvement.
In terms of grading, the strength of evidence for this topic was low. Risk of bias was
generally high due to study design issues - we identified only one true cluster RCT17 Major
issues affecting risk of bias for many studies included low response rates for surveys and
incomplete reporting (not reporting all units or hospitals where interventions were conducted,
and not reporting results for all parts of the culture survey but focusing only on those that were
statistically significant). Results were inconsistent: findings in half of the studies were not
statistically significant and significant findings were difficult to compare due to variations in
measurement methods. With respect to directness, the intervention was often not specifically
designed to improve patient safety culture, and actual patient safety outcomes were infrequently
reported. Finally, with respect to precision, a number of different survey instruments were used
and were often reported differently in the articles, so no conclusions could be drawn (see
Evidence Table on Risk of Bias).
366
in safety culture scores was larger for faith-based hospitals; however statistical analyses of these
reported differences were not reported.23
This review also highlights the largely atheoretical nature of the research to date regarding
the development and implementation of interventions to promote patient safety culture. Only 3 of
the 18 reviewed studies (17%) noted any form of theoretical grounding for their improvement
and implementation strategy. Conceptual theories of patient safety culture and culture change are
foundational features of a clear logic model that guides intervention design and implementation.
Future studies should dedicate greater effort to developing and reporting the underlying
theoretical model driving intervention design and improvement efforts. Such models are a
critical element to furthering our understanding of the role context plays in moderating the
effects of various interventions to promote safety culture (Table 1).
Table 1, Chapter 33. Results of included studies on patient safety culture
Author, Year
20
Abstoss, 2011
Description of PSP
4 culture & 3 systemlevel interventions
Study Design
Pre-post
Adams24
Pizarro,2011
19
Blegen, 2010
Multi-component
intervention
Multi-component
intervention
Crisis resource
management training
Paraprofessionals
communication training
Pre-post
Outcomes: Benefits*
Culture survey: N
Teamwork: Y
Reported errors resulting in harm: Y
Overall reporting rate:
Culture survey: No statistical tests reported
Pre-post
Culture survey: Y
pre-post with
control hospitals
Pre-post
Culture survey: N
Multi-component
intervention
Executive walkrounds
Structured InterDisciplinary Rounds
Structured InterDisciplinary Rounds
Multiple interventions
Multi-component
Pre-post
TeamSTEPPS training
Cluster RCT
Pre-post
44
Cooper, 2008
Donahue, 2011
16
Edwards, 2008
47
15
Frankel, 2008
13
OLeary, 2010
OLeary, 2011
12
25
Paine, 2010
22
Pettker, 2009
21
Pettker, 2011
48
Pronovost, 2005
Riley, 2011
18
Sexton, 2011
23
Thomas, 2005
Tiessen, 2008
17
14
Multi-component
intervention
26
Timmel, 2010
Comprehensive UnitBased Safety Program
*Overall results statistically significant Yes or no
Pre-post
Concurrent
control
Concurrent
control
Pre-post
Pre-post
Cluster RCT
Pre-post
Pre-post
367
368
Low
Evidence or
Potential for
Harmful
Unintended
Consequences
Uncertain
Estimate of
Cost
Low-tomoderate
(varies)
Implementation Issues:
How Much do We
Know?/How Hard Is it?
References
1.
2.
3.
4.
5.
10.
11.
12.
13.
14.
6.
7.
369
15.
16.
17.
18.
19.
20.
21.
22.
23.
370
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
43.
35.
44.
36.
45.
46.
37.
38.
39.
40.
47.
41.
48.
49.
42.
371
372
Figure 1, Chapter 34. Hospital organization, nursing organization, and patient outcomes
The model of Despins and colleagues (Figure 2) explicitly posits that better detection of
potential signals of patients at risk of poor outcomes is the mechanism by which more effective
nursing care exerts its beneficial effects; it further elaborates that organizational culture is an
important component of better signal detection (e.g., high reliability organizations instill in their
staff the value they place on safety). Internal factors such as nurse fatigue also play a role in
this model.
Figure 2, Chapter 34. Despins model on patient risk detection
373
In the models proposed by Tourangeau (Figure 3) and by Thornlow (Figure 4), numerous
patient, system, nurse, nurse environment, and other factors are hypothesized to play an
important role in reducing inpatient mortality and other outcomes. The Tournageau model
explicitly posits that the use of care maps/protocols is associated with lowering the risk of
inpatient mortality.
Figure 3, Chapter 34. Tourangeaus model on determinants of 30-day mortality
374
Figure 4, Chapter 34. Thornlows model on cascade iatrogenesis: postoperative respiratory failure
375
Table 1, Chapter 34. Pooled odds ratios of patient outcomes corresponding to an increase of one
registered nurse full-time equivalent per patient day
No. Avoided Events/1000
Outcome
Studies
Odds Ratio (95% CI)
Hospitalized (95% CI)
All Patients
Mortality, intensive care units
5
0.91 (0.86; 0.96)
5 (2; 8)
Mortality, surgical patients
8
0.84 (0.8; 0.89)
6 (4; 8)
Mortality, medical patients
6
0.94 (0.94; 0.95)
5 (4; 5)
Hospital-acquired pneumonia
4
0.81 (0.67; 0.98)
1 (0; 2)
Pulmonary failure
5
0.94 (0.94; 0.94)
1 (1; 1)
Cardiopulmonary resuscitation
5
0.72 (0.62; 0.84)
2 (1; 2)
Intensive care units
Hospital-acquired pneumonia
3
0.7 (0.56; 0.88)
7 (3; 10)
Pulmonary failure
4
0.4 (0.27; 0.59)
7 (5; 9)
Unplanned extubation
5
0.49 (0.36; 0.67)
6 (4; 8)
Cardiopulmonary resuscitation
3
0.72 (0.62; 0.84)
2 (1; 2)
Surgical Patients
Failure to rescue
5
0.84 (0.79; 0.9)
26 (17; 35)
Nosocomial bloodstream infection
5
0.64 (0.46; 0.89)
4 (2; 5)
Table was adapted from Kane et al., 20077
Kane RL, Shamliyan TA, Mueller C, et al. The association of registered nurse staffing levels and patient outcomes - Systematic
review and meta-analysis. Medical Care. 2007 Dec;45(12):1195-204. Permission granted by Wolters Kluwer Health.
The EPC authors also conducted an indirect analysis of the potential for a dose-response
relationship. This analysis (Figure 5) assessed the effect across studies of additional RN-level
nurses per shift. In each case, comparisons of quartiles of nurse staffing levels showed the
expected relationship. In other words, if the association between nurse staffing and mortality is
causal, the difference in the risk for death should be greater between the 1st and the 3rd quartile
of nurse staffing than it is between the 1st and the 2nd quartile, because the difference in staffing
between the 1st and 3rd quartiles is greater than between the 1st and 2nd quartiles.
Figure 5, Chapter 34. Pooled odds ratio of quartiles of nurse staffing levels
376
The EPC review concluded that a consistent relationship has been demonstrated but
identified numerous limitations in the literature with respect to establishing that this relationship
is causal. Ultimately, the authors concluded that the arguments for a causal relationship are
mixed, and they called for future research to address the role of nurse staffing and competence
on the effectiveness of patient care, taking greater cognizance of other relevant factors such as
patient and hospital characteristics and quality of medical care.
The Tourangeau search identified literature published through 2009 and was restricted to
studies that used hospital-related mortality as the outcome; the authors identified 17 studies (10
of which were not included in the Kane review, seven published since 2007). Although the
Tourangeau review was narrative (not a meta-analysis like the EPC review), the two had broadly
similar results: 14 of 17 studies found a statistically significant relationship between nurse
staffing variables and lower mortality rates (see Evidence Table in Appendix D). In addition,
Tourangeau and colleagues identified mixed findings for mortality among five studies assessing
the characteristics of the nurse work environment and work relationships, three studies assessing
nurses responses to work and the work environment (e.g., burnout), and seven studies assessing
nurses educational preparation and experience. Only one study assessed any nursing process-ofcare variables; it found a cross-sectional relationship between the use of care maps and lower
hospital-associated mortality, with an estimated effect size of 10 fewer deaths per 1000 acute
medicine discharged patients. Like the EPC review, the review by Tourangeau concluded that a
strong relationship exists but that more research is needed to understand the reasons that this
relationship between higher nurse staffing and lower hospital mortality might be causal; that is,
they called for a theoretical model that explains the relationship in ways that can be tested and
refined.
Thus, these two reviews came to broadly similar conclusions: Mostly cross-sectional studies
consistently report that higher RN staffing is associated with lower hospital-related mortality.
However, as Kane and colleagues ask, does this association reflect a causal relationship? If it
does not, then an intervention that simply hires more RN-level nurses may not achieve the
desired result. Indeed, mandates for fixed nurse-patient ratios have been critiqued as being an
inflexible solution which is unlikely to lead to optimal use of resources.9
Any number of factors might confound the observed relationship: In cross-sectional studies,
hospitals that are better in a variety of other ways might also be better staffed with RN-level
nurses. For example, one published study of electronic health record (EHR) implementation
showed that hospitals with EHRs have higher nurse staffing ratios and lower patient mortality.10
Longitudinal studies overcome these kinds of limitations in cross-sectional studies, but
imprecision in the measures of nurse staffing and of the severity of patient illness (which may
increase the risk of death via other, non-nursing-sensitive ways) constitute potential threats to the
validity of the association between nurse staffing and mortality.
Update Review
To supplement the two existing reviews, we used the Web of Science to conduct an update
search for articles published from 2009 onwards that cited any of four landmark articles in this
field. Our update search identified 546 titles, and 4 articles came from reference mining. From
550 titles, we identified 9 longitudinal studies and 1 new systematic review.11-20 The systematic
review included studies that assessed nurse staffing ratios and outcomes restricted to adult ICU
settings20 and reached conclusions similar to the previous reviews: a consistent relationship
between increased nurse staffing and better patient outcomes in observational studies, evidence
377
that falls short of causality. One longitudinal study narratively reported that increased nurse
staffing was related to significantly (P 0.01) decreased rates of decubiti, pneumonia, and
sepsis, but data were not presented.14
We discuss the 1 cross-sectional study because it addresses the effect of an intervention to
change nurse staffing ratios, implemented in response to a 2004 California law requiring
minimum nursepatient ratios in acute care hospitals.21 This legislation mandated patientnurse
staffing levels of 5:1, 4:1, and 2:1 for medical or surgical units, pediatric units, and ICUs,
respectively. The California legislative mandate does not require nurse staffing to be met with
RNs (that is, licensed vocational [practical] nurses can also meet the mandate).
Aiken and colleagues21 assessed the relationship between nurse staffing and mortality in
2006, 2 years after the California mandate, comparing data from California with those of two
states without mandates New Jersey and Pennsylvania. Data about workloads were drawn
from a survey of RNs in the three states22,336 nurses in totalwith a response rate of 35.4
percent. Hospital data came from the American Hospital Association, and patient and outcome
data came from State hospital discharge databases.
The authors reported that their survey data showed substantial compliance with the California
mandate, with 88 percent of medical/surgical, 85 percent of pediatric, and 85 percent of ICU
nurses reporting that on their last shift they were within the mandated staffing ratios. This level
of compliance is higher (sometimes considerably) than the values of 19 percent, 52 percent, and
63 percent for the same settings in New Jersey and 33 percent, 66 percent, and 71 percent in
Pennsylvania. In logistic regression analyses adjusted for a large number of patient
characteristics and three hospital characteristics (bed size, teaching status, and technology use),
Aiken and colleagues found statistically significant relationships between the estimation of the
average number of patients per nurse and two outcomes: 30-day mortality and failure to rescue
(Table 2).
Table 2, Chapter 34. Odds ratios indicating the effect of nurse staffing on 30-day inpatient
mortality and failure to rescue, in California, New Jersey, and Pennsylvania
State Hospital
Sample
California
These associations were found for all three states. The authors also provided several
measures of nurse-assessed practice environment characteristics taken from their survey
responses, such as a reasonable workload and enough staff to get work done; all consistently
favored California over New Jersey or Pennsylvania. The authors concluded that, 2 years after
the California mandate, nurse patient care loads were significantly lower in California than in
either New Jersey or Pennsylvania; on average, these loads were one patient fewer, and in the
378
medical/surgical units they were closer to two patients fewer. California nurses were also more
likely to report favorable practice environment characteristics.
Although the study by Aiken and colleagues21 collected data after the implementation of
Californias staffing mandate, it did not test the effect of that mandate per se because it had no
comparison data from the period before the mandate went into effect. The possibility that the
relationship is causal is blunted by a longitudinal study that examined measures from before and
after the California mandate and showed the expected changes in nurse staffing and proportion of
licensed staff per patient but no improvement in two patient outcomes believed to be nursingsensitive: falls and pressure ulcers.11,13 In fact an unexpected statistically significant increase in
pressure ulcers was associated with a greater number of hours of care for the patient (which may
have been due to greater detection). This study did not assess mortality.
Five additional longitudinal studies add further information to this picture. The first is a
longitudinal assessment of nurse staffing and hospital mortality and failure to rescue in 283
California hospitals between 1996 and 2001, which had access to direct measures of nurse
staffing.15 In multivariable models that included numerous hospital market characteristics as well
as risk adjustment using the Medstat Disease Staging Methodology to produce a predicted
probability for complications or death, the authors found that an increase of one RN FTE per
1,000 inpatient days was associated with a statistically significant 4.3 percent decrease in
mortality.
The second longitudinal study assessed care at 39 Michigan hospitals between 2003 and
2006; it included adults admitted through the emergency department with acute myocardial
infarction, heart failure, stroke, pneumonia, hip fracture, or gastrointestinal bleeding.17 This
study simultaneously controlled for high hospital occupancy on admission, a weekend
admission, seasonal influenza, and nurse staffing levels. Each factor had a statistically significant
increased effect on in-hospital mortality. Each additional RN FTE per patient day was associated
with a 0.25 percent decrease in mortality.
The third longitudinal study assessed the effect of a mandate in three Western Australia
public hospitals to implement a new staffing method, the Nursing Hours Per Patient Day
(NHPPD).18 The study assessed three time periods: 20 months before implementation 7 months
of a transition period, and 2 months post implementation. The authors found that the total
nursing hours and RN nurse hours increased during the observation period. However, the
percentage of total nursing hours provided by RNs decreased (from 87% to 84%). Also, the
article stated that although the nursing hours increased for all three hospitals (in the postimplementation period), the changes were not statistically significant, Mortality rates improved
during this time period. Among a host of other outcomes, some improved, others did not, and
some changes were inconsistent across hospitals. Although the study was described as an
interrupted time series, it was analyzed as a before-and-after study.
The fourth longitudinal study assessed changes in nurse staffing over 9 years in 124 Florida
hospitals and related these to changes in Agency for Healthcare Research and Quality Patient
Safety Indicators.19 The study used both initial staffing ratios and changes in staffing ratios.
Results were mixed but generally favored better patient safety outcomes with higher RN staffing
levels.
The methodologically strongest longitudinal study is discussed here in more detail.16 In this
study, Needleman and colleagues used data over time from a single hospital to assess the
association between naturally occurring differences in levels of RN staffing within the same
hospital and inpatient mortality. This study is further characterized by a careful matching of
379
nurse staffing on a shift-by-shift basis with the actual patients cared for during that shift.
Knowing the actual patients cared for allowed for more sophisticated adjustments at the patient
level of risk-of-death. The study was carried out at a tertiary academic hospital between 2003
and 2006 and included 197,691 admissions and 176,696 nursing shifts, across 43 hospital units.
The patients themselves averaged 60 years of age, and about 50 percent were covered under
Medicare. The variable of interest was exposure of the patient to nursing care that was below the
target level (for that type of unit) for that shift, in other words the proportion of shifts below
target level staffing, on a patient basis. An additional exposure variable was a high turnover
shift (in other words, a shift with many admissions, discharges, or transfers). The authors found
that exposure to each shift of below-target staffing or high turnover was associated with a 2 to 7
percent increase in mortality, with higher levels of risk if the high-turnover or below-target shift
occurred in the first 5 days after admission (see Table 3). For patients who were not in an ICU,
this increased risk rose to 12 percent or 15 percent.
Table 3, Chapter 34. Risk of death associated with exposure to a shift with an actual RN staffing
level 8 hours or more below target, high patient turnover, and other variables
Variable
Total of 197,961 patients
Each shift with RN staffing level below target or high turnover
during first 30 days after admission
Shift with RN staffing level 8 hr or more below target
Shift with high patient turnover
Each shift with RN staffing level below target or high turnover
during first 5 days after admission
Shift with RN staffing level 8 hr or more below target
Shift with high patient turnover
Total of 171,041 patients with no shifts in an ICU
Each shift with RN staffing level below target or high turnover
during first 30 days after admission
Shift with RN staffing level 8 hr or more below target
Shift with high patient turnover
Each shift with RN staffing level below target or high turnover
during first 5 days after admission
1.02 (1.01-1.03)
1.04 (1.02-1.06)
1.03 (1.02-1.05)
1.07 (1.03-1.10)
1.04 (1.03-1.06)
1.07 (1.02-1.13)
The data from Needleman and colleagues contribute to the causality determination because
the study is longitudinal within one hospital, thus controlling for the hospital effect potentially
present in all cross-sectional studies, and has detailed measures of exposure and confounding
variables. These results, and the dose-response analysis from the EPC review, are the two
strongest pieces of evidence in support of causality.
The longitudinal assessments from California11 and Western Australia18 reported an increase in
pressure ulcers associated with increased nurse staffing, although this development may reflect
increased detection. Almost no other studies mentioned an explicit assessment of potential
unexpected adverse outcomes.
381
medical center that has a lower-than-expected in-hospital mortality rate and a reputation for
excellence. Therefore, the association between increased RN staffing and lower mortality, if it is
causal, is potentially applicable even to high-performing hospitals.
Moderate
Evidence or
Potential for
Harmful
Unintended
Consequences
Low
Estimate of
Cost
High
Implementation Issues:
How Much do We
Know?/How Hard Is it?
A lot/Not difficult
Table 5, Chapter 34. Summary table for increasing nurse-to-patient staffing ratios to prevent falls,
pressure ulcers, and other nursing sensitive outcomes (other than mortality)
Scope of the
Strength of
Problem Targeted by Evidence for
the PSP
Effectiveness
(Frequency/Severity) of the PSPs
Common/High
Low
Evidence or
Potential for
Harmful
Unintended
Consequences
Low
Estimate of
Cost
High
382
Implementation Issues:
How Much do We
Know?/How Hard Is it?
A lot/Not difficult
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
383
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
384
22.
23.
24.
25.
385
to treatment of such high risk, rapidly developing conditions is another important frontier for
those seeking to improve consequential diagnostic delays.
Problems in care related to diagnosis are particularly prevalent among precipitating causes
for lawsuits, with studies reporting 25 percent to 59 percent of malpractice claims attributable to
diagnostic errors.5,10,11 A recent study of 91,082 diagnosis-related malpractice claims from 1986
to 2005 estimated payments summing to 34.5 billion dollars (inflation-adjusted to 2010 dollars),
well over one billion dollars per year. The mean per-claim payout was $378,858 (interquartile
range: $72,250 to $472,000).12 Diagnosis-related claims made up 29.1 percent of total claims and
accounted for the highest proportion of total payments (35.6%). In terms of severity, lethal
injuries accounted for 40 percent of total payments. Another study of 10,739 malpractice claims
from the 2005-2009 National Practitioner Data Bank found that diagnosis-related reasons
accounted for 45.9 percent of paid claims from outpatient settings (95% confidence interval [CI],
44.4 to 47.4), the most frequently cited reason from that setting. Diagnostic reasons were the
second-most frequently cited for paid claims in the inpatient setting (21.1%; 95% CI, 20.0 to
22.3) and when both settings were involved (26.7%; 95% CI, 23.9 to 29.5).13
Some have asserted that diagnostic errors are more likely to be preventable and more likely
to result in patient harms than other types of errors (e.g., treatment-related errors, such as wrongsite surgery or incorrect medication dose), making the problem particularly important as well as
useful to address.14 Given this potential, the purpose of this review is to assess the multitude of
interventions to prevent diagnostic errors and better understand their effectiveness.
386
Patient factors: systems may be designed around areas that are more prone to risk (e.g.,
improved staffing with translators).
Human/clinician factors: interventions may aim at errors of planning separately from
errors of execution, and may also be designed to address cognitive error, skill-set error,
task-based error, and/or personal impairment.
Outside care systems, ED access, and triage: consideration of these three framework
areas aims to understand patterns of failure and errors that affect patients before their
arrival in the ED or initiation of care.
Teamwork: interventions in this area focus on communication, coordination, conflict
resolution, personnel assignment practices (e.g., considerations of capability, workload),
and training.
Local ED environment, hospital environment, hospital administration and third parties,
and community level: systems and resources at each of these four additional levels of the
framework have potential for effective interventions to reduce diagnostic errors within
the ED and after the patient leaves.
Within the above framework, human and clinician factors have received significant attention
from researchers interested in diagnosis. Cognitive factors may affect diagnostic accuracy
through rote over-learned actions or through purposive reasoning and decisionmaking processes.
The cluster of automatic or quasi-automatic decisionmaking processes may be classified as
heuristics, or rule-based decisionmaking processes. Heuristics aid in making decisions quickly
and are important for keeping cognitive capacity high for other, more demanding, cognitive
tasks. However, the very thing that makes heuristics helpful, decisions based on logical
assumptions gained from experience, can also lead to systematic bias and incorrect
decisionmaking when assumptions are wrong.16 Other cognitive processes affecting diagnosis
involve working memory in conjunction with learned knowledge, or more plainly, information
that is purposefully stored, recalled and used for completing a current goal. An example of these
cognitive processes can be seen in physicians listening to their patients describe symptoms. The
physician cognitively stores symptomatic information in the short term until she or he can
classify the symptoms into a more general descriptive category of a diagnosis. This process is
also subject to error when attention is pulled away from the task at hand or cognitive capacity is
altered for others reasons (e.g., lack of sleep). The process of metacognition involves continued
focusing and re-focusing attention on these cognitive processes so as to reflect on ones own
potential for biases, incorrect assumptions, and reduced cognitive capacity.17 Ultimately, both
human factors and the systems within which they operate have long been recognized as unique
contributors to human error.18
PSPs relevant to diagnostic error are also being actively developed by those bringing more
attention to this important patient safety target, and drawing on previous work in the research
domains of medical problem solving, decision analytic/normative decisionmaking, and clinical
diagnostic decision support.19 As health information technologies become more pervasive,
electronically-supported workflow and system redesign might target preventing or mitigating
diagnostic errors. PSPs in this area would be akin to computerized physician order entry with
clinical decision support, though more aptly named something like computerized diagnosis
management.
387
388
potential for clinical consequence. The Supplementary Evidence Table (see Appendix D, Table
2) provides basic descriptions of targeted diagnostic errors, intervention descriptions, patient
outcome, study design and results with respect to the effectiveness of the proposed interventions.
Drawing from frameworks proposed by others, we classified interventions into one or more
of the following six types (Figure 1):
Technique (introduction of novel technologies for testing, adaptations of testing
equipment, or changes in medical interventions potentially affecting diagnostic
performance)
Additional Review Methods (introduction of additional steps from the interpretation
through reporting of test results)
Personnel Changes (introduction of additional health care members and/or replacing
certain professionals with others)
Educational Interventions (implementation of educational strategies)
Structured Process Changes (implementation of feedback systems or additional stages in
the diagnostic pathway)
Technology-based Systems Interventions (implementation of technology-based tools at
the system levelcomputer assistive diagnostic aids, decision support algorithms, text
message alerting, pager alerts, etc.)
Figure 1, Chapter 35. Interventions by type
Interventions by Type
(% of Total)
Technique
13%
Educational
9%
Technology-based
Systems
22%
Structured Process
22%
Personnel
5%
Additional Review
M ethods
29%
This pie chart illustrates the percentage of studies as categorized to the six intervention types: Technique, Educational,
Technology-Based Systems, Personnel Changes, Additional Review Methods, and Structured Process Changes.
All six of the evaluative studies identified by Singh and colleagues,23 many of the evaluative
studies identified by Graber and colleagues,24 and most of the studies included in our systematic
review, reported beneficial effects along the diagnostic pathway for a broad array of intervention
types. Because the evidence is predominantly from uncontrolled before-after study designs or
other uncontrolled study types (Table 1) with markedly different outcomes, the strength of the
evidence about interventions to reduce diagnostic errors is insufficient to draw any strong
conclusions. Furthermore, the magnitude of difference attributable to interventions varied by
study and clinical process. For example, some researchers demonstrated what would be
moderate-to-large effects on diagnosis if the assumption of causality were made (e.g., Perno and
colleagues, 2005),25 although methodologies were not designed to test causality, whereas other
389
studies were designed to demonstrate the absence of change in diagnostic outcomes despite
intervention (e.g., Thomas and colleagues, 2003).26
Table 1, Chapter 35. Study design distribution
Study Design
Description
Number of
Studies*
14
Randomized,
controlled trial
(RCT)
A standard randomized controlled trial involving two groups; a control and the
intervention group.
Experimental
Design
Study with a concurrent usual care control group, or another method for
controlling and comparing between experimental and usual care nonexperimental groups (but not including the pre/post method)
12
Pre/Post
16
Other
58
*Number of studies adds to more than 95 because several had multiple designs.
As a result of the state of the science in this area, no meta-analyses have been conducted.
Pooled analysis may be feasible in the near future as the evaluative literature is growing rapidly
in some intervention categories. Figure 2 shows particular increases for several classes of
interventions: Additional Review Methods, Technology-based Systems Interventions, and
Structured Process Changes. The other intervention types have not been studied much over the
entire period.
Figure 2, Chapter 35. Intervention studies by year
The graph illustrates a timeline of the included studies broken down by the six intervention types.
Few studies (5 randomized, controlled trials and 8 other designs) have evaluated patient-level
clinical outcomes to reduce diagnostic errors.9,27-38 Diagnostic errors have a complex relationship
with direct patient outcomes because they can play a role at many different time points in a
patients care; that is, many opportunities exist to catch diagnostic errors. If a diagnostic error is
caught at any of these opportunities, then negative effects on clinical outcomes could potentially
be avoided. Thus, examining the direct relationship between diagnostic errors and clinical
390
outcomes is complex and explains why many of the articles do not report on hard patient
outcomes. The remainder of this section summarizes the findings of the review.
Diagnostic Techniques
The studies of interventions related to medical techniques (n=14)26,31,72-83 demonstrated that
technologies as well as diagnostic test selection might either enhance diagnosis (e.g., visual
enhancements via ultrasound-guided biopsy, changes to number of biopsy cores, cap-fitted
colonoscopy) or impede it (e.g., medical interventions for pain relief in patients with abdominal
pain). In the latter cases, the interventions hypothesized to impede diagnosis did not have that
effect, and interventions expected to enhance diagnostic accuracy did not always do so.
Personnel Changes
Six studies36,37,67,69,84,85 compared the impact on diagnosis of substituting one type of
professional for another, or adding another professional to the care team. The three studies67,69,85
that added a specialist to examine the interpretation of a test result reported an increase in case
detection, although the studies were quite small and targeted narrow patient populations.
391
Educational Interventions
Ten studies employed educational interventions35,61-64,86-90 for various targets: consumers,
community doctors, and intensive care unit doctors and nurses. Strategies targeted at
professionals produced improvements. Only two studies targeted consumers (parents, candidates
for screening) and both intervened on a behavior that occurs much earlier than actual diagnosis
(e.g., awareness of symptom seriousness with the intent of reducing office visits in ways that
would not adversely affect diagnosis)86
392
comparative designs. The review by Singh and colleagues identified seven studies of patient
preferences or satisfaction with different options for receipt of test results.23 However, they also
found no studies that tested ways to reduce error using an intervention that affected test
notification. One of the articles identified in the Singh review by Casalino and colleagues found
a 7.1 percent rate of apparent failures to inform patients of an abnormal test result, and identified
an association between use of simple processes by physician practices for managing results and
lower failure rates.120 A systematic review of failures to follow-up test results with ambulatory
care patients reported that failed follow-ups ranged from 1 percent to 62 percent depending on
type of test result, and these failures were associated with missed cancer diagnoses. Electronic
record systems appeared to exert a mild protective effect against failed follow-ups, although the
authors note the pool of literature was small in this analysis.121
393
with interventions that have ongoing incremental costs. Finally, information technology PSPs to
reduce diagnostic errors may be relatively expensive, though these costs could vary as well.
tests. Eventually reaching the correct diagnosis with inefficient testing strategies (for example,
some sequences of multiple test ordering) is not the appropriate pathway to improved diagnostic
safety. Our review found a paucity of studies that assessed both sensitivity and specificity of
interventions addressing diagnostic performance in the context of mitigating diagnostic errors.
Thus, although we found several promising interventions, evaluations need to be strengthened
before any specific PSPs are scaled up in this domain.
Alongside the literature scoping the problem and generating ideas for potential solutions,
some are also working on policy level efforts. Singh and Vij describe potential institutional-level
policies for communicating test results within the clinical team and to the patient.123 These types
of policies respond to national attention (e.g., the Joint Commission Patient Safety Goals),
spotlighting this part of the diagnostic pathway as ripe for intervention. They note that the area of
notifying patients about their test results is an emerging area for intervention testing.
In conclusion, our review demonstrates that the nascent field of diagnostic error research is
growing, with new interventions being tested that involve technical, cognitive, and systemsoriented strategies. The framework of intervention types developed in the review provides a basis
for categorizing and designing new studies, especially randomized, controlled trials, in these
areas. A summary table is located below (Table 2).
Table 2, Chapter 35. Summary table
Scope of the
Strength of
Problem targeted by Evidence for
the PSP
Effectiveness
(Frequency/Severity) of the PSPs
Common/High
Emerging
practice (few
studies
available)
Evidence or
Potential for
Harmful
Unintended
Consequences
Uncertain
Estimate of
Cost
Varies
Implementation Issues:
How Much do We
Know?/How Hard Is it?
Varies
References
1.
2.
395
3.
4.
5.
6.
7.
8.
9.
10.
11.
396
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
397
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
398
48.
49.
Carew-McColl M. Radiological
interpretation in an accident and emergency
department. Br J Clin Pract. 1983;37(1112):375-7. Epub 1983/11/01. PubMed
PMID: 6671078.
50.
51.
56.
57.
58.
59.
60.
61.
62.
52.
Kwek BH, Lau TN, Ng FC, Gao F. Nonconsensual double reading in the Singapore
Breast Screening Project: benefits and
limitations. Annals of the Academy of
Medicine, Singapore. 2003;32(4):438-41.
Epub 2003/09/13. PubMed PMID:
12968545.
53.
54.
55.
399
63.
64.
65.
66.
67.
68.
69.
400
70.
71.
72.
73.
74.
75.
76.
77.
78.
79.
80.
81.
82.
83.
84.
85.
86.
87.
88.
401
89.
90.
91.
92.
93.
94.
95.
402
96.
97.
98.
99.
100.
101.
102.
103.
104.
105.
106.
107.
108.
403
109.
110.
111.
112.
113.
114.
115.
116.
117.
118.
119.
120.
121.
404
122.
123.
405
to rate care in the context of a bad outcome as substandard), if the evaluators are not blinded to
outcome.15
Chart review was often used as the benchmark for estimating the extent of medical harms in
hospitals or as the gold standard in patient safety studies to quantify AE rates.15,16 However,
chart review is generally resource-intensive.15 Incomplete documentation in the medical record
can affect the ability to detect the potential causes of AEs.15 Near misses that produce no injury
are rarely detected by this method.15,16 The reliability (precision) of the judgments about the
presence of AEs by chart reviewers could also be low.15
Incident reporting systems are a popular mechanism that the majority of hospitals rely on to
uncover internal threats to patient safety.1-4,20,21 Since IOM endorsed using incident reporting
systems in its landmark report on patient safety, 27 states and the District of Columbia have
established hospital AE reporting systems.13,22 Reporting systems include surveys of providers
and structured interviews and can provide rich information about medical errors that lead to AEs.
Incident reporting systems can identify latent errors (system problems) not uncovered by some
other methods; thus, they can be used to improve patient safety.13,15,16 In comparison with
comprehensive chart reviews, incident reporting is also relatively inexpensive.15
However, like other methods for detecting safety problems, incident reporting has its own
limitations. Incident reporting systems alone cannot reliably measure incidence and prevalence
rates of errors and AEs.20,23 Providers may not report errors because of busy schedules, concerns
about potential lawsuits, fear their reputations could be tarnished, or misperceptions about what
constitutes patient harm.20 As a result, reported incidents may represent only a portion of serious
incidents and may misguide detailed investigation efforts to less important targets.16,18,20,23,24
Additionally, the rates of incidents reported over time may not reflect real changes in safety in an
institution, because an increased rate may simply indicate an improved commitment by the
institution to identify medical errors rather than a true rise in medical hazards.23
In recent years, with the adoption of electronic medical records, computerized surveillance,
including using electronic triggers, has become an increasingly popular method for identifying
certain types of medical errors or AEs, particularly those related to use of medications.25-28 By
integrating multiple data sources (e.g., electronic medical records, laboratory, pharmacy, billing),
computerized surveillance may efficiently detect medical errors and AEs and could provide realtime information for preventing harm to patients from errors in medical treatment.25-28 Use of
electronic medical record-based surveillance of diagnostic errors was also reported.29 However,
the accuracy and reliability of the tools for computerized surveillance need further study.30 The
initial cost of the systems remains another barrier to implementation.25
Similarly, other methods for detecting and monitoring patient safety problems (e.g., chart
audits assisted with trigger tools, direct observation of patient care, executive walk rounds,
administrative data analysis, data warehouses) also have their strengths and weaknesses. We
have identified several documents that provide an overview of these methods based on
systematic or targeted literature review.14-16 Table 1 summarizes the purposes of different patient
safety problem detecting methods. Tables 2 and 3 summarize the strengths and weaknesses of
these methods. Because the original documents used different taxonomies for the methods, we
compiled the adapted tables together in this chapter to provide a more comprehensive overview.
As a result, some contents in these tables may overlap.
406
Table 1, Chapter 36. Overview of the purposes of different methods for detecting patient safety
problems
Method
Harm
Active Errors
Direct observation
NR
NR
NR
NR
NR
NR
NR
NR
NR
NR
NR
14
407
Table 2, Chapter 36. Advantages and disadvantages of different methods used to measure errors
and adverse events in health care (from the Thomas and Petersen study)
Method
Advantages
Disadvantages
Administrative data
analysis
Chart review
Clinical surveillance
408
Hindsight bias
Reporting bias
Focused on diagnostic errors
Infrequently and nonrandomly utilized
Hindsight bias
Reporting bias
Nonstandardized source of data
Reporting bias
Hindsight bias
Table 3, Chapter 36. Advantages and disadvantages of different methods for hospitals to monitor
for internal patient safety problems (from the Shojania study)
Method
Traditional incident
reporting
Advantages
Process already ubiquitous
Can identify latent errors
(system problems)
Stimulated/facilitated
incident reporting
Patient complaints
Malpractice claims
409
Disadvantages
Underreporting of serious incidents
Frequent reporting of events
not suited to individual analysis
(e.g., falls)
Can be demoralizing when staff
do not perceive meaningful
improvements resulting from
incidents they have reported
Cannot assess changes in safety
(over time)
More labor intensive than traditional
incident reporting
Greater engagement of staff
increases importance of making
meaningful improvements (i.e., even
more demoralizing than usual
if improvements not made)
Cannot assess changes in safety
May be dismissed by clinicians as
service problems
May require more up-front work
(compared with incident reporting) to
identify incidents worth analyzing in
detail for potential safety
improvements
Cannot assess changes in safety
Heavily biased toward detecting
diagnostic issues and procedural
complication (though these are
usually not detected by other
systems)
Cannot assess changes in safety
Probably similar to malpractice
claims, but not clear
Demoralizing to frontline staff if
management focuses only on
improved public relations
(management cares) and does not
seriously address the problems
identified
Tempting for management to focus
on easy fixes (e.g., related to
equipment) not deeper problems or
those requiring substantial
investments of resources
(e.g., staffing, skill mix, or work-load
problems)
Table 3, Chapter 36. Advantages and disadvantages of different methods for hospitals to monitor
for internal patient safety problems (from the Shojania study) (continued)
Method
Chart audits (commonly
operationalized with
trigger tools)
Electronic triggers
(e.g., drug-lab
combinations, use of
antidotes suggestive of
medication errors)
Performance indicators
derived from
administrative data
Data warehouses
Modifying traditional
morbidity and mortality
rounds with modern
patient safety framework
Advantages
Types of events captured may be
more likely to engage frontline
clinicians (especially physicians)
Produce rates that can be monitored
over time, not just counts or
frequencies susceptible to changes in
reporting biases
Very efficient
Potentially high sensitivity capture for
the events captured
Data easily available
In principle, event rates can be
tracked over time, but in practice
probably applies only for frequent
event types in large health care
systems
Discrepancies between
clinical and autopsy
diagnoses
Monitoring pathologic
discrepancies
(e.g., between cytology
and histology or
antemortem biopsies and
autopsies)
Corrected laboratory
results/reports
Relatively efficient
Can identify patterns of problems
amenable to substantial improvement
projects
Event rates can be followed over time
Relatively efficient
Event rates can probably be followed
over time
410
Disadvantages
Requires willing clinicians to
participate
Many important events not
documented in charts and
contributing factors for documented
events typically unclear
Many triggers have low specificity
Captures only certain types of events
(small subset of events involving
medications or laboratory tests)
Trade-offs between sensitivity and
specificity
Low signal-to-noise ratio
Various methodologic problems
leading to misleading
characterizations of performance
Managers and clinicians tend to
distrust these data (often with good
reason)
Requires intensive effort to
investigate if poor performance is real
and further effort to determine causes
Requires substantial up-front
investments and appropriate clinical
and methodological expertise
Requires organizational culture and
management structures conducive to
driving change on the basis of these
novel data
Care required to avoid traditional
focus on individual errors and
blaming other departments
New processes required to follow-up
systematically on issues identified
(traditional rounds heavy on
discussion, with follow-up occurring
only haphazardly)
Addressing problems identified often
requires host department to engage
and collaborate with other
departmentsdepartures from
traditional norm
Likely to succeed only in select
hospitals because of low autopsy
rates and decreased interest in
autopsies among clinicians and
pathologists at most hospitals
Requires interest on the part of
pathologists to undertake this
nontraditional form of quality
assurance and willingness of clinical
departments to collaborate in
improvement projects
Fairly narrow focus
Requires interested laboratory
medicine personnel
Table 3, Chapter 36. Advantages and disadvantages of different methods for hospitals to monitor
for internal patient safety problems (from the Shojania study) (continued)
Method
Natural language
screening of electronic
portions of medical
records
Advantages
Relatively efficient once implemented
Reasonable sensitivity and specificity
for certain types of events
Direct observation
(e.g., audits of hand
hygiene compliance,
medication administration,
operating room
procedures daily rounds)
Active surveillance
(combination of chartbased trigger tool applied
in quasi-real time,
stimulated reporting, and
other interactions with
frontline staff)
Telephone calls to patients
(can be automated)
Disadvantages
Requires appropriate technical
expertise and initial investment of
time to develop and refine
combinations of search terms with
acceptable sensitivity and specificity
for safety problems
Somewhat labor intensive (but short
periods of measurement may provide
ample data)
Requires appropriately trained
observers
Care must be taken not to create
mistrust among frontline staff
Somewhat labor intensive (but short
periods of measurement may provide
ample data)
Requires appropriately trained
observers
Care must be taken not to create
mistrust among frontline staff
Requires appropriate technology and,
even with automation, still requires
investment of personnel time (e.g., at
least one nurse case manager and a
physician) to respond in real-time to
clinical problems
As Tables 2 and 3 have demonstrated, health care organizations have been using a wide array
of methods to detect AEs and medical errors.14-19 Many of these methods (e.g., trigger tools) can
be further categorized by the targeted problems (e.g., medication-related medical errors or
iatrogenic infections), tools, algorithms, and data source used. Given the limited timeframe for
this review, we focus this chapter on general approaches to detecting patient safety problems that
involve using multiple methods (e.g., incident reporting, executive walk rounds, clinical
surveillance, chart review, and trigger tools) to collect data.
We primarily reviewed studies that compared the utilities of different methods. However,
comparison studies that used any method as a gold standard to validate another method were not
included for this chapter, because, in essence, these studies still focused on one individual
method (i.e., the method being validated). We believe that understanding the strengths and
weaknesses of various methods is crucial for decisionmakers who need to form an effective
strategy for monitoring patient safety problems that is appropriate for their organizations.
Readers who are seeking information on individual methods can refer to studies and reviews
specifically focusing on those methods. As we reviewed the literature for this chapter, we
identified a large number of publications focusing on individual methods, particularly in the
areas of incident reporting, chart review, and trigger tools. Some systematic or targeted reviews
provided insightful summaries about commonly used methods.13,18,22,31,32
care organizations must systematically identify and mitigate patient safety risks and hazards with
an integrated approach to continuously drive down rates of preventable patient harm.33 As
several landmark studies have suggested, medical errors are often a system failure where care
practices are inconsistent among health care professionals.6,34,35 By systematically uncovering
these errors and analyzing their causes, health care institutions can identify defects in processes
of care and design system changes to prevent the errors.18,19,23
In the 1999 report, To Err is Human: Building a Safer Health Care System, the IOM also
acknowledged the need to learn from medical errors and recommended establishing mandatory
incident reporting systems as part of an approach to improving safety.6 The report noted that one
of the causes of medical errors is lack of reliable data on the number of medical errors, which
limits the ability to identify the problems origins and develop initiatives to resolve the problem.
A subsequent IOM report, Crossing the Quality Chasm: A New Health System for the 21st
Century, reinforced the need for reliable data and noted a need for evidence-based policies and
practices.36 By performing root-cause analyses (an in-depth examination of the data to identify
factors in the care process that contribute to the errors) and implementing corrective action plans,
health care organizations may be able to address system and process failures to ensure that
potential errors are prevented in the future.5,7,22,23
412
As discussed, we primarily reviewed studies that compared the utility of different methods.
Our search identified one systematic review published by the World Health Organization (WHO)
in December 2003.14 This study by Michel reviewed methods for assessing the nature and scale
of harm caused by health systems. The objective of the study was to identify the strengths and
weaknesses of available methods according to a defined set of criteria. These criteria included
the following:14
Effectiveness in capturing the extent of harm (in different environments).
Availability of reliable data (judged by interobserver reliability).
Suitability for large-scale or small, repeated studies. (Large-scale studies refer to national
and regional studies. Small, repeated studies are carried out for a limited period at the
hospital or local level.)
Costs (financial, human resources, time, and burden on system).
Effectiveness in influencing policy (focused on national, regional, or local policy or
strategic programs).
Effectiveness in influencing hospital and local safety procedures and outcomes.
Synergy with other domains of quality of care.
This set of criteria was defined by the WHO Working Group in Patient Safety: Rapid
Assessment Methods for Assessing Hazards in December 2002. The first four criteria focused
on the intrinsic characteristics of the methods, their validity, reliability, and cost. The last three
criteria were more related to the ability of the methods to trigger improvements in safety cultures
and the quality of safety programs. The study reviewed 262 relevant studies. With the exception
of comparative studies available for the assessment of effectiveness in capturing the extent of
harm, the literature consisted mostly of descriptive studies. For the review, Michel considered
the data reported in the included studies as well as the opinions of the authors of the studies.
The study rated each method on all seven criteria to produce a summary of its key strengths
and limitations.14 When valid information was available, the author rated the criteria from 1
(least favorable) to 4 (most favorable). A study was defined as valid when an appropriate
description of the method (sampling strategy, data collection, and data analysis) in line with
current standards was available. The lowest level (1) indicates low effectiveness, suitability, or
availability, or it means very high cost. Where the amount of evidence-based data was small, the
author noted to be confirmed. The evidence-based ratings for each method in the seven areas
are provided in Table 4. In the absence of valid data, the author used a subjective rating scale
from 1 (least favorable) to 4 (most favorable), based on the opinions of the studies being
reviewed. These opinion-based ratings are provided in Table 5. Both Table 4 and Table 5 were
based on literature from developed countries. The author also reviewed literature from
developing countries, but that information is not discussed in this chapter.
The WHO study revealed that the methods for assessing the nature and scale of harm caused
by health systems have different purposes (Table 1), strengths, and limitations (Table 4 and
Table 5).14 The main conclusion of the study was that these methods do not compete with each
other. Instead, they complement each other by providing different levels of qualitative and
quantitative information. The list of methods and the illustrative ratings (Table 4 and Table 5)
provided by the study may serve as a starting point for choosing appropriate methods for
detecting harms caused by health organizations. The author suggested that identification of
appropriate methods must take into account the distinct environmental factors faced by each
health care organization or region.
413
This WHO study also had limitations.14 First, the studies included for review varied in
quality and quantity. For some methods of interest, such as interviews with health care providers,
analysis of administrative data, or confidential inquiries, few studies were available for review.
Some other methods that the author thought might be useful for detecting safety problems, such
as single case analysis and focus group discussions, were not covered by the study. Second, the
rating systems and criteria used in the study for judging the strengths or weaknesses of the
methods were not adequately validated. The assessment of the methods was generally subjective
rather than objective. Third, because of the wide variety of studies reviewed, the author was not
able to use explicit criteria for quality assessment of the studies.
In addition to the WHO review, we also identified several primary studies that compared the
utility of various methods for monitoring AEs or medical errors. In 2007, Olsen and colleagues
compared the use of incident reporting, pharmacist surveillance, and local real-time record
review for the recognition of clinical risks associated with hospital inpatient care.37 Using the
three methods, they prospectively collected data on AEs on 288 patients discharged from an
850-bed general hospital in the National Health System in the UK. The study found little overlap
in the nature of events detected by the three methods. Record review detected 26 AEs and
40 potential AEs (PAEs) occurring during the index admission. Incident reporting detected
11 PAEs and no AEs. Pharmacy surveillance found 10 medication errors, all of which were
PAEs. The study concluded that incident reporting does not provide an adequate assessment of
clinical AEs and that a variety of methods need to be used to provide a full picture of the safety
condition in a health care organization.
In 2008, Wetzels and colleagues compared the validity and usefulness of five methods for
identifying AEs in general practice.38 The five methods included physician reported AEs,
pharmacist reported AEs, patients experiences of AEs, assessment of a random sample of
medical records, and assessment of all patients who died. In this prospective observational study,
a total of 68 events were identified using these methods. The patient survey identified the highest
number of events and the pharmacist reports identified the fewest. No overlap among the
methods was detected. The authors concluded that a mix of methods is needed to identify AEs in
general practice.
A study by Ferranti and colleagues compared results from two adverse drug event (ADE)
detection methodsvoluntary reporting and computerized surveillanceat a large academic
medical center.39 This 2008 study analyzed the medications most likely to cause harm and
evaluated the strengths and weaknesses of each detection system. During a 7-month period,
computerized surveillance detected 710 ADEs (6.93/1,000 patient days), whereas voluntary
reporting identified 205 ADEs (1.96/1,000 patient days). For each major drug category
(anticoagulants, hypoglycemia, narcotics and benzodiazepines, and miscellaneous), the two
methods detected significantly different event rates.39 Most surveillance-identified events were
hypoglycemia-related, whereas most voluntarily-reported events were in the miscellaneous
category. Of all unique ADEs (875), only 40 were common between the systems. The studys
findings underscored the synergistic nature of the two ADE detection approaches. Although
surveillance provides quantitative data to estimate the actual rate of ADEs, voluntary reporting
contributes qualitative evidence to prompt future surveillance rule development and identify
areas of emerging risk. The authors concluded that the two detection methods should be used
together to provide a full picture of ADE-related patient safety problems.
In 2010, Levtzion-Korach and colleagues published a study that examined and compared five
AE detection methods in one hospital.40 The methods included a Web-based voluntary incident
414
reporting system, medical malpractice claims, patient complaints, the hospital risk management
database, and executive walk rounds. These methods varied in the timing of the reporting
(retrospective or prospective), severity of the events, and profession of the reporters. The five
disparate data sources at the hospital captured about 15,000 problems. The authors systematically
classified the detected problems into 23 categories using a taxonomy that they developed. The
study found that each method identified important safety problems that were generally not
captured by any of the other methods.40 The following are the common categories of safety
problems detected using the five methods compared in the study:
Spontaneous reporting: patient identification issues, falls, and medication problems
Malpractice claims: issues with clinical judgment related to diagnosis and treatment,
communication, and technical skills and problems with medical records (incomplete,
illegible, or missing)
Patient complaints: issues with communication, ancillary services (e.g., patient transport,
kitchen, housekeeping), and administration (admission and discharge processes,
scheduling)
Risk management: issues with technical skills, patient and family behavior (compliance
issues, unusual behavior by a patient or family members), administration, and clinical
judgment
Executive walk rounds: problems with equipment, electronic medical records and other
such technologies, and infrastructure (work environment, security)
Communication problems were common among patient complaints and malpractice claims.
Clinical judgment problems were the leading category for malpractice claims. Walk rounds
identified issues with equipment and supplies. AE reporting systems highlighted identification
issues, especially mislabeled specimens. The authors concluded that, to obtain a comprehensive
picture of their patient safety problems and to develop priorities for improving safety, hospitals
should use a broad portfolio of approaches and then synthesize the messages from all individual
approaches into a collated and cohesive whole.
In another 2010 study, the Office of Inspector General of the Department of Health and
Human Services compared the usefulness of five safety event screening methods: nurse reviews,
analysis of present-on-admission (POA) indicators, Medicare beneficiary interviews, hospital
incident reports, and analysis of patient safety indicators.41 The study used a sample of 278
Medicare beneficiary hospitalizations selected from all Medicare discharges from acute care
hospitals in two selected counties during a 1-week period in August 2008. The investigators
compared events flagged by each screening method to the 120 events identified and/or confirmed
through physician reviews. The study found that nurse reviews and POA analysis identified the
greatest number of safety events. Nurse reviews identified 93 of the 120 confirmed safety events
and POA analysis identified 61 events. Beneficiary interviews identified 22 events, and the
remaining two screening methods identified 8 events each. Of the 120 events, 55 (46%) were
identified by only one screening method. Nurse reviews identified 35 events (29% of the 120
events) not flagged by any other screening method. POA analysis alone flagged 14 events (12%
of the 120 events).
We also reviewed a study by Tinoco and colleagues that compared a computerized
surveillance system (CSS) with manual chart review (MCR) for detecting inpatient ADEs and
hospital-associated infections (HAIs).25 The authors retrospectively analyzed the events detected
using the two methods by type of events. From a sample of 2,137 patient admissions between
415
October 2000 and December 2001, the authors identified AEs that were detected only by MCR,
only by CSS, or by both methods. The study found that CSS detected more HAIs than MCR
(92% vs. 34%); however, a similar number of ADEs was detected by both systems (52% vs.
51%). The agreement between systems was 26 percent and 3 percent for HAIs and ADEs
respectively. The study also found that MCR detected events missed by CSS using information
in physician narratives and that some events found by MCR were missed by CSS. The authors
concluded that integrating information from physician narratives with CSS using natural
language processing would improve the detection of ADEs more than HAIs.
A compelling theme emerged among the findings of the studies reviewed for this section.
That is, different methods for detecting patient safety problems overlap very little in the safety
problems they detect. These methods complement each other and should be used in combination
to provide a comprehensive safety picture of the health care organization. Detailed information
on the studies reviewed in this section (except for the WHO report14) is provided in Appendix D.
Because the body of evidence consists of studies of different designs, the overall strength of
evidence is not assessed.
416
Table 4, Chapter 36. Evidence-based rating of the main methods used in developed countries for estimating hazards in health care systems
Criteria
Effectiveness in
capturing the
extent of harm
Availability of
reliable data
Methods Based on
Reporting
Incident
External
Reporting
Audit and
Systems
Confidential
Inquiries
No evidence
available
4 for harm
assessment, to
be confirmed*
3, to be
confirmed*
No evidence
available
Electronic
Medical
Records
Administrative
Data
Autopsy
Reports
Mortality
and
Morbidity
Conferences
No evidence
available
2, to be
confirmed*
No evidence
available
No evidence
available
3, to be
confirme
d*
1
No evidence
available
No evidence
available
No
evidence
available
2
No evidence
available
Not
applicable
4, to be
confirmed*
No
evidence
available
No
evidence
available
No
evidence
available
No
evidence
available
No
evidence
available
Suitability for
large-scale
studies
Suitability for
small, repeated
studies
Costs
4, to be
confirmed*
1 for prospective;
2 for crosssectional
No evidence
available
No evidence
available
Effectiveness in
influencing policy
No evidence
available
No evidence
available
3, to be
confirmed*
3, to be
confirmed*
No
evidence
available
4
No evidence
available
Effectiveness in
No evidence
3, to be
No evidence
3, to be
3, to be
3
No evidence
influencing
available
confirmed*
available
confirmed*
confirmed*
available
hospital and local
safety
procedures and
outcomes
Synergy with
4
4
4
2
No evidence
3
4
4
other domains of
available
quality of care
Adapted from Michel P. Strengths and weaknesses of available methods for assessing the nature and scale of harm caused by the health system.14
Rating scale from 1 to 4, the most favorable level being 4
* to be confirmed where the amount of evidence-based data is small
417
3, to be
confirmed*
Table 5, Chapter 36. Subjective rating, where there was no evidence-based data, of the main methods used in developed countries for
estimating hazards in health care systems
Ad Hoc Studies Based on
Epidemiological Designs and
Systematic Data Collection
Review
Studies
Direct
of
Based on
Observation
Medical
Interviews
Records
With HealthCare
Providers
Incident
Reporting
Systems
External
Audit and
Confidential
Inquiries
Studies of
Claims and
Complaints
4 for harm
assessment,
2-3 for cause
analysis
2-3
2-3
1 for global
assessment,
3 for limited
focus
2-3
(videotaping)
Costs
Criteria
Effectiveness in
capturing the extent of
harm
Availability of reliable
data
Methods Based on
Reporting
Electronic
Medical
Records
Administrative
Data
Autopsy
Reports
Mortality and
Morbidity
Conferences
3-4
Not
applicable
1-2
3-4
Effectiveness in
*
3-4
3-4
*
*
*
3-4
influencing policy
Effectiveness in
*
2-3
3-4
*
1-2
*
*
influencing hospital
and local safety
procedures and
outcomes
Synergy with other
*
*
*
*
2-4
*
*
domains of quality of
care
Adapted from Michel P. Strengths and weaknesses of available methods for assessing the nature and scale of harm caused by the health system.14
Rating scale from 1 to 4, the most favorable level being 4
* Evidence-based rating is available (see Table 4)
418
3-4
the hospitals expenditures on these systems were estimated to be a one-time cost of $120,000
and an annual cost of almost $1 million. Additionally, we identified some general discussions
about which detection methods are generally more expensive or labor-intensive (Table 2 and
Table 3). Our search did not identify any full economic evaluation (e.g., cost-effectiveness
analysis from the publics perspective) of the burden related to the implementation of various
methods for detecting AEs or medical errors.
Table 6, Chapter 36. Estimated costs of systems for detecting patient safety problems in one
hospital
Incident
Reporting
Patient
Complaints
Risk
Management
Malpractice
Claims
Executive Walk
Round
One-time
expense
$72,400
$42,580
$0
Not directly
supported by the
institution
$0
Annual support
$9,000
$3,395
$0
Not directly
supported by the
institution
$0
0.5 FTE PS
manager:
$43,340
12 FTE PS
analyst:
$540,000
Not directly
supported by the
institution
0.2 FTE PS
manager:
$17,380
Software
Manpower
Annual support
0.2 FTE RM
analyst: $18,000
0.3 FTE PS
analyst: $12,780
0.1 FTE PS
analyst: $4,500
A weekly hour of
CEO, CMO,
CNO, and COO:
$10,500
Sum
One-time
expense
$72,400
$42,580
$0
Not directly
supported by the
institution
$0
Annual support
$74,840
$543,395
$318,500
Not directly
supported by the
institution
$40,660
Low
Evidence or
Potential for
Harmful
Unintended
Consequences
Negligible
Estimate of
Cost
High
421
Implementation Issues:
How Much do We
Know?/How Hard Is it?
Moderate/Difficult
References
1.
2.
3.
4.
5.
6.
7.
8.
422
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
27.
19.
28.
20.
29.
30.
31.
32.
33.
34.
21.
22.
23.
24.
25.
26.
423
35.
36.
37.
424
38.
39.
40.
41.
425
Postdischarge
interventions
Bridging
interventions
Bridging interventions included both pre- and post-discharge components, and often
emphasized longitudinal relationships in the pre- and the post-discharge periods, as well as the
role of the patient or caregiver in maintaining safe transitions.15
The purpose of this systematic review is to analyze published literature to determine the
effectiveness of the kinds of interventions described in Table 1 to reduce adverse events, ED or
unscheduled acute care visits, and readmissions after hospital discharge of adult patients, and to
assess the feasibility of implementing successful interventions on a larger scale. We included
randomized controlled trials (RCTs) and non-randomized clinical controlled trials (CCTs) that
evaluated one or more of the above interventions in adult general medical patient populations,
utilized at least one intervention prior to discharge, and reported rates of ED visits, readmissions,
426
or adverse events (AEs) after discharge. We included studies that reported costs if they also
reported one of the other targeted outcomes.
427
specific disease process. Unlike the Naylor review, this review did not consider interventions
that occurred after the patient was discharged.
While some aspects of transitional care interventions for disease-specific populations may
apply broadly to general medical populations, others may not be generalizable or may be less
effective. In CHF patients, for example, a clear link exists between dietary and medication
adherence and readmission risk; therefore, many successful interventions incorporate extensive
patient and caregiver counseling around diet, medication adherence, and weighing daily at home.
However, an elderly patient who is debilitated after a lengthy hospitalization for pneumonia may
not derive the same level of benefit from medication and dietary counseling, as would a younger
CHF patient, but might benefit from an intervention emphasizing restoring functional status and
close clinical follow-up.
As several recently published systematic reviews evaluated the role of transitional care
interventions in disease-specific populations and because the outcomes of such interventions
appear to be different in disease-specific and more undifferentiated patient populations, we chose
to focus our review on studies that evaluated only interventions conducted in adult general
medical populations. In contrast to another recent review15 that evaluated only studies of
interventions to reduce readmissions, we also included studies that sought to reduce adverse
events or ED utilization after discharge.
428
429
daily living, medical treatment, social support, and followup with a home visit and telephone
contact in the post-hospitalization period.28
These six studies share several similarities. Five studies were done in geriatric
populations.27,28,35,43,65,68 All had bridging interventions that included five or more separate
interventions, including a dedicated transitional provider across the continuum of care,
individualized personal health records, and post-hospitalization outreach to patients. All six
studies also involved patient contact at multiple points during and after hospitalization. These
interventions likely require a considerable amount of time, resources, and additional staff
(dedicated transitional provider) to facilitate the coordination of care from hospital to home.
Although the relative intensity of the interventions could not be measured directly, the
multifaceted nature of these interventions means they likely were more intensive than those
described in studies that did not find reduced readmission rates. The CTI is the only program
shown to reduce readmissions in multiple studies in different health care settings.27,43,65,68
430
were blinded to group allocationreview electronic medical records for unscheduled hospital
readmissions and ED visits to determine if they were drug related. Schnipper and colleagues
used a combination of structured screening via patient report by the Bates method70 and chart
review by blinded physician reviewers using the Naranjo algorithm to assess causality.71
431
other studies that demonstrated reductions in 30-day readmissions or ED visits were singlecenter studies that have not been replicated in other settings or patient populations.
One study73 evaluated the implementation of the CTI in ten California hospitals, using a
qualitative approach to identify key factors associated with successful implementation.
Leadership support and early engagement of hospital and community stakeholders were
identified as important steps in ensuring early implementation success; maintaining a cadre of
funded transition coaches was thought to be essential for ensuring CTI sustainability.
433
reducing AEs, ED visits, and readmissions after discharge, and determine what is known about
the influence of contextual and implementation factors on the success of these interventions.
Only a limited number of relatively high-intensity bridging interventions appear to reduce
readmissions and ED visits, and only one of these (the Care Transitions Intervention) has been
implemented in multiple contexts. Pharmacist-led interventions do appear successful at reducing
ADEs after discharge, but the overall literature base of interventions specifically targeting
common AEs after discharge is small. The studies we identified unfortunately provided little
information about implementation factors, contextual factors affecting the success of the
intervention, or costs of implementation. Such information will be needed to allow health care
system leaders and policymakers to plan strategically as they consider implementing programs to
prevent readmissions and other harms associated with transitions of care. A summary table is
located below (Table 1).
Table 1, Chapter 37. Summary table
Scope of the
Strength of
Problem Targeted by Evidence for
the PSP
Effectiveness
(Frequency/Severity) of the PSPs
Common/Moderate
Low
Evidence or
Potential for
Harmful
Unintended
Consequences
Negligible
Estimate of
Cost
Moderate-tohigh
Implementation Issues:
How Much do We
Know?/How Hard Is it?
Little/Difficult
References
1.
2.
3.
4.
5.
6.
434
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
435
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
43.
34.
44.
45.
35.
46.
47.
48.
49.
50.
51.
52.
36.
37.
38.
39.
40.
41.
42.
436
53.
54.
55.
56.
57.
58.
59.
60.
61.
62.
437
63.
64.
65.
66.
67.
68.
69.
70.
71.
72.
73.
438
439
communication), and specific simulation-based training protocols have been developed for
enhancing team performance. These protocols include Anesthesiology Crisis Resource
Management.13 Likewise, simulation may provide a way to assess the efficiency and safety of
system-level practices that may be difficult to control in real time or unsafe to test empirically.
Simulation has received increased recognition since the release of Making Health Care
Safer and seminal Institute of Medicine reports on preventable medical errors and mortality.14,15
The Agency for Healthcare Research and Quality (AHRQ) continues to fund a number of
projects dedicated to enhancing patient safety through simulation.16,17 Accrediting bodies from
the American Association of College of Nurses, American College of Surgeons, American
Association of Anesthesiologists, and Society for Simulation in Healthcare have all supported
simulation training centers.18-20
Studies evaluating the relationship between the benefits of simulation exercises and patient
safety outcomes, including potential harms, have not been thoroughly evaluated. Therefore, the
purpose of this chapter is to systematically review evidence on the benefits and harms of using
simulation to improve patient safety in medicine.
440
Simulation is particularly useful for research into processes that cannot be varied in a study
without undue harm to patients. Specific considerations in implementing simulation are detailed
in a later section of this chapter.
On a basic level, simulations improve patient safety by allowing physicians to become better
trained without putting patients at risk and, importantly, by providing a protected time for
reflection and debriefing. The challenge is matching the best simulation method and training
details for the desired learning objectives, while recognizing the costs of each method.24 Because
simulation is a broad technique, rather than a specific technology, faculty training and time are
often considered a more important investment than are specific expensive simulation equipment.
Given that most clinicians are not trained in simulation and debriefing, specific practitioners with
interest in the area must be appropriately trained to effectively use simulation techniques, as well
as any specific desired technologies, in order to accomplish the relevant training or systems
probing goals. The significant investment of time requires that faculty are provided time to
develop and teach meaningful simulations. Besides some procedural skills that may require
mainly repetitive practice, most simulations require an extensive debriefing component, during
which much of the learning takes place. Learning to effectively facilitate debriefings is often the
most time-intensive component of faculty training.26,27
The simulation itself needs to feel real enough for participants to be able to suspend
disbelief, acting and thinking much as they would in a similar but real scenario.25 If the learning
objective is mainly to practice cognitive skills for diagnosis or treatment, a verbal simulation
such as What would you do if may be sufficient. In contrast, if time pressure and/or team
communication are the focus, a more accurate replication of the actions and team presence
become important for the simulation experience.
441
casualty incidents),32 with rare clinical events, and with advancing to mastery levels in a
specialty.7,8
Second, simulation may be particularly valuable in duplicating complex high-stakes
scenarios that involve care teams and complex factors that affect performance.33,34 During an
Anesthesiology Crisis Resource Management simulation (including realistic emergency
scenarios, team communication, and complex decisionmaking), Blum and colleagues planted
probes of clinically pertinent information known only to a single member of the care team.35
These researchers found that participants shared only 27 percent of the probes with the team,
electing instead to share redundant information already available to others. Authority and power
differentials between senior and junior clinicians may also hinder effective communication, and
reluctance to challenge superiors may result in medical errors in high-stakes scenarios. The field
of aviation uses a method called the two-challenge rule to train all team members in a technique
known as advocacy-inquiry, giving junior team members the language to speak up while
seeking clarification. To study use of the two-challenge concept in debriefing anesthesiology
trainees, Pian-Smith and colleagues structured emergency simulations with contraindicated
decisions made by attending clinicians.36 The researchers discovered that trainees were initially
reluctant to challenge their superiors, but after debriefing, they were significantly more likely to
make an effective and clear challenge to contraindicated decisions by attending clinicians.
Third, errors are allowed in simulation and utilized as learning experiences through reflection
and debriefing. Training with real patients requires that supervising clinicians intervene if certain
errors occur, disallowing trainees to carry out the remainder of the procedure or to protect time
for debriefing. Additionally, as Fanning and colleagues note, reflective practices are considered a
cornerstone of life-long education in medicine.26 Simulation may facilitate highly accurate
measures of specific care-related behaviors and processes (technical or non-technical) for the
purpose of debriefing and reflection.26,27 Debriefing has exerted an increased performance effect
over self-study in high-fidelity simulated assessments.37
Fourth, simulation may be used to test new technologies, especially complex ones that
involve new learning curves in even the most experienced of clinicians.38 Likewise, new team
processes or innovative practices may be simulated prior to implementing in real time.39
Simulation may also help resolve disputes between best care practices. For example,
cardiopulmonary arrest of a woman pregnant for greater than 20 weeks requires cesarean
delivery within 5 minutes of onset, to protect both mother and fetus. However, debate has existed
as to whether the procedure should be performed in the labor room or in an operating room after
transporting the patient. Clearly, one would not consider placing actual patients in harms way
just for the purpose of settling this debate in research. However, utilizing simulation in a
randomized design, Lipman and colleagues determined that the average time to incision was
three and half minutes longer in teams instructed to transport patients to the operating room.40
Lastly, teams can simulate patient care flow in situ, for critical events41-43 as well as new
facility usual care, to check for the proper equipment (e.g., severe hemorrhage drills on various
hospital units or new operating rooms, respectively).44,45 Researchers have utilized simulationbased communication training to enhance inter-agency processes such as telephone referrals,46
but this use remains a relatively under-studied area.
442
studies in this arena. The measurable outcome may vary substantially based on the specific
behavior or process under investigation. For example, small gains may be expected on patient
outcomes from additional simulation exercises when assessing clinicians who must first
demonstrate a minimum acceptable level of competency before practicing on actual patients.
Likewise, comparing experienced to expert clinicians requires a highly sensitive and specific
methodology that is still being developed currently among simulation experts.7,8 A plethora of
simulation research has been conducted on what translational science refers to as T1, or
laboratory-only, outcomes.12,47 As noted above, T1 studies, or research about simulation,
support the logic for why simulation matters for patient safety. This section on beneficial effects
of simulation focuses on studies that reported T2 or T3 outcomes and that were published since
Making Health Care Safer, or those studies that translate interventions directly to patient-level
and systems-level outcomes, respectively.12,47 Studies are grouped into simulation exercises that
assessed patient outcomes related to practitioner technical performance, team-level, and systemlevel outcomes. Following these sections is an in-depth review of literature on patient outcomes
related to central venous catheterization.
Practitioner Performance
This section focuses on the technical aspects of physician performance during procedures.
Although technical skills are a crucial component to effective and safe health care, there is
potential for simulation to improve cognitive and other decision-making processes that impact
the delivery of services.
Diagnostic procedures. In a randomized trial with first-year gastroenterology fellows,
simulation-trained fellows out-performed traditionally-trained fellows during standardized
assessment of performance on the first 80 colonoscopy procedures (p = 0.03). The difference
between groups did not persist beyond 80 procedures, and the researchers determined that both
groups required the entire 200-colonoscopy training experience to achieve a desirable level of
procedure mastery.48 Another randomized study of first-year gastroenterology fellows reported
similar results, with simulation-trained clinicians successfully reaching the cecum in 38 percent
of their first 15 colonoscopies compared with 20 percent in the control group (p = 0.027).
Differences between groups were also observable through 30 procedures, but differences did not
persist beyond 30 procedures.49 Other studies have reported earlier acquisition and higher
performance in first colonoscopies performed among trainees who received simulation
training,50-52 and one reported that simulation-trained clinicians reached the cecum 4.5 times
more often on average than did clinicians in the control group during their first 10 procedures
(95% CI, 1.89 to 11.60, p = 0.001).50 Similar effects of simulation on increased early
performance enhancement, and, subsequently, absence of differences between study groups,
were reported for upper gastrointestinal endoscopy.53 In a randomized study, bronchoscopy
simulation exerted no effect on procedure time.54 Safety outcomes reported in these studies
focused primarily on patient discomfort (e.g., insufflation), and simulation training was
associated with lower discomfort in one study,50 no difference in another,48 and higher patient
discomfort in a third.49
In a before and after design, thoracenteses performed after implementing simulation in a
training curriculum involved fewer pneumothoraces (8.7% before vs. 1.1% after, p = 0.003) and
procedures advancing to thoracostomy (6% before vs. 0% after, p = 0.003) than those performed
prior to simulation-based training.55 In a similar type of research design examining
443
cordocentesis, procedure time was lower (6.4 min vs. 13.2 min, p < 0.001) and success rate was
higher (98.8% vs. 94.8%, p < 0.001) after implementing simulation training. There was no
difference between the before- and after-simulation study periods in procedure-related fetal loss
or overall fetal loss.56
Surgical procedures. A meta-analysis of laparoscopic training with virtual reality simulators
reported that procedure time was no faster, but was more accurate (standardized mean difference
[SMD] 0.68, 95% CI 0.05 1.31) in simulation-trained clinicians compared with traditional
video-trained clinicians. These authors reported no difference between simulation-trained and
other-trained clinicians in conversion rate to open surgery.57 Surgical residents who were
randomized to simulation training on laparoscopic cholecystectomy exhibited fewer errors in
exposing (control mean = 53.4, simulator mean = 15.0, p < 0.04), clipping (control mean = 7.1,
simulator mean = 1.9, p < 0.008), and dissection (control mean = 29.5, simulator mean = 11.5, p
< 0.03) during their first ten cholecystectomies. Three-fold fewer total errors and an eight-fold
decreased variation in error making totals were found among simulation-trained clinicians.58
Another study reported that warming up with laparoscopic simulation led to increased
observed structured assessment of technical skills (OSATS) global rating on actual
cholecystectomies in trainee- and experienced clinicians (warm-up mean = 28.50, control mean
= 19.25, p = 0.042), but no significant differences were reported on any one technical skill.59
Compared with no simulation training among surgery residents, simulation training for
extraperitoneal hernia repair was associated with larger increases in OSATS scores over a
baseline procedure for knowledge of procedure (p < 0.05), knowledge of instruments (p = 0.05),
and use of assistants (p < 0.05), but not global score.60 Researchers have utilized simulation for
other surgical modalities as well. In one study, urology residents trained with simulators
performed prostate resection faster (p = 0.025), and with higher performance scores on a
structured assessment (p = 0.021) than those not trained with simulation.61
Other procedures and processes. Pediatrics interns (n = 38) were randomized to simulation
training on basic procedural skills or training as usual for bag-mask ventilation, venipuncture,
peripheral venous catheter placement, and lumbar puncture. Interns in the simulation group
exhibited a higher, but not significantly higher, rate of successful venipuncture (p = 0.08). No
significant differences were observed between groups on the other procedural skills, although
performance scores on all measures were higher in the simulation group.62 Simulation has shown
additional benefits over traditional training in other areas of patient safety. Bachelors level
nursing students randomized to traditional pharmacology coursework or coursework with
additional simulation training were observed for medication administration errors in subsequent
external training placements. Students who trained with simulation made fewer medication
administration errors (7 of 31 total errors observed, p < 0.05), and these results were consistent
across both maternal health (control group = 8 errors, simulation group = 0 errors) and medicalsurgical settings (control group = 16 errors, simulation group = 7 errors).63 In another study,
paramedic students who received simulation training for endotracheal intubation performed
similarly to traditionally-trained students on their first 15 intubations for overall success rate,
success rate on first attempt, and in complications resulting from the intubation procedure.64
Although central venous cathether (CVC) placement is a specific procedure, the in-depth
review of literature on CVC is reviewed below. Again, the physician is not only a technician but
an actor that must balance being cognitively engaged with direct patient care as well as the
444
system within which she or he works. These systems include interpersonal dynamics such as
dyadic relationships (e.g., doctor-nurse pairs), larger team relationships (e.g., cardiac
resuscitation), and the care culture and environment (e.g., patient safety culture, and available
resources). Simulation has the potential to impact care processes and relationships at each of
these levels.
445
debriefing was provided to second-year internal medicine residents (n = 38) whose performance
on subsequent resuscitations was compared with third-year residents (n = 40) who did not
receive simulation training. Based on American Heart Association standards, simulation-trained
residents made 68 percent correct responses compared with 44 percent in the non-simulationtrained residents (mean difference = 44%, p < 0.001).41
Harms
Studies included in this review generally provided additive or supplemental interventions to
training as usual, and no study reported data indicating increased potential for or actual harm to
patients that resulted from implementing simulation techniques. However, it is conceivable that
simulation exercises would place demand on valuable resources that could be applied elsewhere
446
in patient safety efforts. Such considerations were not evaluated in literature captured for this
review.
447
Table 1, Chapter 38. Literature on simulation training for central venous catheterization
Study
Andreatta et
93
al. (2011)
Study Design
Randomized
controlled,
blinded, trial
Barsuk et al.
88
(2009)
Non-randomized
observational
cohort with
historical cohort
for controls
Barsuk et al.
76
(2009)
Pre-post, nonrandomized
observational
cohort
Barsuk et al.
73
(2009)
Observational
cohort study
Simulation
Participants
PGY-1 and PGY-2
radiology residents
(n = 32) performed
32 peripherally
inserted central
catheters
Minimum proficiency/mastery
model on standardized checklist of
CVC insertion with expert panel
review of performance using a
high-fidelity partial task trainer with
computer interface, simulation
exercises performed with deliberate
practice and individualized
feedback components
Minimum proficiency/mastery
model on standardized checklist of
CVC insertion with expert panel
review of performance using a
high-fidelity partial task trainer with
computer interface, simulation
exercises performed with deliberate
practice and individualized
feedback components
Minimum proficiency/mastery
model on standardized checklist of
CVC insertion with expert panel
review of performance using a
high-fidelity partial task trainer with
computer interface, simulation
exercises performed with deliberate
practice and individualized
feedback components
448
Table 1, Chapter 38. Literature on simulation training for central venous catheterization (continued)
Patient Safety Simulation
Practice
Minimum proficiency model with
deliberate practice and
individualized feedback from
attending surgeon on a partial-task
high-fidelity trainer, added to
standard lecture on CVC technique
Study
Britt et al.
89
(2009)
Study Design
Randomized
controlled trial
Evans
85
(2010)
Randomized
controlled,
blinded, trial
Khouli, et al.
90
(2011)
Randomized
controlled,
blinded, trial
Simulation
Participants
Junior surgical
residents (n = 34,
PGY-NR) that had
not completed a
trauma ICU rotation
performed 73
ultrasound-guided
and subclavian
CVCs
PGY-1 and PGY-2
residents from
emergency
medicine, internal
medicine, general
surgery, anesthesia,
and obstetricsgynecology (n =
115) performed a
total 495 internal
jugular, subclavian,
and femoral vein
CVCs
PGY-2 and PGY-3
internal medicine (n
= 47), and
subsequently
emergency residents
(n = 58 total) with
prior certification in
CVC placement,
number of CVCs
performed NR
449
Table 1, Chapter 38. Literature on simulation training for central venous catheterization (continued)
Patient Safety Simulation
Practice
Didactic training on anatomy and
CVC techniques coupled with
supervised practice on fresh
human cadavers, whereby
supervisors provided immediate
individualized feedback on
performance as well as video tape
review with debriefing
Simulation
Participants
Medical students
th
trained in their 4
year, performing
CVC during PGY-1,
number of total
CVCs performed by
health system
ranged from 1,682
1,884 annually
Internal medicine
residents (n = 150,
PGY-NR) performed
54 internal jugular
and subclavian
CVCs
Study
Martin et al.
(2003)
Study Design
Pre-post
observational
cohort design,
compared with
historical cohort
Miranda et
94
al. (2007)
Prospective
controlled cohort
design
Sherertz et
91
al. (2000)
Pre-post, nonrandomized
observational
design
3 year medical
students and PGY-1
physicians (n > 100,
actual number NR),
total CVCs
performed at system
during study period
5,099
Smith et al.
87
(2010)
Randomized
controlled trial
Internal medicine
residents PGY-1 and
PGY-2 (n = 52),
number of CVCs
performed NR
rd
450
Table 1, Chapter 38. Literature on simulation training for central venous catheterization (continued)
Study
Velmahos et
92
al. (2004)
Study Design
Randomized
controlled trial
Simulation
Participants
Surgical interns (n =
28), number of
CVCs performed NR
451
Implementing Simulation
Gaba conceptualized a framework that captures the versatility of simulation.24 In this
framework, each of 11 dimensions assist those looking to implement simulation to define their
needs and goals: (1) the purpose and aims of simulation activity; (2) the unit of participation in
the simulationindividuals or teams; (3) the experience level of simulation participants; (4) the
health care domain in which the simulation is applied; (5) the health care disciplines of personnel
participating in the simulation; (6) the type of knowledge, skill, attitudes, or behavior addressed
in simulation; (7) the age of the patient being simulated; (8) the technology applicable or
required for simulations; (9) the site of simulationin situ clinical setting versus dedicated
simulation center; (10) the extent of direct participation; and (11) the feedback method
accompanying simulation. Gaba further delineates implementation mechanisms for health care
settings as well as a variety of professional and governmental organizations.
In addition to the 11-dimensional framework by Gaba,24 which lays out implementable
aspects of types of simulation, a number of other factors must be considered before using
simulation:95
1. People: are both trainee and training participants available and appropriately trained in
simulation techniques?
2. Time: is sufficient time dedicated to meaningful simulations not only as adjunctive
training experiences provided in the flow of regular care activities?
3. Equipment: are simulation-specific materials (e.g., mannequin) and actual medical
equipment or devices available to recreate a desirable realism in simulated environments?
4. Space: is adequate space available for dedicating to simulated environments? Even in-situ
simulations will require storage and preparation of materials.
5. Supplies: if the simulation requires the use of real medical supplies, are these available?
6. Technical Support: implementation phases, especially of high-tech or complex
simulators, may require support, and there may be upkeep and maintenance to simulation
equipment.
People and time are likely to be the most expensive aspects of simulation in the long run.
However, startup costs vary substantially with the complexity of the simulation, which also
depends on the purpose. Large simulation centers may find financial support in philanthropic
sources or may be subsidized by participants and other organizational entities. Other expenses
will include the upkeep of simulation equipment and space dedicated to simulation.
Rosen and colleagues discussed the importance of considering components most likely to
enhance success in simulation techniques. For example, these authors differentiate the
importance of cognitive fidelity in a simulated exercise from physical fidelity. That is,
simulations that engage the participant in ways that, cognitively, most reflect the actual task are
likely to be more effective.25 Debriefing is considered crucial when implementing simulation,25,26
a recommended best practice according to one review of simulation-based medical education.96
Other national efforts have delineated future directions for simulation research.97 At an
international level, researchers have identified three primary foci to further understanding of best
practices in simulation: instructional design, outcomes measurement, and translational
activities.98
452
453
to assess practices that would be difficult or unsafe to study empirically in real-time with actual
patients. Likewise, simulation has been endorsed for ongoing competency and continuing
education, as well as advancing to mastery-level practices.
Previous systematic reviews have reported that simulation contributes to enhanced
knowledge acquisition and improved clinical performance.29 Simulation techniques have been
utilized in translating results from within simulation lab to patient and health care system-level
outcomes.100 Protecting time for debriefing in a learning experience has been suggested as a
crucial component to simulation techniques in a systematic review.101 This review is the first to
examine effects that simulation exercises exert on patient safety outcomes, and in particular,
outcomes with patients outside of simulation laboratory settings (i.e., during clinical care).
Reviewing the literature on simulation approaches broadly resulted in several limitations.
First, it is possible broad search strategies missed studies that may otherwise be captured with
targeted and comprehensive strategies dedicated to each simulation technique, clinical specialty,
or application. Second, given relative infancy of the research into simulation exercises, the field
may be prone to selective reporting of studies with positive findings, leading to potential
publication bias in our review. Third, we limited our assessment of quality of evidence to study
design and did not perform structured assessment of the strength of evidence. Therefore, overall
strength of the evidence for simulation exercises to improve patient safety should be interpreted
with caution based on the current review.
Simulation is a versatile technique that continues to garner momentum in a variety of clinical
settings and applications, including patient safety strategies. Although evidence is largely
heterogeneous at this time, our review suggests potential for simulation exercises to contribute to
patient safety through increased technical and procedural performance, and through improved
team performance. Limited research using health system-level observations suggests that
simulation may enhance patient safety, although more research is needed on the potential for
simulation to contribute to system-level differences in patient safety outcomes. Systematic
reviews of simulation for specific procedures have begun reporting patient safety outcomes,57,102
and more reviews of this nature would enhance our understanding of the overall contribution of
simulation techniques to patient safety. Future systematic reviews would benefit from
investigators using a consistent framework to describe the intervention, its context and its
implementation, such as Gabas framework.
Simulation implies a spirit of innovation in a world with quickly evolving technologies. As a
specialty field, it may be in an end of the beginning phase, and some of the benefits of
simulation may not be realized from a short-term perspective.24 As a developing field, certain
aspects may remain under discussion, such as standardized definitions of validity in simulated
environments.103 However, the breadth of applications and even purposes has not yet been fully
realized.24 As only one further example of simulations potential role in making care better and
safer, a recent systematic review of simulation and video games for patients post-stroke reported
increased upper extremity functionality.104 A summary table is located in Table 2.
454
Evidence or
Potential for
Harmful
Unintended
Consequences
Uncertain
Estimate of
Cost
Implementation Issues:
How Much do We
Know?/How Hard Is it?
Common/Moderate-to- Moderate to
Moderate
Moderate
high
high for
specific topics
*
Cost varies based on the simulation technique and resources involved. For example, exercises involving fully simulated
operating room environments may have higher costs to implement relative to those that require a simple mannequin or patient
actors.
References
1.
2.
3.
4.
5.
6.
7.
8.
455
9.
10.
11.
12.
13.
14.
15.
16.
17.
29.
18.
30.
31.
19.
20.
32.
21.
33.
34.
35.
36.
37.
22.
23.
24.
25.
26.
27.
28.
456
38.
39.
40.
41.
42.
43.
44.
45.
46.
457
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
458
67.
68.
69.
70.
71.
72.
73.
74.
75.
76.
77.
78.
79.
80.
81.
82.
83.
84.
85.
459
86.
87.
88.
89.
90.
91.
92.
93.
94.
95.
96.
97.
98.
99.
100.
460
101.
102.
103.
104.
Introduction
In health care, informed consent refers to the process whereby the patient and the health care
practitioner engage in a dialogue about a proposed medical treatments nature, consequences,
harms, benefits, risks, and alternatives.1 Informed consent is a fundamental principle of health
care.
The process of informed consent can be considered a patient safety issue from several
perspectives. At the extreme, performing a procedure on a patient without his or her consent has
been considered by the courts to be a form of battery.2 Informed consent may also be indirectly
related to patient safety in that, when done well, it opens a dialogue between the patient and
provider so that the patient can ask questions, knows what to expect during and after procedure,
and can at least theoretically help to avert medical errors.3
In general, studies have shown improved patient outcomes with effective physician-patient
communication and increased patient empowerment.4,5 Patient education has also been
associated with preventing medical errors.6
A review on this topic conducted in 2001 found few studies linking informed consent with
health outcomes and few studies on the impact of procedures used to obtain informed consent on
the quality of consent obtained; studies suggested that the value of informed consent might be
modestly enhanced by augmenting standard patient provider discussions with additional learning
and retention aids and that the process of consent can be modestly enhanced by using structured
interviews and asking patients to recall and restate the key elements of the discussion.
This update review focuses on what we have learned about the informed consent process and
the effectiveness of interventions that have been implemented to try to improve it. We conducted
a search of the health care and health services literature for the time interval 2001 to present and
reviewed all studies relevant to informed consent in the clinical setting.
461
U.S. hospitals, the mean reading level was 12.6 years and only 7% of the forms had an 8th grade
reading level or below.20 Regarding the content of informed consent documents, a separate
survey of a random sample of consent forms from 157 U.S. hospitals showed significant
variability in content and 74% omitted the nature of the procedure, risks, benefits, or
alternatives.21 Perhaps related to the readability of these documents, studies have shown patients
often do not read the consent forms provided to them and, in one study, patients who reported
reading the consent forms given to them were no better informed than those who did not. 22,23
Patients with limited English proficiency (LEP) are at particularly high risk for receiving
inadequate informed consent. In a study of 30 Latina women who were offered amniocentesis at
8 prenatal clinics without trained interpreters, the informed consent process contained all, or
nearly all, of the essential informed consent elements for 9% of the LEP compared with 68% of
the non-LEP women.24 When charts of 74 LEP Spanish and Chinese-speaking patients were
compared with those of 74 English-speakers, all of whom underwent thoracentesis, paracentesis,
or lumbar puncture at a teaching hospital where trained interpreters in Spanish and Chinese were
available, 28% of LEP patients had informed consent documented compared with 53% of
English speakers.25
463
after the informed consent process (median score on a knowledge test one point out six less at
both time periods, p<0.001).23 Among 54 patients who underwent head and neck surgery,
patients who recalled more than 50% of the complications they were told were, on average 7.6
years younger than those who recalled less than 50%.26 However, the association between older
age and less informed consent recall may be related to a lower average educational attainment
among older people. In a study of 200 patients with cancer who underwent informed consent for
radiation therapy, chemotherapy, or surgery, recall by age did not vary when adjusted for
educational attainment.22
Limited health literacy is also likely associated with less comprehension of informed consent.
Health literacy is the capacity to obtain, process, and understand basic health information and
services needed to make appropriate health decisions. In the 2003 National Assessment of Adult
Literacy (NAAL) 36% of Americans had basic or below basic health-related literacy.31 Older
age, lower educational attainment, and being African-American or Latino are all associated with
lower health literacy levels.32 In a study of men using a CD-ROM shared decisionmaking tool
about prostate cancer treatment options, lower levels of health literacy were associated with
lower prostate cancer knowledge after using the tool (Pearson correlation =0.65, p<0.001).33 In a
study of consent documents for a research study on chemotherapy agents, patients reading at or
below an 8th grade level had, on average, 28 percentage point lower comprehension scores
compared with participants with higher reading levels, even when the consent form was modified
to a 7th grade reading level.34 Similarly, in another study of a research consent form, modified to
a 6th grade reading level, patients with lower literacy were significantly less likely to respond
correctly to comprehension questions asked after a first reading of the consent form, adjusting
for other sociodemographic factors.35
Studies have also shown that minority race or ethnicity may be an independent risk factor for
having lower levels of comprehension in the informed consent process. Among 396 patients
being consented for surgery, African-American patients scored an adjusted average 9 percentage
points lower than white patients in a comprehension test administered immediately after the
consent process. This association was independent of education, age, and health literacy score.36
In the study of a research consent form cited above, in addition to those with lower health
literacy, patients who were Black or Asian/Pacific Islander were also less likely, when adjusting
for other factors, to respond correctly to comprehension questions.35
Other factors associated with lower informed consent recall are lower intelligence levels and
having cognitive dysfunction.23 Patients with cognitive dysfunction are particularly vulnerable in
the informed consent process. Cognitive dysfunction may be a long-term state (e.g., dementia) or
transient (e.g., after a medical procedure.) In one study of 302 acutely-ill medical inpatients, 48%
were estimated to have cognitive dysfunction such that they potentially lacked capacity to give
informed consent.37 However, not all patients with cognitive dysfunction lack informed consent
capacity: A structured assessment must be done to determine competence.38,39 If a patient does
not have capacity and his or her cognitive dysfunction is not expected to improve (or a decision
needs to be made prior to it improving), a surrogate decisionmaker must be established, except in
an emergency situation where the physician can determine the choice a reasonable person
would make.38
464
materials; using decision aids; using video educational tools; using interactive computer-based
educational tools; having structured discussions; and using repeat back methods.
465
language with an illustration, given to patients at the pre-operative visit, resulted in patients
having 25% higher knowledge scores on admission for the procedure (p=0.004) compared with
patients who received a structured verbal discussion.45 Finally, in a randomized-controlled trial
of information leaflets describing risks and benefits, sent by mail 2 weeks prior to surgery to
patients scheduled for elective orthopedic procedures, the group receiving the leafelets had a
median comprehension score 30 percentage points higher than those who did not (p<0.001).46
466
467
linked to patient satisfaction. Despite its importance, multiple studies have demonstrated that, in
practice, the informed consent process is often incomplete and patient recall and comprehension
of the discussion is usually low. Patients who are older, less educated, LEP, are of minority race,
or have cognitive dysfunction or low intelligence levels are particularly vulnerable in the
informed consent process.
Multiple methods have been proposed for improving the informed consent process. Studies
have shown that, in general, providing patients with simplified supplemental written materials,
using decision-aids, using video educational tools, and using the repeat back method improves
informed consent patient recall and comprehension. Studies using interactive computer programs
have had mixed results and further research is needed in this area. Studies using structured
informed consent discussion have also been limited. Studies of simplifying the informed consent
documents that have been done for research-related forms have shown mixed results on patient
recall and comprehension, but generally improve satisfaction; studies examining this effect
among informed consent documents in nonresearch clinical settings are lacking. A summary
table is located below (Table 1).
Table 1, Chapter 39. Summary table
Scope of the
Strength of
Problem Targeted by Evidence for
the PSP
Effectiveness
(Frequency/Severity) of the PSPs
Common/Moderate
Moderate
Evidence or
Potential for
Harmful
Unintended
Consequences
Negligible
Estimate of
Cost
Low
Implementation Issues:
How Much do We
Know?/How Hard Is it?
Moderate/Not difficult
References
1.
2.
3.
4.
5.
6.
468
7.
8.
9.
10.
11.
12.
23.
13.
24.
14.
25.
15.
26.
16.
27.
17.
28.
18.
29.
30.
31.
32.
19.
20.
21.
22.
469
33.
34.
43.
44.
35.
45.
36.
46.
37.
47.
48.
49.
50.
51.
52.
38.
39.
40.
41.
42.
470
53.
57.
54.
58.
59.
55.
56.
Solomon J, Schwegman-Melton K.
Structured teaching and patient
understanding of informed consent. Crit
Care Nurse. 1987;7(3):74-9.3665530.
60.
471
Introduction
Deficiencies in communication and teamwork have long been cited as a frequent contributor
to adverse events. Precise estimates of the extent of the problem are difficult to make, given
definitional issues as well as reporting and measurement problems. However, a variety of studies
support the notion that teamwork and communication are critical components of safe health care
systems. Previous reviews have reported linkages between various aspects of teamwork (e.g.,
situational monitoring, communication, leadership, trust, shared mental models) and clinical
performance.1-3 For example, observational studies in the surgical domain have shown increased
odds of complications and death (odds ratio 4.82; 95% confidence interval, 1.30 17.87) when
surgical teams exhibit less frequent teamwork behaviors (e.g., less information sharing during
intraoperative and handoff phases, and less briefing during handoffs).4 Reviews of malpractice
claims indicate that communication problems are major contributing factors in 24% of cases that
result in such claims.5 Other studies using root cause analysis to examine contributing factors
have found teamwork and communication issues cited as root causes in 52% to 70% of adverse
events.6,7 Additionally, teamwork and communication dimensions of safety culture have been
significantly related to adverse clinical events.8,9
The 2001 Making Health Care Safer report reviewed the topic of team-training in a review
entitled, Crew Resource Management and Its Application in Medicine. This review discussed
early conceptualizations of team-training in other high reliability industries such as aviation and
summarized early studies attempting to translate team-training principles developed elsewhere
into health care settings. The development and implementation of team-training programs has
grown dramatically in the last decade with improvements in the content and methods of
training.10 Additionally, there is over 30 years of evidence examining team performance
processes and the impact of team-training across a wide variety of highly complex, high-risk
work environments.11 This review provides an update on the implementation and effectiveness of
team-training in health care.
While there has been no previous comprehensive formal systematic review dedicated
uniquely to team-training in health care to date, a systematic review of interventions to improve
team effectiveness in health care found that the majority involved some form of team-training
(42 of 48 reviewed studies).10 Several systematic reviews with narrowly defined foci have
investigated the effectiveness of team-training for obstetric emergencies,12 for enhancing
communication in surgery,13 and classroom-based team-training interventions13,14 for example.
Additionally, several narrative reviews have investigated the content, design, and delivery of
team-training and the impact of team processes in health care.2,15,16 We draw on results from
these previous reviews to describe articles on interventions involving team-training.
What Is Team-Training?
Team-training is defined as a constellation of content (i.e., the specific knowledge, skills, and
attitudes that underlie targeted teamwork competencies), tools (i.e., team task analysis,
472
performance measures), and delivery methods (i.e., information, demonstration, and practice
based learning methods) that together form an instructional strategy.17 In this sense, teamtraining is a systematic methodology for optimizing the communication, coordination, and
collaboration of health care teams that combines specific content with opportunities for practice,
formative feedback, and tools to support transfer of training to the daily care environment.
As described in the National Quality Forums 34 Safe Practices for Better Healthcare in the
2010 Update, teamwork training and skill building is defined as follows:18
Healthcare organizations must establish a proactive, systematic,
organization-wide approach to developing team-based care through
teamwork training, skill building, and team-led performance
improvement interventions that reduce preventable harm to
patientstraining programs should systematically address and
apply the principles of effective team leadership, team formation
[and team processes]
Borrowing from other high reliability communities, the concept of team-training in health
care originated in the form of Crew Resource Management (CRM), a specific team-training
strategy focused on developing a sub-set of teamwork competencies generally related to hazard
identification, assertive communication, and collective management of available resources (e.g.,
people, tools, and information).19-21 However, the practice of team-training has become much
more broadly conceptualized in health care as the science dedicated to understanding team
processes, and performance has grown. Today, team-training is an overarching term that
encompasses a broad range of learning and development strategies, methods, and teamwork
competencies. The critical element is that the learning activity focuses on developing, refining,
and reinforcing knowledge, skills, or attitudes that underlie effective teamwork. This
differentiates team-training activities from technical or procedural learning activities that are
focused on developing technical clinical skills (e.g., cognitive skills such as differential diagnosis
and procedural skills).22 Prior narrative reviews of team-training interventions in health care have
found that the most commonly targeted teamwork competencies include communication,
situational awareness, leadership, role clarity, and coordination.13-16,23,24
473
session. Master Trainers then train administrators and frontline personnel within their own
organization using the customizable TeamSTEPPS curriculum. A slightly different approach was
utilized in the large-scale implementation of team-training throughout the Veterans
Administration (VA). As part of the VA National Center for Patient Safety Medical Team
Training (MTT) program learning sessions for participating VA medical centers were facilitated
directly by an interdisciplinary team (physician, nurse) of dedicated MTT faculty.35,36 Both
strategies, however, include local facility change teams, implementation of on-the-job tools (e.g.,
process checklists, scripts) to support training transfer, and measurement and evaluation
processes as integral implementation components.
474
decreases in equipment issues/case delays (from 24% to 7% of cases, p < .0001), decreased
handoff issues (from 5.4% to 0.3% of cases, p < .0001), and most notably a reduction in
mortality (p = .01).36,40,41 Additionally, a dose-response relationship was established such that for
each quarter the program was in place at a facility, a decrease of 0.5 deaths per 1000 procedures
(p = .001) was observed. Implementation of a related team-training program jointly developed by
the Agency for Healthcare Research and Quality and the Department of Defense,
TeamSTEPPS, has been associated with increased efficiency in clinical processes for multidisciplinary trauma teams (e.g., decreased times from arrival to surgery from 130.1 to 94.5
minutes (p < .05), endotracheal intubation from 10.1 to 6.6 minutes (n.s.), and CT scan from 26.4
to 22.1 minutes (p < .01)42) as well as an 83% reduction in medication and transfusion errors
(p < .001) and a 70% reduction in needlestick injuries and exposures (p < .05) in a U.S. Combat
Support Hospital deployed in Iraq.29 Other studies have also reported significant reductions in
clinical decision time (p < .05)43 associated with team-training, as well as one study showing a
reduction in adverse clinical events and a 50% reduction in high severity malpractice claims
(pre-training 11 high severity claims, post-training 5 high severity claims, no statistics
reported).44
Overall, the systematic review by Buljac-Smardizic10 concluded that the majority of studies
reviewed were of low to moderate level quality; however, eight of the reviewed team-training
studies were categorized as high or moderate quality (i.e., RCT or high quality pre-post study).
In the review by Rabl38 of classroom-based team-training interventions 15 of the 18 reviewed
studies were uncontrolled and 17 studies were rated at a moderate or high risk for bias.
476
professionals, from basic through continuing education, are needed to continue building this base
of evidence. A summary table is located in Table 1, Chapter 40.
Table 1, Chapter 40. Summary table
Scope of the
Strength of
Problem Targeted by Evidence for
the PSP
Effectiveness
(Frequency/Severity) of the PSPs
Common/High
Moderate
Evidence or
Potential for
Harmful
Unintended
Consequences
Low
Estimate of
Cost
Moderate
Implementation Issues:
How Much do We
Know?/How Hard Is it?
Moderate/Moderate-todifficult
References
1.
2.
3.
4.
5.
6.
7.
8.
477
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
478
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
479
Introduction
Adverse drug events are one of the most common types of harmful errors in both hospitalized
and ambulatory patients. Studies have shown that preventable adverse drug events occur in 7 to
10 of every 100 hospital admissions,1-3 and may even occur more frequently in the ambulatory
setting.4 Prescribing errors are likely responsible for at least half of these events.5,6
480
electronic health records (EHR) by 2015, and will include penalties for failing to achieve that
standard by 2016. The meaningful use criteria requires in part that EHRs must include one
clinical decision support rule applied to a specialty or high-priority condition, as well as the
ability to track compliance with that rule.
Given that only CPOE systems with an integrated CDSS meet the HITECH criteria for
meaningful use, this brief update review will assess the state of the evidence regarding the
effectiveness, cost, and implementation issues related to CPOE systems with CDSS capabilities
(CPOE+CDSS).
The 2001 Making Health Care Safer report reviewed evidence on the effectiveness of
CPOE+CDSS, as well as isolated CDSS, at improving medication safety.9 The review defined
level 1 outcomes as adverse drug events (ADEs), and level 2 and 3 outcomes as medication
errors and change in prescribing practices, respectively. These definitions were used in order to
distinguish the effects of CPOE and CDSS on clinical outcomes (e.g., preventable ADEs) and
surrogate outcomes that may not have caused patient harm (e.g., medication errors).
The 2001 review included four studies of CPOE+CDSS, three of which were conducted at
the same academic medical center. These studies all found improvement in level 2 and 3
outcomes, but did not document a reduction in preventable ADEs. All of the studies included in
the report evaluated homegrown, institution-specific systems (as opposed to commercial
system purchased from vendors) and often focused on safety of a specific medication or
medication class (such as antibiotics). These factors limit the generalizability of these studies to
general ADE prevention and to other institutions. The review also noted the high cost and
complex implementation issues that accompany CPOE+CDSS, stating, CPOE requires a very
large up-front investment with more remote, albeit substantial returns. In addition, CPOE affects
clinicians and workflow substantially. Its complexity requires close integration with multiple
systems, such as the laboratory and pharmacy systems. Failure to attend to the impact of such a
large-scale effort on organizational culture and dynamics may result in implementation failure
(page 71).
The overall conclusion of the review was that CPOE+CDSS can lower the rates of
medication errors and can promote appropriate prescribing, but evidence of its impact on actual
patient-level harm was limited. This conclusion proved to be somewhat controversial. In
response, followup commentaries10 took issue with the fact that CPOE+CDSS received only a
medium strength of evidence recommendation in the report. The objection to this conclusion
centered around the argument that CPOE+CDSS are difficult and costly to evaluate in controlled
trials, particularly when evaluating a relatively infrequent single adverse event such as an ADE,
and that the face validity of such systems indicated that proof of clinical benefit should not be
required before wider adoption. The evidence reports authors responded that using evidence to
evaluate the effectiveness and generalizability of these patient safety practices was essential to
their appropriate prioritization and application.11
481
482
Hospital in Boston). The conclusions regarding CPOE+CDSS in the 2001 edition of Making
Health Care Safer thus appear to stand largely unchanged a decade later.
483
related consequences, and how the transition from paper-based systems to an electronic system is
handled.
Some studies have successfully tailored alerts by incorporating patient-specific
characteristics into algorithms for displaying drug warnings. Seidling and colleagues24
implemented a tailored alert system at a German hospital and found a reduction in prescribing
errors; this study is notable because providers accepted nearly 25 percent of warnings, much
higher than rates generally reported in the literature. However, efforts to tailor drug warnings are
currently limited by the lack of standardized consensus definitions for drug-drug interactions that
are likely to lead to ADEs and unclear malpractice implications for users and manufacturers of
CDSS systems25 should patients be harmed if an alert is not provided. Recent commentaries25,26
have called for better guidance and legal protections to allow greater tailoring of alerts to
minimize alert fatigue and improve the safety performance of decision support systems, and a
recent consensus conference27 identified the key issues in developing more effective alert
mechanisms.
At the institutional level, it is clear that careful attention must be paid to the implementation
process of CPOE+CDSS, particularly with regard to how systems are integrated into existing
clinician workflow. Unfortunately, no clear consensus exists on the optimal implementation
methods in either the hospital or ambulatory setting. The CDSS five rights provides a
framework on implementation to improve medication management and outcomes by linking
each intervention with a specific objective. This framework includes each right be addressed to
ensure an optimal CDS program: right information, to the right person, in the right format,
through the right channel, at the right point in workflow.28 The Agency for Healthcare Research
and Quality has published the online Guide to reducing unintended consequences of electronic
health records (www.ucguide.org), and several case studies of implementation of commercial
CPOE+CDSS systems have also been published29-31 These reports likely provide the most useful
guides for decisionmakers regarding implementation issues.
We did not identify any formal cost-effectiveness analyses of CPOE+CDSS published in the
past 5 years. Individual institutions with homegrown CPOE+CDSS systems have estimated
considerable cost savings32 due to ADE prevention and optimizing medication use, but these data
may not be generalizable to other settings and systems. A 2009 review of the costs and benefits
of health information technology33 found a paucity of meaningful data on the cost-benefit
calculation of actual IT implementation, and concluded, although there is some empirical
evidence to support the positive economic value of an EHR system and the component parts of
EHRs, projections of large cost savings assume levels of health IT adoption and interoperability
that we are nowhere near achieving.
484
Low-tomoderate
Evidence or
Potential for
Harmful
Unintended
Consequences
Low-to-moderate
Estimate of
Cost
High
Implementation Issues
How Much do We
Know?/How Hard Is it?
Moderate/Difficult
References
1.
2.
3.
4.
5.
6.
7.
485
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
486
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
487
Clinical Staff1,7
Trace all lines back to their point of origin to verify that correct connections are made
Recheck connections and trace all lines to their point of origin after the patients arrival
to a new care area or as part of a handoff process
Do not force connections
Only use adapters in accordance with hospital policy for a specific indication
Label certain high-risk catheters as to the type of catheter (e.g., epidural, intrathecal)
Route lines with different purposes in unique and standardized directions (e.g., IV line
towards patients head, enteral feeding line towards patients feet)
Identify and manage conditions that may contribute to worker fatigue, which could result
in inattentiveness when making connections
Non-Clinical Staff1,7
The risk reduction strategies suggested for enteral feeding misconnections for clinical and
non-clinical staff include the following.
Clinical Staff1,7
Do not use standard luer syringes for oral medications or enteral feedings
Do not modify or adapt IV or enteral feeding devices
Route lines with different purposes in unique and standardized directions (e.g., route IV
lines towards patients head, route enteral feeding lines towards patients feet)
Identify and manage conditions that may contribute to worker fatigue
Review identification labels before administering solutions to ensure that the intended
delivery route is correct
488
Placing labels with warnings - WARNING for Enteral Use Only Not for IV Use
Non-Clinical Staff1,7
Ensure that an adequate number of distinctly labeled enteral pumps are purchased to
reduce or eliminate the use of infusion pumps for enteral administration to adult patients
Reinforce existing purchasing policies that mandate purchasing only enteral feeding sets
that are incompatible with female luer connectors
When possible purchase only non-IV compatible enteral feeding containers
Secure enteral administration sets with enteral feeding containers (e.g., with rubber band)
or pre-attached sets from the manufacturer before sending them to the patient care unit
Perform pre-purchase evaluations of enteral feeding systems under the guidance of a
multidisciplinary task force before purchasing decisions are made
According to ECRI Institute, the single most important work practice solution for clinicians
is to trace all lines back to their origin before connecting or disconnecting any devices and
infusions.8 Additional strategies that may be useful include ensuring proper lighting when
making connections, contacting manufacturers to determine if luer fittings can be replaced with
different connector types, storing medications for different delivery routes in different locations,
and using a color-code system.1
489
490
spinal drug errors.11 The clinicians were also concerned with color-coding since visual
discrimination is not possible in poor lightning conditions, non-translucent devices preventing
the ability to see what a needle is doing and if it has reached the right place during injections, and
drug leakage.11 Overall, if the identified design issues are resolved, clinicians would be willing to
adopt the non-luer devices because they believe patient safety will benefit from
implementation.11
In 2004, the Columbus Childrens Hospital conducted a Health Failure Mode and Effects
Analysis (HFMEA) to identify the inherent risks of use and labeling of various enteral,
parenteral, and other tubing types in patient care and the potential for harm.10 Woods and Shultz
(2006) found three common themes causing all failure modes: non-standardized labeling of
tubing, lack of knowledge of nomenclature or alias, and inconsistent inventory.10 Labeling of
tubing with infused mechanisms happened 85% of the time in the pediatric intensive care unit,
53% of the time during surgery, and 93% of the time during interventional radiology. The risk
reduction methods subsequently implemented involved the standardization of the labeling
process throughout the organization (e.g., color of labels, content on label, size and placement of
tubing), and the development of an online pictorial catalog listing all available supplies by
category and the nomenclature used by materials management and common names used by
clinical staff.10 The third risk reduction method conducted an inventory with the help from
clinical staff and materials management to identify currently used tubing, connectors that fit
properly and those that needed to be removed from practice, and devices that needed to be
purchased. According to the authors, several recommendations have been implemented and no
tubing misconnections have been reported.10
Another point of consideration when implementing the PSP for tubing misconnections is
cost. In 2008, Peter Angood, the chief patient safety officer and vice president of The Joint
Commission stated that the cost of acquiring new devices, identifying risky connections and
practices, and implementing training and testing will no doubt impact hospitals.12 However,
the cost of not making such changes could, of course, be much greater in terms of lives lost
erroneous connections between tubes and catheters can create catastrophic outcomes, even
death says Angood.12
491
cognitive, and organizational issues of devices. The importance of designing devices that help
reduce human error is discussed further in this chapter.
As seen in the previously mentioned studies, each facility will have specific needs and
inherent risks that require the implementation of different risk reduction strategies. Again, one of
the most important work practice solutions involves the tracing of lines back to the point of
origin. Regardless of the differences between facilities, it is recommended that facilities perform
a risk assessment to determine their ultimate needs in equipment changes and work practice
policy updates. A summary table is located below (Table 1).
Table 1, Chapter 42. Summary table
Scope of the
Strength of
Problem Targeted by Evidence for
the PSP
Effectiveness
(Frequency/Severity) of the PSPs
Common/Moderate
Low
Evidence or
Potential for
Harmful
Unintended
Consequences
Low
Estimate of
Cost
Low
Implementation Issues:
How Much do We
Know?/How Hard Is it?
Moderate/Not Difficult
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
492
Introduction
Long and unpredictable work hours have been a staple of medical training for centuries, but
the effects of fatigue among residents on patient safety garnered little attention until March 1984,
when a young woman died at a teaching hospital in New York. Her death was attributed in part
to a medication prescribing error made by residents in the midst of a 36-hour shift. This seminal
event led to the passage of regulations in the State of New York limiting residents shift duration
to 24 consecutive hours and overall work week to 80 hours. However, at the time of the original
Making Health Care Safer1 report in 2001, widespread violations of the New York regulations
were common, and it was not unusual for residents still to work 36 hour shifts and over 100
hours per week in other states as well).
A considerable body of evidence from health care and other industries2 links acute and
chronic sleep deprivation to impaired cognitive performance. Some studies have also shown that
sleep deprivation can affect psychomotor skills.3 Working extended duration shifts can be
harmful for both clinicians and patients. Studies have shown that residents who work more than
16 consecutive hours have an increased risk of motor vehicle accidents after their shift4 and of
suffering a needlestick injury during their shift.5 Working over 16 consecutive hours in the
intensive care unit has been shown to result in interns committing more diagnostic and
therapeutic errors.6 Among nurses, shift duration of greater than 12 hours is also associated with
a significantly increased risk of committing errors.7 In response to these and other data, the
Accreditation Council for Graduate Medical Education (ACGME) implemented formal work
hour restrictions for resident physicians in 2003, and made these regulations even more stringent
in 2011.8
Despite these known risks, the extent to which patients are harmed by clinician fatigue is
difficult to determine. Fatigue on the part of an individual provider may only be one of several
latent causes of a preventable adverse event, especially in the complex hospital environment.
Few studies have attempted to directly address the connection between clinician fatigue and
adverse clinical outcomes. Two recent studies9,10 examined whether attending surgeons fatigue
was linked to an increased risk of complications, and reached conflicting results; one study10
found an increased risk of complications when the surgeon had the opportunity to sleep for less
than 6 hours the night prior to the procedure, but the other9 did not find the same association.
Despite the lack of hard data linking fatigue and complications, the traditional residency
work hours that existed prior to 2003 could not be justified from an educational, humanistic, or
patient safety standpoint. The implementation of regulations to reduce residents work hours
have resulted in fundamental changes to residency education over the past decade, and the
regulations effect on patient safety has been extensively studied.
The 2001 report reviewed the evidence linking sleep deprivation and fatigue to medical
errors, and reached four conclusions:
Sleep deprivation and disturbances of circadian rhythm lead to fatigue, decreased
alertness, and poor performance on standardized testing.
493
Although data from non-medical fields suggest that sleep deprivation leads to poor job
performance, this link has not yet been established in medicine.
Forward rather than backward shift rotation [i.e., progressing from day to evening to
night shifts, rather than the reverse], education about good sleep hygiene, and strategic
napping before or during shifts may reduce fatigue and improve performance. High face
validity, low likelihood of harm, and ease of implementation make these promising
strategies, although more evidence of their effectiveness in medicine is warranted.
Given that medical personnel, like all human beings, probably function suboptimally
when fatigued, efforts to reduce fatigue and sleepiness should be undertaken, and the
burden of proof should be in the hands of the advocates of the current system to
demonstrate that it is safe.
In this review, we assess the evidence that has accumulated since 2001 for the effect of
limiting individual providers hours of service on patient safety outcomes. Although other
countries have significantly more stringent regulations (for example, trainees in the European
Union are limited to 48 hours per week), the focus of this review will be on studies conducted in
the United States.
494
The revised resident duty hour regulations published by the ACGME in 2010, and
implemented July 1, 2011, did not incorporate all of the IOMs recommendations. The 2011
regulations have four key components:
Continued 80-hour work week limit
No more than 16 consecutive hours on duty for first-year residents only. Second-year and
more senior residents can work 24 hours on duty, with an additional 4 hours allowed for
transitions of care
No more than 6 consecutive night shifts
Continued minimum of 4 days off per month and on-call frequency of no more than once
every 3rd night
Although excessive work hours are linked to errors among nurses, regulation of nurses work
hours is less uniform. Currently, 16 states do restrict mandatory overtime for nurses, but many
nurses still routinely work more than 12 hours per shift.7 There are also no regulations on
working hours for practicing physicians, despite some data indicating that many practicing
physicians work schedules that would be prohibited were they still residents.12
495
did not change.18 Burnout and depression are themselves linked to impaired job performance,
independent of acute or chronic fatigue.19
The 2003 duty hour regulations still allowed all residents to work a maximum of 30
consecutive hours. These extended duration shifts are still longer than those allowed in virtually
any other industry, and studies have found an association between working more than 16
consecutive hours and an increased risk of self-reported errors and attentional failures.20 A 2010
systematic review21 that identified 13 studies in which shift length for clinicians was
purposefully reduced found consistent evidence among the higher-quality studies that both
objectively-measured and self-reported errors decreased after shift length reduction. One
particularly high quality study6 found a significant reduction in serious medical errors for
medical interns assigned to work a 16-hour shift in the intensive care unit, compared with interns
working a traditional 30-plus hour shift. However, the reviewers were unable to reach a firm
conclusion regarding the optimal shift length, due to heterogeneity between shift lengths used in
the primary literature.
Thus, the totality of the evidence on the 2003 ACGME duty hour regulations indicates that
reducing resident duty hours does not improveor worsenspatient safety or mortality. The
association between extended duration (>16 hour) shifts and adverse events ultimately was a
factor in the ACGMEs decision to enact a 16-hour shift length limit for first-year residents as
part of the 2011 regulations.
496
replace the housestaff workforce with alternative providers and the need to provide greater
supervision for residents by senior physicians is likely to be considerable.
Low
Evidence or
Potential for
Harmful
Unintended
Consequences
Moderate (at
least)
Estimate of
Cost
High
Implementation Issues:
How Much do We
Know?/How Hard Is it?
Moderate/Difficult
Includes lack of
training time
References
1.
2.
3.
4.
5.
497
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
498
Part 3. Discussion
Chapter 44. Discussion
Introduction
Progress Since the 2001 Report
Over 2000 years ago, Hippocrates reminded physicians to, first, do no harm. In 1863,
Florence Nightingale wrote, It may seem a strange principle to enunciate as the very first
requirement in a hospital that it should do the sick no harm. Notwithstanding these
commonsensical admonitions, it was not until the turn of this century that a systematic effort to
improve patient safety began, catalyzed by the publication of the IOM report, To Err is
Human.1
The year following the publication of the IOM report, AHRQ commissioned a group of
investigators, led by the UCSF-Stanford EPC, to synthesize the worlds literature on PSPs, an
effort that culminated in the 2001 report, Making Health Care Safer.2 This report was widely
used by clinicians, safety workers, researchers, and policymakers, and it informed a variety of
other initiatives including the National Quality Forums Safe Practices list.
Since 2001, research in the patient safety field has exploded, with literally thousands of
published studies. In fact, some of todays popular safety practicesrapid response teams,
disclosure of errors to patients, or any of the checklistbased interventionshad barely been
invented at the time of the 2001 report.
In light of this maturation of the field, AHRQ asked a group of investigators, many of whom
were involved in producing the earlier report, to synthesize the vast amount of new information
on PSPs that has emerged since the release of Making Health Care Safer. Using a similar
method, which combined explicit criteria, detailed evidence reviews, and an international panel
of expert advisors, this report reviewed the evidence on 41 PSPs. In Table 1, below, we provide a
summary of this evidence, followed by a discussion of the evidence in the context of prior work
on patient safety, and then present priorities for adoption of PSPs
Table 1, Chapter 44. Summary table*
Patient Safety Practice
Scope of the
Problem Targeted
by the PSP
(Frequency/
Severity)
Practices Designed for a Specific Patient Safety Target
Adverse Drug Events
High-alert drugs: patient safety
Common/Moderate
practices for intravenous
anticoagulants;
in-depth review
Use of clinical pharmacists to
Common/Low
prevent adverse drug events;
brief review
The Joint Commissions Do Not
Common/Low
Use list; brief review
Smart infusion pumps; brief
Common/Low
review
Strength of
Evidence for
Effectiveness
of the PSPs
Evidence or
Potential for
Harmful
Unintended
Consequences
Estimate of
Cost
Implementation
Issues:
How Much do We
Know?/How Hard Is
it?
Low
Low-to-moderate
Low
Little/Moderate
Moderate-tohigh
Low
High
Little/Moderate
Low
Negligible
Low
Low
Low
Moderate
Little/Probably not
difficult
Moderate/Moderate
499
Infection Control
Barrier precautions, patient
isolation, and routine surveillance
for the prevention of healthcareassociated infections; brief
review
Interventions to improve hand
hygiene compliance; brief review
Reducing unnecessary urinary
catheter use and other strategies
to prevent catheter-associated
urinary tract infections; brief
review
Prevention of central lineassociated bloodstream
infections; brief review
Scope of the
Problem Targeted
by the PSP
(Frequency/
Severity)
Strength of
Evidence for
Effectiveness
of the PSPs
Evidence or
Potential for
Harmful
Unintended
Consequences
Estimate of
Cost
Implementation
Issues:
How Much do We
Know?/How Hard Is
it?
Common/Moderate
Moderate
Moderate
(isolation of
patients)
Moderate-tohigh
Moderate/Moderate
Common/Moderate
Low
Low
Low
Moderate/Moderate
Common/Moderate
Moderate-tohigh
Low
Low
Moderate/Moderate
Common/Moderate
Moderate-tohigh
Low
Low-tomoderate
Ventilator-associated
Common/High
pneumonia; brief review
Interventions to allow the reuse
Common/Low
of single use devices; brief
review
Surgery, Anesthesia, and Perioperative Medicine
Preoperative checklists and
Common/Moderate
anesthesia checklists to prevent
a number of operative safety
events, such as surgical site
infections and wrong site
surgeries; in-depth review
The use of ACS-NSQIP report
Common/High
cards and outcome
measurements to decrease
perioperative morbidity and
mortality; in-depth review
New interventions to prevent
Rare/Low
surgical items from being left
inside a patient; brief review
Moderate-tohigh
Low
Low
Low
Low-tomoderate
Low
Moderate-to-difficult/
Not difficult
(implementation of a
bundle)-to-moderate
(understanding
organization culture
and context)
Moderate/Moderate
High
Negligible
Low
A lot/Moderate
Moderate-tohigh
Low
Moderate
Moderate/Moderate
Low
Negligible
Little
Low
Negligible
Low if it
simply
involves
more
frequent
manual
counting;
high if RFID
is used
Moderate
Common/Low-tohigh
500
A lot/Not difficult
Moderate/Moderate
Scope of the
Problem Targeted
by the PSP
(Frequency/
Severity)
Common/High
Strength of
Evidence for
Effectiveness
of the PSPs
High evidence
harms may
equal or
exceed
benefits
High
High
Evidence or
Potential for
Harmful
Unintended
Consequences
High (death,
stroke,
hypotension, and
bradycardia)
Estimate of
Cost
Low
Implementation
Issues:
How Much do We
Know?/How Hard Is
it?
NA
Negligible
Low-tomoderate
A lot/Moderate
Moderate
Moderate/Moderate
Moderate
Moderate/Moderate
Common/Low
Moderate
Moderate
(increased use of
restraints and/or
sedation)
Low
Common/Moderate
Moderate
Negligible
Moderate
Moderate/Moderate
Common/Moderate
Moderate-tohigh evidence
it doesnt help
High
(hypoglycemia)
Low-tomoderate
NA
Common/Low
Low
Negligible
Low
Little/Not difficult
Common/High
Moderate
Low
Moderate
Moderate/Moderate
Common/Low
Moderate
Low
Moderate
Moderate/Moderate
Rare/High
Low
Low
Moderate
Little/Moderate
Rare/Moderate
Moderate
Moderate
(pneumonia)
Moderate
Little/Not difficult
Common/Moderate
High
Moderate
(bleeding)
Low
Little/Moderate
Rare/High
Moderate
Negligible
Low
Moderate/Not difficult
501
Scope of the
Problem Targeted
by the PSP
(Frequency/
Severity)
Common/Moderate
Strength of
Evidence for
Effectiveness
of the PSPs
Ensuring documentation of
Moderate
patient preferences for lifesustaining treatment, such as
advanced directives; brief review
Increasing nurse-to-patient
Common/High
Moderate
staffing ratios to prevent death;
in-depth review
Practices Designed To Improve Overall System/Multiple Targets
Increasing nurse-to-patient staff
Common/High
Low
ratios to prevent falls, pressure
ulcers, and other nursing
sensitive outcomes (other than
mortality); in-depth review
Incorporation of human factors
Not assessed
Potentially
and ergonomics in the design of
systematically,
applicable to all
health care practices by hiring an patient safety
but moderateexpert or training clinicians in
to-high
problems
human factors; in-depth review
evidence for
some specific
applications
Promoting engagement by
Common
Emerging
patients and families to reduce
practice (few
adverse events (such as patients
studies
encouraging providers to wash
available)
their hands); in-depth review
Interventions to promote a
Common/Low-toLow
culture of safety; in-depth review
high
Evidence or
Potential for
Harmful
Unintended
Consequences
Low
Estimate of
Cost
Low
Implementation
Issues:
How Much do We
Know?/How Hard Is
it?
Moderate/Moderate
Low
High
A lot/Not difficult
Low
High
A lot/Not difficult
Negligible
Moderate
A lot/Moderate
Uncertain
Low
Little/Moderate
Uncertain
Emerging
practice (few
studies
available)
Low
Uncertain
Lowtomoderate
(varies)
Varies
Negligible
High
Moderate/Difficult
Low
Negligible
Moderate-tohigh
Little/Difficult
Uncertain
Moderate
Moderate
Common/Moderate
Moderate-tohigh for
specific topics
Moderate
Negligible
Low
Moderate/Not difficult
Common/High
Moderate
Low
Moderate
Common/Moderate
Low-tomoderate
Low-to-moderate
High
Moderate/Moderateto-difficult
Moderate/Difficult
Common/Moderate
Low
Low
Low
Moderate/Not difficult
Common/High
Common/Low-tohigh
Common/Moderate
Common/Moderateto-high
502
Scope of the
Problem Targeted
by the PSP
(Frequency/
Severity)
Common/Moderate
Strength of
Evidence for
Effectiveness
of the PSPs
Evidence or
Estimate of
Implementation
Potential for
Cost
Issues:
How Much do We
Harmful
Know?/How Hard Is
Unintended
Consequences
it?
Limiting trainee work hours; brief
Low
Moderate (at
High
Moderate/Difficult
review
least); includes
lack of training
time
Abbreviations: ACS NSQIP=American College of Surgeons National Surgical Quality Improvement Program; NA = not
available; PSP: Patient Safety Practice; RFID = radio-frequency identification.
*In some cases, the text in the PSP column differs slightly from the chapter heading for that PSP. This difference is attributable
to our Technical Expert Panels desire to include the target safety problem (if the practice is in fact targeted at a specific safety
problem), more specification, or an example of the PSP (e.g., adding such as a centralized display of consolidated data to the
PSP designated as operating room integration and display systems).
Rating Scales:
Scope of the problem targeted by the PSP (frequency/severity): frequency = rare or common; severity = low, moderate, or high.
Strength of evidence for effectiveness of the PSPs: low, moderate, or high.
Evidence or potential for harmful unintended consequences: negligible, low, moderate, or high.
Estimate of cost: low, moderate, or high.
Implementation issues: How much do we know? = little, moderate, or a lot; How hard is it? = not difficult, moderate, or difficult.
One of the great challenges in measuring patient safety is determining whether to assess
primary outcomes (harms), intermediate outcomes (errors) or processes (such as adherence
to evidence-based safety practices). Each of these methods has advantages and disadvantages.
Over the past few years, the safety field has increasingly emphasized primary outcomes (namely,
harm measures), and the IHI Global Trigger Tool (GTT) has emerged as an increasingly popular
method for such assessment. In fact, several studies using the GTT3-5 have come to the same
disappointing conclusion: that rates of harm remain high and, at least in a group of North
Carolina hospitals, did not improve during the first several years of the patient safety movement.
Although the Global Trigger Tool has demonstrated better test characteristics than other
outcomes-oriented methods of measuring safety, such as voluntary incident reports and the
AHRQ Patient Safety Indicators,3 one of the main insights to have emerged from recent patient
safety research is that multiple lenses are needed to get a broad, and true, view of progress in
safety. Shojania has called this issue the elephant of patient safety, in that one gets a different
view depending on what part one is looking at.6
Because of the limitations of outcome measures in patient safety, it is important that we
continue to assess the degree to which we now understand and have implemented effective PSPs.
The present report, conducted by many of the same investigators, as 2001s Making Health
Care Safer illustrates both the progress and the challenges in this area of safety research.
Over the past decade, we have achieved greater agreement on what constitutes evidence of
effectiveness and the importance of implementation and context (this new understanding was
codified in a prior AHRQ report, Assessing the Evidence for Context-Sensitive Effectiveness
and Safety of Patient Safety Practices: Developing Criteria7 and in the peer-reviewed articles
that drew on this report).8-12 In the current review, 20, or about half of the PSPs reviewed, had
the strength of evidence for their effectiveness rated as at least moderate, which represents
significant progress since 2001. The evidence base supporting implementation strategies is also
improving. For 26 of the PSPs reviewed in the present report, we judged that there was at least
moderate evidence about how to implement the practice; the area of implementation was so
underdeveloped a decade ago that Making Health Care Safer did not even consider it.
503
However, for almost no PSPs do we understand with confidence the potential role that context
plays in effectiveness. This area remains a major gap in our knowledge base about how to select
and implement PSPs; and it is a particularly crucial gap as institutions and individuals try to
implement best practices, and policymakers, accreditors, and payers seek to create incentives
for implementation via transparency- or payment-related initiatives.
Preoperative checklists and anesthesia checklists to prevent operative and post-operative events
Bundles that include checklists to prevent central line-associated bloodstream infections
Interventions to reduce urinary catheter use, including catheter reminders, stop orders, or nurse-initiated
removal protocols
Bundles that include head-of-bed elevation, sedation vacations, oral care with chlorhexidine, and
subglottic-suctioning endotracheal tubes to prevent ventilator-associated pneumonia
Hand hygiene
Do Not Use list for hazardous abbreviations
Multicomponent interventions to reduce pressure ulcers
Barrier precautions to prevent healthcare-associated infections
Use of real-time ultrasound for central line placement
Interventions to improve prophylaxis for venous thromboembolisms
The conclusions in this report explicitly represent a combination of the available evidence
with the judgment of our technical expert panelists interpreting that evidence.
Additional PSPs were judged by our technical experts as having sufficient evidence about
effectiveness and implementation that they should be encouraged for adoption. Table 3
presents the encouraged PSPs.
Table 3, Chapter 44. Encouraged patient safety practices
504
The 22 PSPs in Tables 2 and 3 represent practices that health care providers can consider for
adoption now. This recommendation particularly applies to the 10 strongly encouraged
practices in Table 2, which, at least in the judgment of our technical experts, providers have
sufficient knowledge to implement and doing so will likely result in safer care. And while future
evaluations will probably strengthen our knowledge base regarding how best to implement these
practices to make them most effective, our technical experts believe that providers should not
delay consideration of adopting these practices while waiting for more research: enough is
known now to permit health care systems to move ahead.
Limitations
Because of limited resources and time, the current report does not cover the entire patient
safety field, which has grown exponentially since the last report, both in the number of potential
PSPs and in the amount of data about individual PSPs). We used an explicit and transparent
process to select the PSPs we did evaluate, and our final list should include most PSPs of highest
priority to policymakers and providers.
Secondly, we did not do in-depth reviews of all the PSPs. Again, in order to make the best
use of the available time and resources, we tailored our methods to the needs of our stakeholders,
targeting those PSPs of greatest interest (or for which there was perceived to be the most new
information) for in-depth reviews; others received briefer reviews. It was crucial that the
decisions about which PSPs would receive in-depth review and which would receive brief
review were made by a broadly representative stakeholder committee. The in-depth reviews,
while thorough, did not conform to all of the criteria in the 2011 IOM report, Finding What
Works in Health Care: Standards for Systematic Reviews,13 nor all the criteria in the EPC
Methods Guide (for example, we did not publicly post a protocol for each individual review).
We used our collective experience as EPCs to adapt existing EPC methods that we judged best
preserved the essence of a systematic review while allowing us to complete 18 in-depth reviews
within 9 months and the available budget.
Additionally, over time, we will likely improve our methods for assessing evidence regarding
how patient safety interventions affect health care processes and outcomes. The methods we used
for this report incorporate new perspectives regarding the importance of implementation and
context, which was the focus of the Context Sensitivity report; likewise, in the future, we can
expect to increase our understanding of the interactions between multiple intervention,
implementation, and organizational variables and how these influence safety outcomes. If future
research reveals that these variables interact in ways that our current understanding of theory and
logic models cannot explain, we will need to modify the methods of evaluating PSPs again.
Lastly, we relied on the judgment of our technical experts at every important step of the
project: Therefore our results are as much a product of these judgments as of our systematic
review methods. Hence, our results might be sensitive to the selection of particular experts on
our technical expert panel. However, we mitigated this potential bias by including more than
double the number of experts on our technical expert panel as we typically would for an EPC
review, which allowed us to include a diverse set of stakeholders from the U.S., Canada, and the
United Kingdom; from PSP developers and evaluators to patient safety policymakers to experts
in design and evaluation methods. Rather than regarding the tight linkage between the needs of
the stakeholders and the work of the EPCs as a limitation, we view it as a strength that increases
the likelihood that the results of the review will be meaningful to providers, payors, and patients,
and that the reports results will lead to meaningful change.
505
Conclusions
In 2001, when we published Making Health Care Safer, the literature on PSPs was limited,
for several reasons.
First, the dominant cognitive model for patient safety was that errors represented human
lapses; thus, there seemed little to study. The key PSP, one might say with only slight hyperbole,
was to admonish caregivers to be more careful next time.
Moreover, no business case existed for institutions or individuals to focus on patient safety,
no public pressure was exerted to improve safety, and no research funding was available for
safety studies. The fact that the literature on safety was relatively primitive was anything but
surprising.
This picture changed completely over the ensuing decade. AHRQs Patient Safety Network,
the organizations main portal for safety literature, now lists more than 3000 research studies,
with 400 of these deemed by the editors as Classics. Safety research receives substantial
support from AHRQ and others, the business case for safety improvement has grown, and
policymakers are intensely interested in safety research as they consider what they can do to
promote safety.
We found evidence of this progress in our current review of the literature on patient safety.
There are now over a dozen practices for which the evidence of effectiveness is strong or very
strong, and data are emerging on the contextual factors that so often determine the outcome of
implementation. With the Federal investment in safety (under the Partnership for Patients) of
about $1 billion, and an investment of more than 20 times that amount in information technology
implementation, we are on the cusp of an exploding database of research on safety practices.
Yet recent studies of rates of harm have demonstrated how difficult improving safety really is
and have caused policymakers and researchers to redouble their efforts to identify and implement
safe practices in hospitals, nursing homes, and clinics. Individuals and institutions seeking to
improve safety would do well to scrutinize the practices described in this reportwidespread
implementation is likely to save hundreds, if not thousands, of lives. It will also help us continue
to refine our efforts to identify the factors associated with successful implementation of PSPs,
and to pinpoint, and hopefully prevent, any unintended consequences.
506
Some PSPs were not included in this review because they were not deemed sufficiently
developed, and new PSPs will subsequently be developed. Thus a strategy of surveillance should
be adopted regarding evidence on PSPs. Future research also requires advancing the basic
science of safety measurement such as standardized methods for rare events and for evaluating
studies that assess only process-related outcomes relative to those that assess patient outcomes.
507
individual behavior change. Knowing the details of the intervention also could help readers
of the study determine how much the success of the PSP implementation depended on
contextual issues (e.g., organization or teamwork).
3. Understanding the important items to measure and report on for implementation.
Experts consider having comprehensive information about implementation key to being able
to replicate a PSP. However, little empirical evidence exists about what makes a description
of the PSP adequate for reporting. Assessing what implementers need to know, if they are to
be able to implement or adapt an intervention in their own settings, is critical. Most experts
considered understanding the important items to measure and report on for implementation
to be related to or even the same as reporting the intervention in sufficient detail that it can
be replicated. This view suggests that the distinction between the intervention and the
implementation may be an arbitrary line, and that ideal evaluations of PSP interventions
need to consider the implementation as part of the intervention.
4.
5. Refining a framework for assessing the strength of a body of evidence. The research team
did developmental work on an adaptation of the GRADE and Evidence-based Practice Center
(EPC) systems for assessing the strength of evidence across studies of a PSP. This work
warrants further development.
6. Generating empirical evidence that the contextual factors identified in this project
influence the success of the PSP. The research team acknowledges that most of the
recommendations in the report have a thin empirical evidence base, which simply reflects the
relatively immature state of research in this still relatively young field. Building a stronger
evidence base will help future efforts at refining the recommendations presented here.
508
example, efforts have focused on reducing blood stream infections, improving teamwork, or
enhancing patient engagement.
If health care is to make significant improvements in patient safety, research should inform
and guide these efforts, just as it has done in every other field. We have learned much about how
to improve safety, yet we need to learn much more. Acquiring this knowledge will require
investments in patient safety research, including basic methodological research. To date,
investments in patient safety research have fallen far short of the magnitude of the problem.
To achieve progress in improving patient safety, research is needed in a number of areas:
basic patient safety research to develop new tools and measures and ensure that the tool
matches the problem;
a larger number of valid measures of patient safety;
better methods to measure context and how an intervention was implemented;
methods to identify and provide the necessary skills, resources, and accountability (i.e., a
safety management infrastructure) at each level of the health care system; and
more effective and less burdensome methods of improvement so that clinicians,
researchers, and administrators work on reducing all types of harms patients are at risk of
suffering rather than a select few. Below we briefly discuss each of these.
Basic patient safety research. Largely driven by an appropriate desire to reduce patient harm
from medical errors immediately, the field has often invested in quick fixes that may have lacked
sufficient theory or validated evaluation tools. For example, although the need to evaluate
context when implementing patient safety interventions is widely recognized, few validated
instruments have been developed to accomplish this task. Just as the hundreds of thousands of
deaths from heart disease or cancer each year inspire rather than obfuscate the need for basic
research, so too should the large number of deaths from preventable harm. To improve patient
safety, the Federal Government will need to invest in basic patient research, to diagnose
different types of safety problems, to match the theories and methods to the type of problem, to
better evaluate implementation efforts and their surrounding context, and to evaluate whether
patient safety is indeed improving
The future research needed to advance the science of basic patient safety research was
covered in Context Sensitivity,7 where this technical expert panel rated as highest priority
topics such as development of a theory-based taxonomy with which to describe and evaluate
key elements of interventions, contexts, and targeted behaviors and understanding the
important items to measure and report in implementation.
Larger number and more valid tools to measure safety. Despite more than a decade of effort,
the health care system remains unable to quantify the magnitude of preventable harm or to use
valid tools to evaluate progress in improving patient safety over time or among provider
organizations. Moreover, for most of the patient safety harms discussed in this report, the field
lacks valid, broadly accepted definitions forand the mechanisms to monitorprogress in
reducing patient harm. Thus, despite significant efforts to improve patient safety over the last
decade, no oneincluding patients, providers, researchers, payors, and policymakersknows
whether care is safer. The need for evaluation tools is urgent.
Better methods to measure context and to describe an intervention. In contrast to most
clinical research interventions, patient safety interventions are iterative, evolve over time, are
509
context dependent, and are strongly influenced by the organizations in which they are
implemented and the personnel involved. Both this report and Context Sensitivity have
demonstrated that too often, the intervention is insufficiently described and the context is barely
mentioned; the result is stalled learning and reduced generalizability. Research is needed to
better understand how an intervention was implemented over time, the most salient iterations the
intervention has undergone, and the critical contextual domains that may have supported or
mitigated the improvement effort. For example, while leadership and teamwork are widely
regarded as important in implementing patient safety interventions, the field lacks consensus
both on how best to measure these domains and on a theory that explains how various domains
of context support or hinder the success of an intervention.
The topics more patient safety measures for ambulatory care and better measures of the
major causes of harm, which were judged as high priority by our technical expert panel, fall into
this domain.
Methods to build a safety management infrastructure. Health care is largely organized
around the care of individual patients, yet patient safety requires the management of populations
of patients and accountability for complications. While physicians have profound individual
accountability for their patients, especially for complications that are directly related to their
care, their accountability often diminishes for complications less directly related to the care they
provide, or complications that are influenced by the care of a care team or how care is organized;
infections are an example of such complications.
If patient safety requires the management of a population of patients, an infrastructure should
be in place to help monitor risks and prioritize interventions, to implement interventions, and to
monitor progress. At multiple levels of a provider organization (clinic or unit; department or
region; hospital or group practice), an infrastructure is needed to ensure that safety leaders have
sufficient resources, skills, and accountabilities to improve safety. Little is known regarding the
specifics of the infrastructure that is needed. Nevertheless, the existing infrastructure is largely
underdeveloped.
Specifically, researchers and managers must determine how much physician, nurse, and other
staff support are needed at each unit/clinic, department/groups of clinics, hospital/health system
levels to ensure patient safety; what skills they need, and how they should be held accountable
for the safety of the care provided. For example, if an employees job is to evaluate progress in
patient safety, does he require training in clinical epidemiology? If an employees job is to
implement interventions, does she require training in human factors engineering and
implementation science? Finally, how will managers and researchers create a cascading
accountability system in which unit and clinical leaders hold individual clinicians accountable,
department or regional leaders hold unit/clinical leaders accountable, and hospital or health
systems leaders hold department and regional leaders accountable. Although this type of
infrastructure exists in other industries, little is known about its benefits and costs. Investments
are needed to understand, implement, and evaluate a safety management infrastructure in health
care.
The topic, more data about the costs of patient safety practices, which was judged high
priority by our technical expert panel, falls into this domain.
More effective and efficient system interventions to reduce multiple, rather than single
types of harm. Our technical expert panel strongly advocated for a systematic approach to
510
preventing harm and exploring the concept of mutually reinforcing practices that reduce many
kinds of harm.
The Center for Medicare and Medicaid Services (CMS) is now embarking on an ambitious
national effort to reduce 10 types of preventable harm. However, because of the burden of
implementing the interventions, hospitals are generally selecting interventions to reduce only a
subset of the harms. Yet most patients are at risk for all 10 harms and many others as well.14
In health care, too many improvement efforts rely on the heroism of clinicians rather than
safely designed systems. In other industries, as the amount of information has increased,
technological improvements are implemented, with the result that productivity and safety
increase. Health care productivity remains flat and clinicians use technologies that generally do
not talk to each other. For example, the infusion pump does not talk to a respiratory monitor. If
such intercommunication occurred, a respiratory monitor would automatically shut off an
infusion pump if a patient developed a dangerously low respiratory rate from an infusion of
narcotics, a common cause of respiratory arrest. Preciously few examples exist of safe design in
health care. Such an approach will require close collaboration with systems engineers.
Patient harms do not occur in isolation, and they are not independent. Rather, they are
interdependent; thus, the solutions must be as well. Hospitalized patients are at risk for multiple
complications. For example, a patient on a breathing machine after surgery is at risk for 9 of the
10 complications targeted for reduction by CMS (the tenth, which is an obstetric complication,
would not apply in this case).
Complex patients suffer the same fate. Because these patients have a variety of chronic
diseases, they are at risk for a variety of harms, yet few efforts have been undertaken to
systematically reduce the risk for all types of harm.
Yet there is an alternative. Health care could more fully embrace systems engineering. The
need to develop and demonstrate a system framework that addresses the universal and
fundamental challenges in contemporary health care delivery remains a critical challenge. In the
2005 report, Building a Better Delivery SystemA New Engineering/Health Care Partnership,
the National Academy of Engineering (NAE) and the IOM noted:
a systems approach to healthcare delivery could transform the U.S.
health care sector from an underperforming conglomerate of
independent entities into a high-performance system in which
every participating unit recognizes its interdependence and
influence on every other unit
Ideally, the new and improved health care delivery system should include: 1) an integrated,
ubiquitous, distributed, responsive, expansive, flexible, affordable and resilient system; 2)
personalized delivery facilitated by secure information flow and optimized information that runs
smoothly, efficiently, and safely.
The IOM/NAE report highlighted the observation that health care is significantly underengineered and called for greater input from systems engineers to make health care safer. Future
research in patient safety needs to take a systems approach, focusing on all the harms a patient
suffers, clarifying the therapies that may reduce harm, and ensuring that patients always receive
them. This research could include three key areas: a focus on engaging patients and their
families, ensuring patients receive therapies to reduce harm, and creating a learning and
accountability system. Few examples of successful collaboration between engineering and
medicine currently exist. One example of such a collaboration is the Systems Engineering
511
Initiative for Patient Safety (SEIPS) program at the University of Wisconsin-Madison, which
brings human factors and systems engineers together with clinicians to work on solving complex
patient safety problems.15 This model, developed by Carayon, views the work system as an
interaction among people, tasks, tools and technologies, organization, and environment. Another
example, occurring at Johns Hopkins University, involves a collaboration with the Johns
Hopkins University Whiting School of Engineering, Applied Physics Laboratory, the School of
Medicine, the Bloomberg School of Public Health, the Carey Business School, the School of
Nursing, and the private sector. The purpose of this collaboration is to develop a model to
eliminate preventable harm.
As a part of these collaborations, clinicians and researchers are exploring ways to apply
systems engineering to patient safety. For example, providers might consider all potential harms
of being hooked up to a breathing machine. The resulting list includes 9 of the 10 harms
enumerated by CMS and several more, including patient-centered harms from loss of dignity,
autonomy, and respect. The clinicians and engineers would then consider the tasks or treatments
to prevent those harms and the barriers to performing those tasks and ultimately design a system
that ensures patients receive the recommend therapies and monitors and improve performance.
For example, clinicians could be provided visual displays indicating when a treatment is due
and when it has been completed. As discussed above, at present, few of the technologies
responsible for care processes are integrated: these technologies, which include the medical
devices including the bed, the ventilator, the infusion pump, the monitors, and the electronic
health record, do not communicate. Research into how to better integrate systems engineering
into health care could help improve safety and allow clinicians and managers to work on
preventing all types of patient harm.
A part of adopting a systems approach dictates including patients and their families. Patients
and their families are an integral part of the health care system, yet often, are not adequately
engaged or provided with sufficient information. Future research needs to explore how best to
engage and activate patients to help improve safety and how best to view patients as an integral
part of the health care system.
Efforts to improve patient safety are often fragmented into specific clinical or academic
disciplines. While each approach and method is necessary, it is unlikely any one, by itself, will
be sufficient to address the entire patient safety problem and reduce all types of harms. These
systems approaches that link clinicians, human factors and systems engineers, social scientists,
health services researchers, informatics specialists, economists, and biostatisticians, offer hope
for realizing broad improvements in patient safety. Although such diverse groups pose
management challenges, they have enormous potential to rebuild the health care chasse, which
remains largely broken. If we are to make progress, science must guide us, an endeavor that will
require investments in patient safety research. The research agenda outlined above may point
health care in the right direction.
References
1.
2.
512
3.
4.
9.
10.
5.
11.
6.
12.
7.
13.
14.
15.
8.
513
Abbreviations/Acronyms
AAGBI
Association of Anaesthetists of Great Britain and Ireland
AANA
American Association of Nurse Anesthetists
AAPM
American Association of Physicists in Medicine
ACCP
American College of Chest Physicians
ACGME
Accreditation Council for Graduate Medical Education
ACR
American College of Radiology
ACS NSQIP
American College of Surgeons National Surgical Quality Improvement
Project
ADE
Adverse Drug Event
ADL
Activities of Daily Living
ADR
Adverse Drug Reaction
AE
Adverse Event
AHRQ
Agency for Healthcare Research and Quality
AIR
American Institute for Research
AMI
Acute Myocardial Infarction
AMSTAR
A measurement tool to assess systematic reviews
AORN
Association of periOperative Registered Nurses
APN
Advanced Practice Nurses
APOE4
Apolipoprotein E4
ARRA
American Recovery and Reinvestment
ASA
American Society of Anesthesiologists
ASATT
American Society of Anesthesia Technicians and Technologists
ASRT
American Society of Radiologic Technologists
AV
Audio/Video
BATS
Brain Attack Teams
BC/BS
Blue Cross/Blue Shield
BG
Blood Glucose
BIS
Bispectral Index
BMI
Body Mass Index
BPMH
Best Possible Medication History
BPOC
Barcode-Enabled Point of Care
C
Comparator
CAN-NSQIP
Canadian National Surgical Quality Improvement Collaborative
CAUTI
Catheter-Associated Urinary Tract Infections
cc
Milliliter
CCAs
Critical Care Areas
CCOT
Critical Care Outreach Team
CCT
Clinical Controlled Trial
CCTA
Cardiac Computed Tomography Angiography
CCU
Coronary Care Unit
CDC
Centers for Disease Control
CDC HICPAC
Centers for Disease Control and Prevention Healthcare Infection Control
Practices Advisory Committee
CDSS
Clinical Decision Support Systems
CHF
Congestive Heart Failure
CHG
Chlorhexidine
514
CI
CI-AKI
CKD
CLABSI
CMS
CNA
CNS
CPOE
CPRS
CPSI
CRBSI
CRCPD
CRM
CSO
CSRS
CSS
CT
CTA
CTI
CTPA
CTSQC
CUSP
CVA
CVC
CWOCN
DEER
DERS
DM
DNR
DoD
DRG
DVT
EAST
EBA
EBP
ED
eMARs
EMR
EPC
EPOC
ESA
FDA
FHA
FMEA
FPTK
FSCI
FTE
FTE
Confidence Interval
Contrast-induced Acute Kidney Injury
Chronic Kidney Disease
Central Line-Associated Bloodstream Infections
Centers for Medicare & Medicaid Services
Certified Nursing Assistant
Clinical Nurse Specialist
Computerized Provider Order Entry
Computerized Patient Record System
Canadian Patient Safety Institute
Catheter-related Bloodstream Infection
Conference of Radiation Control Program Directors
Crew Resource Management
Constant Special Observation
Cardiac Surgery Reporting System
Computerized Surveillance System
Computed Tomography
Computed Tomography Angiography
Care Transitions Intervention
Computed Tomographic Pulmonary Angiography
Connecticut Surgical Quality Coalition
Comprehensive Unit Based Safety Program
Cerebrovascular Accident
Central Venous Catheters
Certified Wound Ostomy Continence Nurse
Diagnostic Error Evaluation and Research
Dose Error Reduction System
Diabetes Mellitus
Do Not Resuscitate
Department of Defense
Diagnosis-Related Group
Deep Venous Thrombosis
Eastern Association for the Surgery of Trauma
European Board of Anesthesiology
Evidence-based Practices
Emergency Department
Electronic Medication Administration Records
Electronic Medical Record
Evidence-Based Practice Center
Effective Practice and Organization of Care
European Society of Anesthesiology
Food and Drug Administration
Florida Hospital Association
Failure Mode and Effects Analysis
Fall Prevention Tool Kit
Florida Surgical Care Initiative
Full Time Employment
Full-time Equivalent
515
GAO
GI
GRAM
HAI
HAPU
HAT
HELP
HFE
HFMEA
HH
HICPAC
HITEC
HR
HS
HSOPS
ICDSC
ICU
IEA
IHI
IIDs
IIT
INR
IOM
IOCM
ISMP
ISQIC
IT
IV
JC
JCAHO
KPNCRNC
kV
LEB
LEP
LMWH
LVN
MCR
MCSQC
MDS
MET
MHCS
MICU
MRSA
mSv
MS
MTT
NAAL
NAC
NCHS
NCSQC
NGC
NHSN
NPSA
NPSG
NPUAP
NQF
NR
NS
NSAIDs
NVASRS
OEMs
OR
OR
OSATS
PAC
PACS
PAE
PCA
PDA
PDSA
PE
PGY
PICC
POLST
PPACA
PPV
ProFaNE
PSC
PSCHO
PSP(s)
PTT
PU
QA
QD
QI
QM
QS
RCA
RCT
RED
RFID
RN
RR
RR
RRS
RRT
RSNA
SAQ
SCHA
SCIP
SCOAP
SCR
SCS
SHEA
SHS
SICU
SMD
SQAN
SSI
STS
SUD
SURPASS
T
TBD
TCPS
TEP
TIA
TIW
TSA
TSQC
TURP
U
U.K.
U.S.
UCSF
UHC
UP
VA
VAP
VRE
VRSA
VTBI
VTE
WHO
WOCN
WON
518
A-1
A-2
Use of human factors principles in evaluation of medical devices (Medical device related adverse
events)
Refining performance of medical device alarms (e.g., balancing sensitivity and specificity of
alarms, ergonomic design) (Adverse events)
Transitions in care (broader topic)
Information transfer between inpatient and outpatient pharmacy (Adverse events related to
discontinuities in care)
Handoff protocols (broader topic)
Standardized, structured sign- outs for physicians (Adverse events during cross- coverage)
Use of structured discharge summaries (Adverse events related to information loss at discharge)
Protocols for notification of test results to patients (Failures to communicate significant abnormal
results (e.g., pap smears))
Use of bar coding (Adverse events due to patient misidentification)
Sign your site protocols (Performance of invasive diagnostic or therapeutic procedure on
wrong body part)
Team training (broader topic)
Application of aviation style crew resource management (e.g., Anesthesia Crisis Management;
MedTeams) (Adverse events related to team performance issues)
Simulator-based training (Adverse events due to provider inexperience or unfamiliarity with
certain procedures and situations)
Limiting individual providers hours of service (Adverse events related to fatigue in health care
workers)
Fixed shifts or forward shift rotations (Adverse events related to fatigue in health care workers)
Napping strategies (Adverse events related to fatigue in health care workers)
Specialized teams for inter-hospital transport (Adverse events due to transportation of critically
ill patients between health care facilities)
Mechanical ventilation (Adverse events due to transportation of critically ill patients within a
hospital)
Asking that patients recall and restate what they have been told during informed consent
(Missed, incomplete or not fully comprehended informed consent)
Use of video or audio stimuli (Missed, incomplete or not fully comprehended informed consent)
Provision of written informed consent information (Missed, incomplete or not fully
comprehended informed consent)
Computer- generated reminders to discuss advanced directives (Failure to honor patient
preferences for end-of-life care)
Use of physician order form for life- sustaining treatment (POLST) (Failure to honor patient
preferences for end-of-life care)
A-3
A-4
A-6
B-1
Yes
No
Cant answer
Not applicable
Yes
No
Cant answer
Not applicable
Yes
No
Cant answer
Not
applicable
Yes
No
Cant answer
Not applicable
*The AMSTAR criteria was taken from Shea et al, 2007 Development of AMSTAR: a measurement tool to assess the
methodological quality of systematic reviews. BMC Med Res Methodol 7: 10.
B-2
C-1
LANGUAGE:
English
SEARCH STRATEGY:
intravenous heparin OR heparin NEAR/5 infus*
AND
error* OR cross infection OR infection control OR safe* OR overdos* OR adverse OR
(infection OR infections OR iatrogenic AND (prevent OR prevention OR preventive OR
preventing)) OR protocol* OR nomogram* OR inpatient coagulation service OR inpatient
coagulation services OR human factors OR decision support
AND
article/it OR article in press/it OR conference abstract/it OR conference paper/it OR
review/it
NUMBER OF RESULTS: 791
====================================================================
DATABASE SEARCHED & TIME PERIOD COVERED:
Cochrane 2000-11/21/2011
LANGUAGE:
English
SEARCH STRATEGY:
heparin AND (intravenous OR infusion) in Title, Abstract or Keywords
AND
adverse OR Error* OR Safe* OR overdos* OR ((infection OR infections OR iatrogenic) AND
(prevent OR prevention OR preventive OR preventing)) OR protocol* OR nomogram* OR
inpatient coagulation service OR inpatient coagulation services OR human factors OR
decision support in Title, Abstract or Keywords
NUMBER OF RESULTS: 803 (Cochrane Reviews [11], Other Reviews [4], Clinical Trials
[781], Methods Studies [1], Economic Evaluations [6]
====================================================================
DATABASE SEARCHED & TIME PERIOD COVERED:
CINAHL 2000-11/21/2011
LANGUAGE:
English
SEARCH STRATEGY:
heparin AND (intravenous OR infusion)
AND
C-2
Intervention
Comparator
Usual practice
Outcomes
Timing
Settings
Inclusion/exclusion criteria:
No restrictions were made by language, country of study, or indication for use of heparin.
C-3
SEARCH STRATEGY:
clinical pharmacist*
AND
adverse OR harm* OR side effect* OR safe* OR reaction*
NUMBER OF RESULTS: 84
SECTION B. Methods
Titles and abstracts were reviewed by a physician health services researcher with experience in
both systematic reviews and in clinical pharmacist services. Included studies were those most
relevant to clinical pharmacist interventions on medication errors and adverse drug events in
various health care settings. The focus was on studies that addressed the possible association
between clinical pharmacist activities and improved prescribing practices and/or assessed
whether such activities might lead to reduced medication errors and adverse drug events.
Included studies were narratively summarized by the author.
implementation were the primary resources and given priority in the order above. The synthesis
was narrative.
Date limits
Platform/provider
PubMed
Cochrane Library
Wiley
ECRI Institute
Health Devices
ECRI Institute
2000-November 9, 2011
http://www.ihi.org
Joint Commission
http://www.jointcommission.org/
2000-November 9, 2011
2000-November 9, 2011
http://patientsafetyauthority.org/Pages/Default.aspx#
C-5
Conventions:
PubMed
[mh] =
[majr] =
[pt]
=
[sb] =
[sh] =
[tiab] =
[ti]
=
MeSH heading
MeSH heading designated as major topic
publication type
subset of PubMed database (PreMEDLINE, Systematic, OldMEDLINE)
MeSH subheading (qualifiers used in conjunction with MeSH headings)
keyword in title or abstract
keyword in title
Safety
Controlled
Vocabulary
Infusion pumps[majr]
medication errors[mh]
safety[mh]
safety
management[mh]
risk management[mh])
Keywords
infusion pump*
smart pump*
IV pump*
drug delivery system
drug infusion system
(infusion OR medication OR intravenous] OR IV OR drug OR smart) AND
(pump] OR pumps])
error*
mistake*
safe
safety
risk*
malfunction*
overdos*
wrong
C-6
PUBMED
English language, human, remove overlap
Set
Number
Concept
Search statement
Infusion pumps
infusion pumps[majr]
Safety issues
1 AND 5
Combine sets
keywords for in
process citations (i.e.
not yet indexed with
MeSh headings and
therefore not captured
in set 7)
Combine sets
keywords in title (to
ensure retrieval of any
relevant records that
were not picked up in
the previous sets)
3 AND 6
10
Combine sets
7 OR 8 OR 9
11
Eliminate records on
proton pump inhibitors
10 NOT proton*
12
13
Apply limits
Total Downloaded
Total Retrieved
Total Included
53
150
18
SECTION B. Methods
Titles and abstracts were reviewed by a health services research methodologist with experience
in both systematic reviews and medical devices. Included studies were those most relevant to
evaluation of smart pumps and related protocols for reduction of medication errors and adverse
drug events in various health care settings. The focus was on studies that compared medication
C-7
error rates and adverse drug events following implementation of these technologies in hospitals
compared to a control period when the technologies were not active or in place. Potential barriers
to implementation (e.g. user compliance) were also assessed. Included studies were narratively
summarized by the author.
interventions from 1950 to 2008 to decrease catheter-associated urinary tract infections by means
of the MEDLINE and Cochrane databases (using Ovid), the PubMed Journals and Medical
Subject Heading (MeSH) datasets, the ISI knowledge databases (Web of Science and Biosis
Previews) and the CINAHL and EMBASE databases. The MEDLINE and Cochrane database
searches were conducted by exploding and combining the following Medical Subject Heading
(MeSH) terms: urinary tract infection, urinary catheterization, indwelling catheter, inpatient,
reminder system, device removal, intervention studies. The MeSH reminder system was also
searched separately. We included the following terms in a keyword search (with wildcard
indicated with *): urinary tract infection; ((urin* or uret*) and cath*)) or catheter*; nosocomial
or inpatient or hospital*; reminder, removal, and intervention. We used similar strategies with
the other databases. A research librarian provided guidance to improve search completeness.
This search yielded 6679 citations, including many duplicate citations. As our initial search was
broad and yielded many guidelines and reviews published regarding prevention of catheterassociated urinary tract infection, we also evaluated these articles reference lists for additional
studies; 1 additional reference was located in this manner. More detailed review was required for
118 articles to determine whether they met inclusion criteria. After applying inclusion and
exclusion criteria to focus on human studies of adults admitted to acute care hospitals reporting
at least one outcome involving catheter use or CAUTI events as a result of the intervention, and
with a comparison group (either pre- versus post-intervention or a separate control group); this
yielded 16 studies for further review. Two authors of the systematic review (J.M. and M.M.)
independently reviewed and abstracted data from the 16 articles that appeared to meet inclusion
criteria, including setting, study population, inclusion/exclusion criteria, definitions used, health
outcomes, and quality issues. A third investigator (S.S.) resolved any differences in abstraction
and reviewed the joint decisions made to exclude 2 of the 16 articles that no longer met inclusion
criteria after further review. As a result, this systematic search in 2008 yielded the 14 articles
reviewed in the previously published meta-analysis.
To update the prior literature search for Chapter 9, a search was performed of MEDLINE and
Cochrane databases (using Ovid) and PubMed for intervention studies (published from August
2008 to February 2012) to reduce use of unnecessary urinary catheters in the acute care of adults,
using the same detailed search strategy as employed in the 2008 search. Yet, unlike the 2008
search which was focused on removal of recently placed indwelling catheters (which excluded
emergency environments), the patient population for the 2012 search was expanded to include
emergency department patients because use of interventions to restrict initial placement was an
additional topic of interest for Chapter 9. The 2012 search results were also supplemented with
prior lists of articles excluded from the prior 2008 search that were focused on emergency
department interventions. A secondary evaluation of the CINAHL database was also performed
for interventions developed and implemented by nurses related to urinary catheter use. In light of
the somewhat different terminology on the topic found in the nursing literature, we searched
CINAHL using variations of the following terms: reminder, removal, urinary catheter, nurse
empowered, nurse directed, nurse protocol. No date limits were employed in the CINAHL
search, which retrieved 5 records. Overall, the MEDLINE and CINAHL searches yielded 479
citations, including 353 from MEDLINE through Ovid, 9 additional from PubMed, 117 from the
Cochrane EBM databases, and 7 duplicates. Studies were included if at least one outcome
involving catheter use or CAUTI events (Table 1 in Chapter 9) was reported as a result of the
intervention with a comparison group. A review of reference lists for additional studies was also
C-9
performed, yielding one additional study. After applying inclusion and exclusion criteria to focus
on human studies of adult patients with at least one outcome involving catheter use or CAUTI
events reported as a result of the intervention, and with a comparison group, this updated search
yielded 12 intervention studies published since the prior meta-analysis.
SECTION B. Methods
As summarized in the previously published meta-analysis for the 14 selected studies from 2008
or earlier, a systematic review process was performed. Correspondence with 24 authors was
initiated to clarify details regarding the interventions and outcomes with responses received from
11 authors, and 4 authors provided unpublished numeric data necessary for statistical pooling.
Two physician reviewers performed a detailed abstraction of the 14 studies. Details of the
statistical analyses for obtaining the pooled effects are detailed in the prior published analyses,
and were not replicated or expanded for writing Chapter 9.
A similar review and abstraction process was performed by one physician (J.M.) for the 12
recent articles in the updated search. No contact was initiated with authors, and theses articles
were analyzed and compared in a narrative process rather than a meta-analysis.
Details of the 14 prior and 12 recent studies are summarized in the Appendix Table for Chapter
9, regarding study design, patient population size and care environment, and details of the
interventions used to either avoid inappropriate placement or to prompt removal of unnecessary
catheters. Other important interventions that could possibly influence the outcomes of the studies
were also summarized in this table. Important outcomes of the 14 prior studies as previously
published in the meta-analyses were summarized in Figure 2; similar outcomes for the 12 recent
studies were summarized in Figure 3.
SEARCH STRATEGY:
Medical Subject Headings (MeSH) Bacteremia and Catheterization, Central Venous, and the
MeSH subheadings Prevention & control and Adverse effects, as well as the keywords
central line-associated bloodstream infection, central line, and central venous catheter.
Search terms included variations of the keywords bacteremia, bloodstream infection,
central line, central venous catheter, prophylaxis, and prevention, using wildcards and
truncation to capture alternate spellings and endings.
NUMBER OF RESULTS:
1,087 unique manuscripts were retrieved by the search of which 337 articles were relevant for
this report.
SECTION B. Methods
All relevant titles and abstracts were reviewed by a physician health services researcher (VC)
with experience in both systematic reviews and in the topic of central line associated
bloodstream infections (CLABSI). Studies included were those most relevant to prevention of
CLABSI; in addition, studies that reported on local and national policies, economic impact and
interventions associated with CLABSI reduction were included in this report.
C-11
Date limits
Platform/provider
2001-November 3, 2011
ECRI Institute
2001-November 3, 2011
PSNet
2001-November 3, 2011
PubMed
2001-November 2, 2011
FDA
2001-November 8, 2011
JCAHO
2001-November 3, 2011
NCQA
2001-November 3, 2011
MeSH heading
MeSH heading designated as major topic
C-12
[pt]
[sb]
[sh]
[tiab]
=
=
=
=
publication type
subset of PubMed database (PreMEDLINE, Systematic, OldMEDLINE)
MeSH subheading (qualifiers used in conjunction with MeSH headings)
keyword in title or abstract
Controlled Vocabulary
Keywords
Equipment
surgical instruments
equipment and supplies
disposable equipment
surgical
medical
cutting
instrument*
tool*
equipment*
device*
trocar*
scalpel*
shaver*
raz*
drill*
catheter*
syringe*
needle*
mask*
gown
glove*
endoscop*
instrumentation
Reprocessing
equipment reuse
reuse
reusing
reus*
reprocess
reprocessing
reprocessed
reprocess*
recycled
recycling
recycle*
repurposed
repurposing
repurpose*
Single-Use
single use
single-use
disposable
one-time
one time
C-13
PubMed
English language, human, date limit: January 1, 2001-November 2, 2011
Set
Number
Concept
Search statement
Equipment
Reprocessing
Single Use
Combine
S1 AND S2 AND S3
Total Downloaded
Total Retrieved
Total Included
75
11
15
SECTION B. Methods
Titles and abstracts were reviewed by a health services research methodologist with experience
in both systematic reviews and medical devices. Included studies consisted of systematic reviews
and clinical studies that compared patient outcomes following use of new versus reprocessed
single-use devices as well as laboratory studies that tested an array of reprocessed single-use
devices for microbiological contamination. Data regarding potential cost-savings of reprocessed
single-use devices was also presented. Included studies were narratively summarized by the
author.
Date limits
Platform/provider
OVID SP
MEDLINE
OVID SP
PreMEDLINE
OVID SP
C-14
Name
Date limits
Platform/provider
EBSCOHost
Cochrane Library
Wiley
PubMed
Scopus
Science Direct
C-15
[tiab] =
Controlled Vocabulary
Anesthesia/
Anesthesiology/
Exp perioperative care/in, ae, mt, st
Checklist
Checklist/
World Health Organization/
Context/Setting
Equipment
Electronics, medical/
Equipment failure/
Equipment failure analysis/
is.fs.
Hospital
Procedures/Administration/Protocol
Incidence
Exp incidence/
C-16
Keywords
Analges$
Anaesthe$
Anesthe$
Sedat$
Checklist$
safety checklist$
Checkout
Checkt-out
WHO
world health
*world health organization
Intraoperat$
Operat$
Patient$
operating room$
intraoperat$
Perioperat$
Preoperat$ or pre-operat$
Perioperat$ or peri-operat$
Postoperat$ or post-operat$
Surg$
Surgical suite$
Alarm$
Apparatus
Check-out or checkout
Electronic checklist$
Equipment
Error$
Failure
fault
Machine$
Monitor$
System$
Administrat$
Organization$
Patient care
Patient safety
Polic$
Protocol$
Standard$
Decrease$
Incidence
Prevalence
Reduc$
Number
Concept
Interventions
Controlled Vocabulary
Checklist/
Intraoperative complications/pc
Medical errors/pc
Postoperative complications/pc
Obstacles
Attitude/
Attitude of health personnel/
Exp Medical staff, hospital/
Nursing staff, hospital/og, st, ut
Organizational Culture
C-17
Keywords
An?esthesia adj2 check$
An?esthesia safety
checklist
Checklist$
Communicat$
Document$
Guideline$
Implement$
Information adj shar$
Instrument$
Knowledge
Missed step$
Precaution$
Practice$
Preoperative
Pre-op$
proactive
Postoperative procedure$
Safety measure$
Strateg$
Surg$ adj2 check$
Surgical safety checklist
Team briefing$
Tool$
Anesthesiologist$
Anesthetist$
Attitude$
Barrier$
Competen$
Educat$
Gap$
Knowledge
Nurse$
Obstacle$
overload
Resident$
Resource$
Surgeon$
time
Train$
Change$
Context$
Culture$
Direct$
Hospital$
Manage$
Organization$
Staff
Team$
Unit$
Concept
Outcomes
Controlled Vocabulary
Safety management/
Quality Management
Surgery
Keywords
Adverse event$
Adverse effect$
Cardiac arrest
coma
Complication$
CPR
Culture change$
Death$
Heart attack
Infection$
Injur$
Myocardial infarction of MI
Outcome$
Patient safety
Perform$
Renal failure
Resuscitat$
Risk management
sepsis
Transfusion$
Ventilat$
Health$
Healthcare$
Hospital$
Quality
TQM
Total quality management
Operat$
operating room$
Surg$
Surgical suite$
Embase/Medline/Premedline
English language, human, remove overlap
Set
Number
1
Concept
Anesthesia
Context/Setting
5
6
Search statement
(anesthesia/ OR anesthesiology/ OR (anaesthes$ OR
anesthes$).ti,ab.) OR anesthesia department, hospital/og, st OR exp
patient care/og, st
operating rooms/ OR operating room/ OR exp perioperative care/in, ae,
mt, st, og OR exp surgical procedures, operative/ OR exp general
surgery/ OR su.fs. OR surgery department, hospital/og, st OR
operating rooms/og, st OR exp patient care/og, st
((OR OR operating room$ OR operat$ OR surg$ OR surgical
suite$).ti,ab.) OR ((pre?operative OR pre?op OR peri?operative OR
pre?surgical OR intra?op$).ti,ab.)
1 OR 2 OR 3
C-18
Set
Number
9
Concept
Equipment
10
11
12
13
Medical errors
14
Staff
15
Safety checklists
16
17
18
19
20
21
Organizational culture
22
23
24
25
26
27
28
29
30
31
32
Search statement
((exp equipment failure/ OR electronics, medical/ OR is.fs.) AND
check$.ti,ab.) OR ((equipment or machine$ OR apparatus OR system$
OR monitor$) AND (check$ OR checklist OR check-out OR
checkout)).ti,ab.
((anesthesia OR anaesthesia) AND simulation$ AND check$).ti,ab.
9 OR 10
exp incidence/ OR (decrease$ OR incidence OR prevalence OR
reduc$ OR number$).ti,ab. OR ((before and after) OR (preintervention
OR preintervention OR pretest OR pre-test OR postintervention OR
postintervention OR posttest OR post-test)).ti,ab.
medical errors/pc OR postoperative complications/pc OR ((error$ OR
complication$ OR adverse event$ OR intraoperative awareness OR
wrong$) AND (prevent$ OR control)).ti,ab.
exp medical staff, hospital/ OR nursing staff hospital/og, st, ut OR
attitude of health personnel/ OR attitude/ OR ((nurse$ OR anesthetist$
OR anesthesiologist$ OR resident$ OR surgeon$) AND (knowledge
OR attitude$ OR competen$ OR train$ OR educat$)).ti.
(safety checklist$ OR ((an?esthesia OR surg$) adj2 check$) OR
((surg$ OR pre?surg$ OR pre?op$ OR peri?op$ OR intra?op$) AND
(checkout$ OR checkout$))).ti,ab.
Checklist/
15 OR 16
safety/ OR safety management/ OR safety.ti,ab. OR ((preop$ OR preop$ OR periop$ OR peri-op$ OR pre?surg$) AND (safety OR
precaution$)).ti,ab.
8 OR 18
total quality management/ OR health care quality/ OR ((health$ OR
healthcare OR hospital$) AND (quality OR TQM OR total quality
management)).ti,ab.
health care delivery/ OR organizational culture/ OR ((organization$ OR
hospital$ OR unit$ OR team$ OR staff) AND (culture$ OR change$ OR
manage$ OR direct$ OR context$)).ti,ab.
14 OR 20 OR 21
24 AND 12
24 AND 18
25 OR 26 OR 27
11 AND 15
29 AND (13 OR 19)
24 AND 22
28 OR 30 OR 31
C-19
Set
Number
33
34
35
36
37
Concept
Combine with WHO
checklist
Combine for final set
Apply limits
Remove duplicates
Total Downloaded
Search statement
32 OR 23
32 OR 33
Limit 34 to yr=2000-2011
Limit 35 to English language
Remove duplicates from 36
Total Retrieved
Total Included
459
SECTION B. Methods
Patient safety problem: Preoperative checklists can help prevent errors and complications
related to surgery. Checklists are often implemented within a multifactorial strategy of
interventions, therefore they cannot be judged alone as a patient safety practice. The World
Health Organization Surgical Safety Checklist is a prominent example of a preoperative checklist
intended to ensure safe surgery and minimize complications; it has been translated into at least
six languages.1 One family of errors involves wrong site surgery (such as wrong procedure,
wrong site, wrong person), and in 2004, the Joint Commission created the Universal Protocol for
Preventing Wrong Site, Wrong Procedure, Wrong Person Surgery.2 It comprises three sets of
steps: pre-operative verification process, marking the operative site, and a time out
immediately before the operation. A checklist can be used to clarify the details of these three
steps. For anesthesia checklists, in 2008 the American Society of Anesthesiologists provided
general guidelines that should be checked before surgery, and institutions can implement the
guidelines to tailor the checklist to their specific equipment and clinical setting.3
Proposed key questions
1. What is the evidence on the context and implementation of preoperative checklists in
healthcare facilities?
2. What is the evidence on the adoption and diffusion of preoperative checklists in
healthcare facilities?
3. What is the evidence on the effectiveness of preoperative anesthesia checklists in
healthcare facilities?
4. What is the evidence on the context and implementation of preoperative anesthesia
checklists in healthcare facilities?
5. What is the evidence on the adoption and diffusion of preoperative anesthesia checklists
in healthcare facilities?
C-20
PICOTS
Elements
Population
Intervention
Comparison
Outcomes
Timing
Settings
Comments
KQ1 and KQ2: Patients undergoing any surgery.
KQ3, KQ4, and KQ5: Patients undergoing any surgery involving general anesthesia.
Preoperative checklist, either electronic or hard-copy.
KQ1 and KQ2: For a preoperative checklist addressing surgical safety in general, we
examined in detail the World Health Organization Surgical Safety Checklist. For
preoperative checklist specifically designed to implement the Universal Protocol and
prevent wrong-site surgery, any checklist.
KQ3, KQ4, and KQ5: For anesthesia, it must have been an equipment checklist prior
to administering general anesthesia before surgery
KQ1 and KQ2: No comparison required to be reported, but we extracted information
on comparisons that were made.
KQ3, KQ4, and KQ5: Not using a checklist, or a different checklist.
KQ1 and KQ2: No health outcomes necessary to be reported (because these
questions do not involve effectiveness), but we extracted information on outcomes that
were reported
KQ3, KQ4, and KQ5: Rates of intraoperative awareness, any equipment
complications, intraoperative patient complications, postoperative patient
complications
Only examined postoperative events within one month of surgery, because later
events are less likely to have been caused by the surgery itself.
Hospitals and surgical centers
Inclusion criteria:
General inclusion criteria: Full article published in a peer-reviewed journal, Abstracts will be
excluded, English language publications only, published in 2000 or later, preoperative checklist
(either electronic or hard-copy), surgery at either a hospital or a surgical center.
Inclusion criteria for Key Questions 1 and 2:
Patients undergoing any surgery
For preoperative checklists primarily designed to implement the Universal Protocol and
prevent wrong-site surgery: At least 20,000 procedures. This number may change
depending on the size of the literature that meets the inclusion criteria.
For a preoperative checklist addressing surgical safety in general, we examined in detail
the World Health Organization Surgical Safety Checklist.
For preoperative checklist specifically designed to implement the Universal Protocol and
prevent wrong-site surgery, any checklist.
Any study design included, because these questions involve issues of implementation and
adoption, which do not require a comparison set of procedures.
Inclusion criteria for Key Questions 3, 4, and 5:
Patients undergoing any surgery involving general anesthesia
At least 100 procedures. This number may change depending on the size of the literature
that meets the inclusion criteria.
Equipment checklist prior to administering general anesthesia before surgery
Study must either compare the use of a checklist to not using a checklist, or study must
compare checklists. We included any design that made such as comparison (e.g., beforeafter, interrupted time series, or time series with concurrent control group, etc).
C-21
References
1.
2.
3.
C-22
SECTION B. Methods
Evidence from the literature and the ACS NSQIP website was reviewed by a general surgeon
health services researcher with experience in systematic reviews. The synthesis was narrative.
Date limits
Platform/provider
Cochrane Library
Wiley
ECRI Institute
PSNet
PubMed
C-23
exp
= explodes controlled vocabulary term (e.g., expands search to all more specific
related terms in the vocabularys hierarchy)
.de. or
/
=
.fs.
=
.hw. =
.md. =
.mp. =
.pt.
=
.ti.
=
.tw.
=
PubMed
[mh] =
[majr] =
[pt]
=
[sb] =
[sh] =
[tiab] =
Controlled Vocabulary
Foreign bodies[mh]
Retained item
Medical errors
Medical errors[mh]
Surgical
Surgical instrument[mh]
Technology
Keywords
Foreign
Gossypiboma
nothing left behind
Retained
Body
Bodies
Instrument*
Sponge*
Tool*
Error*
Medical
never event
Prevent*
Surgical
Surgery
Surgical
bar code
Bar-code
count
RFID
Tag*
Embase/Medline/Premedline
English language, human, remove overlap
Set
Number
Concept
Search statement
Foreign bodies
Surgery
Surgical[tiab] OR surgery[tiab]
C-24
Set
Number
Concept
Search statement
Medical errors
Surgical tools
Technology
Term
Combine
Combine
S1 AND S5
S6 OR S7 OR S*
10
2000-2011
Total Downloaded
Total Retrieved
Total Included
70
20
13
SECTION B. Methods
Titles and abstracts were reviewed by a health services research methodologist with experience
in both systematic reviews and medical devices. Included studies were those most relevant to the
risk and prevention of retained foreign objects as a result of surgery. We examined studies on
manual counting, as well as those using various forms of radiofrequency identification. Potential
barriers to implementation (e.g. user compliance) and the costs of various technologies were also
assessed. Included studies were narratively summarized by the author.
Date Limits
Platform/Provider
PubMed
CINAHL (Cumulative
Index to Nursing and
Allied Health Literature)
EBSCOhost
ECRI Institute
Health Devices
ECRI Institute
C-25
Name
Date Limits
Platform/Provider
http://www.ihi.org
Joint Commission
http://www.jointcommission.org/
Pennsylvania Patient
Safety Authority PASR
(PA Safety Authoritypatient safety reporting
system)
http://patientsafetyauthority.org/Pages/Default.aspx#
http://www.google.com
C-26
Controlled Vocabulary*
operating rooms[majr]
Integration
Information itself
Transmission of
information
computer communication
networks[majr]
monitoring,
intraoperative/instrumentation[majr]
videorecording[majr]
C-27
Keywords
operating room/s
operating suite/s
surgery suite/s
surgical suite/s
OR/s
C-suite/s
hybrid operating suite/s
supersuite/s
super suite/s
VIOR/s (visually integrated operating room)
surgical field/s
integration
integrated
central control
centralized control
centralised control
common location
single location
plug and play
router
routing
audio
AV
camera
data
EMR
image/s
imaging
PACS
picture archiving
video
communication/s
digital
Concept
Display of
information
Controlled Vocabulary*
data display[mj]
Keywords
boom
cart-mounted
dashboard/s
display/s
flat panel
high definition
HD
LCD/s
monitor/s
television/s
touch screen/s
touchscreen/s
TV/s
screen/s
surgical display/s
streaming
video
VCR
wall-mounted
workstation/s
*Note: none of the MeSH terms were specific enough to retrieve results relevant to the topic and were not incorporated into the
main search strategy.
Concept
Search Statement
Main concept
Setting
Descriptive concept
integrat*[ti]
Combine sets
2 AND (3 OR 4)
Combine sets
1 OR 5*
*Note: no publication limits or safety concepts were applied to the final set of search results because there were few results.
C-28
Concept
Setting
Search Statement
C-suite OR C-suites OR hybrid operating OR
supersuite OR super suite
OR
ORs
operating room AND future
VIOR*
2
Concepts
integrat*[tiab]
central[ti] OR central location OR common location
OR single location OR (visually[ti] AND
integrated[tiab])
router*[tiab] OR routing[tiab]
audio* OR AV
camera*[ti] OR image[ti] OR images[ti] OR imaging[ti]
OR picture archiving OR video*[ti] OR (visually[ti]
AND integrated[tiab])
EMR[tiab]
cart mounted OR touch screen OR touch screens
OR touchscreen* OR workstation*
boom* OR flat panel OR TV* OR television* OR
wall-mounted OR HD OR LCD* OR VCR* OR
streaming
communication[ti] OR communications[ti] OR data[ti]
OR data integration[tiab] OR digital integration[tiab]
monitor[ti] OR monitors[ti] OR surgical display OR
surgical displays OR (surgical[ti] AND display*[ti]) OR
(surgical field AND display[tiab]) OR (surgical field
AND monitor*[tiab])
3
Mesh concepts
Computer communication networks[mj] OR data
display[mj] OR monitoring,
intraoperative/instrumentation[mj] OR systems
integration[mj] OR video-assisted surgery[mj] OR
video recording[mj]
*Note: The terms in this table mainly yielded no results, irrelevant results, or large sets of results with few relevant. Some
references were identified and kept during searches with these terms, but the terms were not useful enough to include in the main
search strategy above.
CINAHL
Set Number
Concept
Search Statement
Setting
Main concept
integrat*
Combine
1 AND 2
Total Downloaded
Total Retrieved
Total Included
140
44 instructions
18
SECTION B. Methods
Titles and abstracts were reviewed by a physician health services researcher with experience in
systematic review. Only full published studies were considered for review (meeting abstracts
were excluded). Only English-language publications were eligible for inclusion. For the
effectiveness and harms of the PSP, we considered including studies of any design (e.g.,
C-29
C-30
Anaesthesia
Circulation
Chest
Anesthesia & Analgesia
Annals of Emergency Medicine
Anesthesiology
Archives of Surgery
NUMBER OF RESULTS AFTER FILTERING: 92
============================================================
SEARCH #2 (RCTS OR META-ANALYSES):
DATABASE SEARCHED & TIME PERIOD COVERED:
PubMed 2000-12/16/2011
LANGUAGE:
English
SEARCH STRATEGY:
ultrasound OR ultrasonograph*
AND
guided OR guidance OR ultrasound-guided OR doppler-guided OR ultrasound-assisted
AND
catheter* OR cannulat*
AND
vein OR veins OR venous OR vascular
AND
randomized controlled trial* OR randomized controlled trial[pt] OR rct* OR double-blind* OR
single-blind* OR double blind OR single blind OR systematic review OR meta-analy* OR
metaanaly* OR meta analy* OR Meta-Analysis[pt]
NUMBER OF RESULTS: 93
NUMBER AFTER REMOVING REFERENCES FROM SPECIFIED JOURNALS
LIST: 74
====================================================================
SEARCH #3
DATABASE SEARCHED & TIME PERIOD COVERED:
SCOPUS 2000-12/16/2011
LANGUAGE:
English
SEARCH STRATEGY:
C-31
TITLE-ABS-KEY((ultrasound OR ultrasonograph*)
AND
guided OR guidance OR ultrasound-guided OR doppler-guided OR ultrasound-assisted
AND
TITLE-ABS-KEY-AUTH(catheter* OR cannulat*)
AND
vein OR veins OR venous OR vascular
AND
SUBJAREA(mult OR medi OR nurs OR vete OR dent OR heal OR mult OR arts OR busi OR
deci OR econ OR psyc OR soci)
MANUALLY FILTERED IN ENDNOTE FOR RCTS OR META-ANALYSES
NUMBER OF RESULTS: 11
====================================================================
SEARCH #4
DATABASE SEARCHED & TIME PERIOD COVERED:
Web of Science 2000-12/16/2011
LANGUAGE:
English
SEARCH STRATEGY:
ultrasound OR ultrasonograph*
AND
guided OR guidance OR ultrasound-guided OR doppler-guided OR ultrasound-assisted
AND
catheter* OR cannulat*
AND
vein OR veins OR venous OR vascular
MANUALLY FILTERED IN ENDNOTE FOR RCTS OR META-ANALYSES
NUMBER OF RESULTS: 12
B. Methods
Titles and abstracts were reviewed by a general internist with experience in systematic reviews
(Figure 1). Relevant articles were narratively summarized. This summary was reviewed by the
second author, a general internist experienced in the implementation and use of ultrasound for
central-line placement, who suggested several additional references and described program
implementation at one health care site.
C-32
* A meta-analysis is also included in these totals, hence numbers sum to more than the total.
C-33
C-34
Date Limits
Platform/Provider
EBSCOHost
Cochrane Library
Wiley
OVID SP
MEDLINE
OVID SP
PreMEDLINE
OVID SP
PubMed
C-35
C-36
Disease/Condition
Intervention program
Controlled Vocabulary
Analgesic agent/adverse drug reaction,
drug interaction, drug toxicity,
pharmacokinetics, pharmacology
Analgesics, opioid/ae
Exp analgesics/adverse effects,
pharmacokinetics, poisoning, toxicity
Exp perioperative care/
Exp Postoperative complication/
Postoperative care/
Delirium/
Delirium/et
Delirium/pc
C-37
Keywords
Anaesthe$
Analgesic$
Anesthe$
Complicat$
Opioid$
Postop$
Sedat$
Concept
Risk/Screening
Controlled Vocabulary
Delirium/et
Exp Risk/ or Risk Assessment/
Setting
Keywords
Assess
Assessment
Checklist
Examination
Examine
History
Interview
Predict
Prediction
Predictor
Risk
Survey
admission
Hospital$
Hospital-acquired
Iatrogenic
Inpatient
Intensive care
ICU
nosocomial
patient
Embase/Medline/Premedline
English language, human, remove overlap
Set
Number
1
2
Concept
Disease/Condition
3
4
Combine sets
Risk
8
9
10
11
12
13
14
15
16
17
Search Statement
*Delirium/
(deliriuim or sundown syndrome or acute confusional
state).ti,ab.
1 or 2
exp risk/ or risk assessment/ or (epidemiology or etiology or
prevention).fs. or (avoid$ or caus$ or risk$ or predict$ or
prevent$).ti,ab.
(delirium/et or (delirium and (cause$ or result$ or outcome$ or
complicat$ or etiology or develop$ or effect$ or event$)).ti,ab.)
(3 AND 4) OR 5
(hospital or hospitals or hospitaliz* or hospitalis* or inpatient$ or
iatrogenic or admission or admitted or ICU or intensive care
or post anesthesia or post anaesthesia or post surgery or
post surgical or postoperative or post operative).ti.
exp hospitalization/ or exp intensive care units/
7 OR 8
6 AND 9
C-38
Set
Number
18
19
20
21
22
Concept
Combine sets for Sedation
Combine sets for Postoperative
or Sedation Complications
Combine sets for Postoperative
or Sedation Complications and
Risk of Delirium
Disease/Condition prevention
and control
Interventions
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Remove duplicates
Search Statement
15 or 16 OR 17
14 OR 18
19 AND 6
3 OR 21
(quality and improv$ and intervention$).mp.
31 AND 32
10 OR 20 OR 30 OR 33
Limit 34 to yr=2000-2011
Limit 35 to English language
36 AND ((research synthesis or pooled).mp. or systematic
review/ or meta analysis/ or meta-analysis/ or ((evidence base$
or methodol$ or systematic or quantitative$ or studies or
search$).mp. and (review/ or review.pt.)))
Remove duplicates from 37
Total Downloaded
Total Retrieved
Total Included
587
301
85
SECTION B. Methods
Inclusion/Exclusion Criteria
General criteria: Only full published studies were considered for review (meeting abstracts were
excluded). Only English-language publications were eligible for inclusion.
C-39
Risk factors:
Included RCTs comparing groups with different risk factors; also prospective and
retrospective cohort studies that perform multivariate analyses of factors associated with
incidence of delirium.
Comparative studies must have at least 20 patients in each arm, while cohort studies must
have at least 20 patients overall.
Effectiveness and harms:
Included RCTs, controlled clinical trials (CCTs), interrupted time series, and controlled
before-after studies (CBAs) where at least the after-intervention portion is prospective;
CBAs are necessary to look at implementation (KQ6).
Studies must have at least 20 patients in each arm.
Implementation and Context:
Abstracted information on implementation and context from effectiveness studies, and
descriptive studies of implementation with an associated effectiveness study
Qualitative research studies addressing implementation of delirium prevention
interventions
Quantitative research studies on implementation of delirium prevention interventions
Date Limits
Platform/Provider
EBSCOHost
Cochrane Library
Wiley
OVID SP
MEDLINE
OVID SP
PreMEDLINE
OVID SP
Controlled Vocabulary
Pressure ulcer/
Skin ulcer/ or skin ulcers/
Decubitus/ or decubitus ulcer/
C-41
Keywords
bed sore$
bedsore$
decubitus adj ulcer$
pressure sore$
Pressure ulcer$
pressure ulcer$
pressure wound$
Skin ulcer$
wound
Concept
Intervention program
Controlled Vocabulary
clinical effectiveness/evaluation
exp health care quality/
models, organizational/
program development/
program evaluation/
quality assurance, health care/
quality of health care or quality
assurance,health care or quality
indicators, health care or health
plan implementation
safety management/methods
Barriers
Setting
Effectiveness/Measure
Exp incidence/
Exp prevalence/
Exp vital statistics/
C-42
Keywords
Checklist
checklist*
Clinical checklist
Clinical documentation
clinical protocol
Implement*
implement*
Initiative
initiative*
Intervention
intervention
Medical checklist*
Medical documentation*
Medical protocol*
Program
program*
Protocol
protocol
quality and improvement and
intervention$
Standard
standard*
train*
Training
barrier$
Barriers
Compliance
compliance$
imped$
(nurse$ or physician$ or staff or
employee) and (educat$ or
train$ or knowledge or attitude$
or competen$ or time)
obstacle$
outcome$
Outcomes
alliance$
coalition$
collaborat$
health care or healthcare medical
health system$
hospital$
network$
decrease
incidence
increase
number
prevalence
rate
Embase/Medline/Premedline
English language, human, remove overlap
Set
Number
1
2
3
4
5
6
7
Concept
Disease/Condition
Search Statement
*pressure ulcer/ or pressure ulcer*.ti,ab. or ((skin ulcers/ or skin ulcer/)
and pressure.ti,ab.)
(exp decubitus/ or exp decubitus ulcer/) and skin.mp.
(pressure adj2 (sore$ or ulcer$ or wound$)).ti,ab.
(bedsore$ or (bed adj2 sore$)).ti,ab.
(decubitus adj ulcer$).ti,ab.
1 or 2 or 3 or 4 or 5
program evaluation/ or program development/ or safety
management/methods or models, organizational/ or clinical
effectiveness/evaluation or quality assurance, health care/ or ((clinical or
medical) adj (protocol* or checklist* or documentation*)).ti,ab.
(quality and improv$ and intervention$).mp.
(implement* or initiative* or program* or intervention or train* or
checklist* or standard* or protocol).mp.
exp health care quality/ or (quality of health care or quality
assurance,health care or quality indicators, health care or health plan
implementation).sh.
7 or 8 or 9 or 10
8
9
10
11
12
Obstacles/Barriers
13
14
15
16
17
18
19
20
21
22
23
24
25
14 and 18
19 and 17
C-43
Set
Number
26
Concept
Combine Obstacles and
Disease and
Intervention/Program
Implementation/Quality
improvement and Setting
and Incidence
Limit
Limit
Combine Disease and
Quality Improvement
Combine for final set
Limit
Limit
Limit
27
28
29
30
31
32
33
Search Statement
22 and 25
Total Downloaded
Total Retrieved
Total Included
454
87
47
CINAHL
Set
Number
Concept
Search Statement
S1
Disease/Condition
#
Downloaded
Program
S3
Obstacles
S4
Protocols/Guidelines
S5
Combine Programs
and Protocols
S2 OR S4
S6
Combine Condition
and Obstacles AND
Programs or
Protocols
S1 AND S3 AND S5
S7
Limit
Limit S6 to 2000-2011
350
From S7 keep
64
C-44
SECTION B. Methods
Inclusion criteria:
Experimental research studies including randomized controlled trials, non-randomized
controlled trials, pre-post studies (or before and after studies), and cohort studies that
evaluated the implementation of a multicomponent pressure ulcer (PU) prevention
programs
Published post-2000 and conducted in the U.S.
Study must report on PU rate (incidence/prevalence)
Studies must report PU rate for at least 6 months post- implementation of prevention
program
Exclusion criteria:
Studies that did not report a baseline (pre-prevention program implementation) PU rate
Studies with less than 50% of patient population at study end
Studies focused on PU risk assessment or singular interventions that prevent PUs
(e.g., special mattresses, skin care items, etc.). These topics are currently covered in a
separate comparative effectiveness review.
C-45
11 intensive care units/ or burn units/ or coronary care units/ or recovery room/ (31083)
12 postoperative complications/ or prosthesis-related infections/ or surgical wound dehiscence/
or surgical wound infection/ (269644)
13 (postoperative$ or post operative$).mp. {mp=title, original title, abstract, name of substance
word, subject heading word, unique identifier} (501444)
14 6 or 7 or 8 or 9 or 10 or 11 or 12 or 13 (643228)
15 5 and 14 (6711)
16 randomized controlled trial.pt. (278973)
17 controlled clinical trial.pt. (79853)
18 randomized controlled trials.sh. (0)
19 random allocation.sh. (66268)
20 double blind method.sh. (103038)
21 single blind method.sh. (13368)
22 16 or 17 or 18 or 19 or 20 or 21 (416935)
23 (animals not human).sh. (4467853)
24 22 not 23 (374747)
25 clinical trial.pt. (452229)
26 exp clinical trials/ (0)
27 (clin$ adj25 trial$).ti,ab. (166370)
28 ((singl$ or doubl$ or trebl$ or tripl$) adj25 (blind$ or mask$)).ti,ab. (103187)
29 placebos.sh. (28486)
30 placebo$.ti,ab. (118661)
31 random$.ti,ab. (460194)
32 research design.sh. (57708)
33 25 or 26 or 27 or 28 or 29 or 30 or 31 or 32 (924508)
34 33 not 23 (807020)
35 34 or 24 (836845)
36 15 and 35 (1138)
37 exp Myocardial Infarction/ (124416)
38 exp Hospitalization/ (120073)
39 exp Inpatients/ (8203)
40 exp Cerebrovascular Accident/ (55676)
41 cerebrovascular disorders/ or brain ischemia/ or exp intracranial embolism and
thrombosis/ or exp intracranial hemorrhages/ (120133)
42 exp myocardial revascularization/ or exp coronary artery bypass/ (64411)
43 37 or 40 or 41 or 42 (330035)
44 5 and 43 (4681)
45 35 and 44 (689)
46 45 not 36 (556)
47 38 or 39 (126877)
48 5 and 47 (1329)
49 35 and 48 (243)
50 49 not (36 or 46) (130)
51 exp Hypoglycemia/ci, ep, et {Chemically Induced, Epidemiology, Etiology} (9091)
52 1 or 2 or 4 (271238)
53 51 and 52 (5827)
C-47
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
14 and 53 (251)
43 and 53 (52)
47 and 53 (89)
54 or 55 or 56 (362)
57 not (36 or 46 or 49) (312)
exp Hypoglycemia/ (18273)
52 and 59 (10136)
14 and 60 (400)
43 and 60 (101)
47 and 60 (126)
61 or 62 or 63 (580)
64 not 57 (218)
65 not (36 or 46 or 49) (193)
exp Costs and Cost Analysis/ (145993)
15 and 67 (51)
exp Economics/ (413412)
ec.fs. (261917)
69 or 70 (488778)
15 and 71 (82)
72 not 68 (31)
from 73 keep 1-31 (31)
C-48
15 5 and 14 (5822)
16 randomized controlled trial.pt. (246761)
17 controlled clinical trial.pt. (77022)
18 randomized controlled trials.sh. (52472)
19 random allocation.sh. (59778)
20 double blind method.sh. (94781)
21 single blind method.sh. (11591)
22 16 or 17 or 18 or 19 or 20 or 21 (418296)
23 (animals not human).sh. (4261058)
24 22 not 23 (382274)
25 clinical trial.pt. (444490)
26 exp clinical trials/ (199910)
27 (clin$ adj25 trial$).ti,ab. (139332)
28 ((singl$ or doubl$ or trebl$ or tripl$) adj25 (blind$ or mask$)).ti,ab. (94254)
29 placebos.sh. (26956)
30 placebo$.ti,ab. (106977)
31 random$.ti,ab. (394441)
32 research design.sh. (50582)
33 25 or 26 or 27 or 28 or 29 or 30 or 31 or 32 (887876)
34 33 not 23 (778635)
35 34 or 24 (798240)
36 15 and 35 (979)
37 exp Myocardial Infarction/ (115916)
38 exp Hospitalization/ (107713)
39 exp Inpatients/ (6673)
40 exp Cerebrovascular Accident/ (44100)
41 cerebrovascular disorders/ or brain ischemia/ or exp intracranial embolism and
thrombosis/ or exp intracranial hemorrhages/ (112871)
42 exp myocardial revascularization/ or exp coronary artery bypass/ (56866)
43 37 or 40 or 41 or 42 (300510)
44 5 and 43 (4061)
45 35 and 44 (657)
46 45 not 36 (544)
47 38 or 39 (113294)
48 5 and 47 (1078)
49 35 and 48 (202)
50 49 not (36 or 46) (114)
51 exp Hypoglycemia/ci, ep, et {Chemically Induced, Epidemiology, Etiology} (8651)
52 1 or 2 or 4 (250082)
53 51 and 52 (5520)
54 14 and 53 (180)
55 43 and 53 (41)
56 47 and 53 (65)
57 54 or 55 or 56 (276)
58 57 not (36 or 46 or 49) (254)
59 exp Hypoglycemia/ (17277)
C-49
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
52 and 59 (9545)
14 and 60 (285)
43 and 60 (86)
47 and 60 (97)
61 or 62 or 63 (445)
64 not 57 (169)
65 not (36 or 46 or 49) (152)
limit 36 to english language (865)
limit 46 to english language (476)
limit 50 to english language (104)
limit 58 to english language (215)
limit 66 to english language (113)
from 67 keep 1-865 (865)
from 68 keep 1-476 (476)
from 69 keep 1-104 (104)
from 70 keep 1-215 (215)
from 71 keep 1-113 (113)
An additional search for adverse effects used the above strategy through line 71, followed
by:
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
Database: EBM Reviews - Database of Abstracts of Reviews of Effects <3rd Quarter 2008>
1 (insulin or hypoglycemic agent$ or hypoglycaemic agent$ or glycemic control or glycaemic
control).mp. {mp=title, full text, keywords} (163)
2 ((critical$ adj6 ill$) or critical care or icu or intensive care or burn unit$ or coronary care).mp.
{mp=title, full text, keywords} (327)
3 (postoperative$ or post operative$).mp. {mp=title, full text, keywords} (705)
C-50
4 2 or 3 (973)
5 1 and 4 (6)
6 from 5 keep 1-6 (6)
Database: EBM Reviews - Cochrane Central Register of Controlled Trials <3rd Quarter
2008>
1 (insulin or hypoglycemic agent$ or hypoglycaemic agent$ or glycemic control or glycaemic
control).mp. {mp=title, original title, abstract, mesh headings, heading words, keyword} (14093)
2 ((critical$ adj6 ill$) or critical care or icu or intensive care or burn unit$ or coronary care).mp.
{mp=title, original title, abstract, mesh headings, heading words, keyword} (6526)
3 (postoperative$ or post operative$).mp. {mp=title, original title, abstract, mesh headings,
heading words, keyword} (36957)
4 2 or 3 (42208)
5 1 and 4 (541)
6 from 5 keep 1-541 (541)
SECTION B. Methods
Search strategy
We searched MEDLINE and the Cochrane Database of Systematic Reviews for literature
published from database inception through January 2010 and obtained additional articles from
consultation with experts and from reference lists of pertinent studies, reviews, and editorials.
We updated this search for the purposes of this report in October 2011. Appendix Table 1
provides the search strategies in detail. We searched clinicaltrials.gov for information about
unpublished studies. All citations were imported into an electronic database (EndNote X2,
Thomson Reuters, New York, NY).
Study selection
Three investigators reviewed the abstracts of citations identified from literature searches. Fulltext articles of potentially relevant abstracts were retrieved for further review. Eligible articles
were published in English and provided primary data on the use of IIT in hospitalized patients.
We excluded studies that evaluated fixed-dose insulin and glucose-insulin-potassium (GIK)
infusions.
To evaluate the efficacy of and hypoglycemia risk associated with IIT in hospitalized patients,
we considered randomized controlled trials that reported at least one of the following
prespecified outcomes: mortality, cardiovascular events, congestive heart failure, disability,
wound infection, sepsis, or renal failure requiring hemodialysis. We defined perioperative trials
as those in which IIT was begun pre-, intra-, or immediately post-operatively and discontinued
less than 24 hours post-operatively.
C-51
Because the safety of IIT may vary based on intervention and implementation characteristics, we
evaluated hypoglycemia rates in controlled and uncontrolled studies of IIT protocols, even if
they did not report other health outcomes (study selection details in Appendix Table 1).
To assess the risk of hypoglycemia associated with IIT, we included controlled and uncontrolled
studies that evaluated IIT protocols in hospitalized patients, even if they did not report health
outcomes. We excluded IIT studies that did not report rates of hypoglycemia [1-12]. In order to
avoid studies with potential selection bias, we excluded prospective cohort studies in which
patients were not consecutively enrolled or in which there was excessive loss to follow-up [1221]. Because tight glycemic control strategies require personnel training and institutional
acceptance, we excluded studies in which the intervention was evaluated over a short period of
time (defined as 6 months or less) as we felt these studies were less likely to provide externally
valid results [22-27].
Data extraction and quality assessment
From each study, we abstracted the following: study design, objectives, setting, demographics
(sex, age, baseline morbidity), subject eligibility and exclusion criteria, number of subjects, years
of enrollment, duration of follow-up, study and comparator interventions, method used to
monitor blood glucose, target range for blood glucose control, outcomes measured, analytic
method used, variables adjusted in the analysis, results of the study and mean blood glucose
achieved in each group, information on concomitant therapy/nutrition, occurrence of
hypoglycemia in each group, and any other adverse events.
The quality of each study was rated as good, fair, or poor based on U.S. Preventive Task Force
Service criteria [28]. When reviewers disagreed about quality rating, consensus was reached
through discussion with all authors.
Meta-analysis
We conducted meta-analyses using IIT studies identified in our original search through January
2010. Studies identified from the update search through October 2011 were described, but not
included in these meta-analyses. The primary outcome of interest was short-term mortality,
defined as mortality occurring within 28 days or during the ICU or hospital stay. If studies
reported more than one of these outcomes, we preferentially used 28-day mortality for the
analysis, followed by hospital- or ICU-mortality. We conducted a sensitivity analysis based on
short-term mortality definition. Secondary outcomes included 90- or 180-day mortality,
infection, length of stay, and hypoglycemia. For each outcome, we abstracted the number of
events and total subjects from each treatment arm and obtained a pooled estimate of relative risk
(RR) using a random effects model [29]. Statistical heterogeneity was assessed by Cochrans Q
test and I2 statistic [30]. All analyses were performed using Stata 10.0 (StataCorp, College
Station, TX, 2007).
We conducted prespecified subgroup analyses comparing ICU with non-ICU studies, and
sensitivity analyses on the following aspects: 1) the proportion of diabetic patients included,
using 25% as a cut-point based on a natural division in the included studies; 2) mean blood
C-52
glucose achieved in the intervention group, using 6.7 mmol/L (120 mg/dL) as the cut-point since
a lower threshold (6.1 mmol/L, 110 mg/dL) would have yielded only one study; and 3) study
quality.
Study yield
From our initial search through January 2010, we identified 3,055 titles and abstracts of which
461 articles selected for full-text review. We included 31 trials conducted among critically ill
patients, patients with acute MI or stroke, or perioperative patients. We also found 29 insulin
protocol studies not reporting health outcomes. Our update search through October 2011
identified an additional 331 titles and abstracts of which 40 articles were selected for full-text
review. We included 2 trials conducted in neurologic intensive care units, 1 trial in gastrectomy
patients, and 1 trial of a subcutaneous insulin regimen in general surgical ward patients. We also
found 10 insulin protocol studies not reporting health outcomes. The yield is summarized in
Figure 1.
C-53
C-54
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
C-55
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
C-56
#1
#2
#3
#4
Search string
((contrast medium OR contrast media OR contrast dye OR radiographic
contrast OR radiocontrast media OR radiocontrast medium OR contrast
agent*) AND (kidney diseases/ci OR kidney/ae OR kidney/de OR nephritis OR
nephropath* OR nephrotox* OR renal insufficiencies[mh] OR renal insufficienc*
OR diabetic nephropathies[mh] OR creatinine OR kidney injury OR kidney
dysfunction OR renal dysfunction)) OR (contrast induced OR contrast
associated AND (renal OR kidney OR nephropath* OR nephrotox*)) OR
(contrast media/ae AND (kidney diseases[majr] OR kidney[majr] OR diabetic
nephropathy[majr]))
#1 limited to Randomized Controlled Trials, English language, publication date
since 1/1/2001
#1 limited to Systematic Reviews, English language, publication date since
1/1/2001
#3 limited to publication date since 1/1/2007
Results
5217
193
53
32
SECTION B. Methods
Based on the large number of systematic reviews identified by the above search, we opted to
perform a systematic meta-review of the existing systematic reviews. We included only
systematic reviews published since January 1, 2007. Two authors independently reviewed the 32
reviews identified through the PubMed search to identify systematic reviews and meta-analyses.
All of the systematic reviews identified (N=20) were assessed for methodologic quality by two
reviewers who independently completed the AMSTAR checklist. Disagreements in this process
were resolved by consensus. The included systematic reviews were grouped according to the
specific CI-AKI preventive intervention studied, and were summarized narratively.
AND
effectiv* OR implement* OR success* OR fail* OR utiliz* OR adopt*
NUMBER OF RESULTS: 1679
====================================================================
DATABASE SEARCHED & TIME PERIOD COVERED:
CINAHL: 2000-8/16/2011
SEARCH STRATEGY:
rapid response team OR rapid response teams OR rapid response system OR rapid
response systems OR medical emergency team OR medical emergency teams OR critical
care outreach team OR critical care outreach teams OR patient at-risk team OR patient atrisk teams OR patient at risk team OR patient at risk teams OR emergency medical team
OR emergency medical teams
AND
effectiv* OR implement* OR success* OR fail* OR utiliz* OR adopt*
OR
patient care AND team* AND (emergency OR emergencies OR rapid OR critical care)
AND
effectiv* OR implement* OR success* OR fail* OR utiliz* OR adopt*
Search modes - Phrase Searching (Boolean)
NUMBER OF RESULTS: 333
====================================================================
DATABASE SEARCHED & TIME PERIOD COVERED:
EMBASE: 2000-11/4/2011
SEARCH STRATEGY:
rapid response team/exp OR rapid response team OR rapid response teams/exp OR rapid
response teams OR rapid response system/exp OR rapid response system OR rapid response
systems/exp OR rapid response systems OR medical emergency team/exp OR medical
emergency team OR medical emergency teams/exp OR medical emergency teams OR
critical care outreach team OR critical care outreach teams OR patient at-risk team OR
patient at-risk teams OR patient at risk team OR patient at risk teams OR emergency
medical team OR emergency medical teams OR ((patient care NEAR/3 team*) AND
(emergency OR emergencies OR rapid OR critical care))
AND
effectiv* OR implement* OR success* OR fail* OR utiliz* OR adopt*
NUMBER OF RESULTS: 594
C-58
====================================================================
DATABASE SEARCHED & TIME PERIOD COVERED:
Cochrane: 2000-11/4/2011
SEARCH STRATEGY:
rapid response team OR rapid response teams OR rapid response system OR rapid
response systems OR medical emergency team OR medical emergency teams OR critical
care outreach team OR critical care outreach teams OR patient at-risk team OR patient atrisk teams OR patient at risk team OR patient at risk teams OR emergency medical team
OR emergency medical teams:ti,ab,kw or patient care AND team* AND (emergency OR
emergencies OR rapid OR critical care) :ti,ab,kw
NUMBER OF RESULTS: 72 (Syst Revs 4, Other Revs 7, Clin Trials 55, Econ 6)
SECTION B. Methods
PICOTS
Elements
Population
Intervention
Comparator
Outcomes
Timing
Settings
Inclusion/exclusion criteria:
Studies from all countries and languages were included
Effectiveness: Included all studies with a comparison group and at least some component of an
RRS. Critical Care Outreach Team studies were included if they also included a pre-intensive
care unit RRS component (general response to all ward patients). Effectiveness studies were only
included after November 2008, the end date for a high-quality systematic review and metaanalysis.
Implementation: Included qualitative and quantitative studies addressing implementation.
Studies were defined as qualitative research studies if they used a formal qualitative
methodology such as interviews, focus groups, or ethnography
C-59
Studies were defined as quantitative implementation studies if they evaluated the impact of a
change or difference in implementation strategy on utilization of the RRS and/or patient
outcomes.
C-61
SECTION B. Methods
These reviews and studies were reviewed by a surgeon health services researcher with clinical
and quality improvement experience with venous thromboembolism. The synthesis was
narrative.
Date limits
Searched November 8, 2011
Searched November 14, 2011
Searched November 14, 2011
PubMed
Platform/provider
ECRI Institute
Agency for Healthcare Research
and Quality
National Library of Medicine
[pt]
[sb]
[sh]
[tiab]
=
=
=
=
publication type
subset of PubMed database (PreMEDLINE, Systematic, OldMEDLINE)
MeSH subheading (qualifiers used in conjunction with MeSH headings)
keyword in title or abstract
Programs
Study design
Controlled Vocabulary
radiation injuries/prevention and
control[majr]
radiation-protective agents[majr]
radiation monitoring[majr]
radiation dosage[majr]
dose-response relationship,
radiation[majr]
radiation protection[majr]
fluoroscopy/adverse effects[majr]
radiography, interventional/adverse
effects[majr]
radiotherapy/adverse effects[majr]
radiation[majr]
death[majr]
mortality[majr]
wounds and injuries[majr]
burns[mesh]
skin transplantation[mesh]
outcome and process assessment
(health care)[majr]
safety management[majr]
risk assessment[majr]
secondary prevention[mesh]
program development[mesh]
program evaluation[mesh]
health plan implementation[mesh]
randomized controlled trial[pt]
controlled clinical trial[pt]
randomized controlled trials[mh]
random allocation[mh]
double-blind method[mh]
single-blind method[mh]
clinical trial[pt]
clinical trials[mh]
research design[mh:noexp]
comparative study[pt]
evaluation studies[pt]
evaluation studies as topic[mh]
follow-up studies[mh]
prospective studies[mh]
cross-over studies[mh]
meta-analysis[mh]
meta-analysis[pt]
outcomes research[mh]
multicenter study[pt]
C-64
Keywords
radiation
fluoroscop*
injuri*
injury*
death
mortality
injur*
harm
burn*
skin graft
skin transplant
grade 3
grade 4
prevent*
reduc*
initiative*
program*
implement*
clinical trial
clinical trials
comparative study
comparative studies
evaluation study
evaluation studies
Pubmed
English language, human
Set
Number
Concept
Search statement
Programs, preventive or
assessment
Study design
Combine
2000-2011
Total Downloaded
Total Retrieved
Total Included
80
56
10
SECTION B. Methods
Titles and abstracts were reviewed by a health services research methodologist with experience
in both systematic reviews and radiation therapy. Our research was limited to studies
implementing practices (e.g., protocols, technical measures) to reduce radiation exposure to
patients from fluoroscopy and computed tomography-guided diagnostic and interventional
procedures. The focus was on studies published from 2005 to the present that compared
outcomes (e.g., radiation dose, imaging time) following implementation of these practices
compared to a control period when the technologies were not in place. Potential barriers to
implementation, technical difficulty of practices and reported harms from patient safety practices
were also assessed. Included studies were narratively summarized by the author.
C-65
C-66
C-67
SEARCH #4:
DATABASE SEARCHED & TIME PERIOD COVERED:
PubMed 2006-9/15/2011
SEARCH STRATEGY:
patient participation[mh] OR patient participation[tiab] OR patient role OR patients role
OR patient complain* OR patients complain* OR patient reporting[tiab] OR patient
reports[tiab] OR patients reporting[tiab] OR patients reports[tiab]
AND
adverse events OR Iatrogenic Disease/prevention and control OR Medical
Errors/prevention and control or Medical Errors/adverse effects or Safety Management or
Cross Infection/prevention and control OR safe* OR medical error OR medical errors OR
mistake* OR medication error OR medication errors
NUMBER OF RESULTS: 514
=====================================================================
SEARCH #5:
DATABASE SEARCHED & TIME PERIOD COVERED:
Cochrane Database of Systematic Reviews 2006-9/15/2011
SEARCH STRATEGY:
PhysicianPatient Relations OR NursePatient Relations OR Patient Participation OR
Patient Education OR informed consent OR patient reporting OR patient reports OR
patients reporting OR patients reports OR complain* OR patient participa* OR patients
participa* OR patient education OR education of patients OR patient role OR patients
role OR patients role* OR health literacy in Title, Abstract or Keywords
AND
adverse events OR Iatrogenic Disease/prevention and control OR Medical
Errors/prevention and control or Medical Errors/adverse effects or Safety Management or
Cross Infection/prevention and control OR safe* OR medical error OR medical errors OR
mistake* OR medication error OR medication errors OR adverse OR dangerous in Title,
Abstract or Keywords
NUMBER OF RESULTS: 909
=====================================================================
SEARCH #6:
DATABASE SEARCHED & TIME PERIOD COVERED:
PubMed 2006-9/19/2011
SEARCH STRATEGY:
C-68
C-69
SEARCH #9:
DATABASE SEARCHED & TIME PERIOD COVERED:
PubMed 2000-10/5/2011
SEARCH STRATEGY:
adverse events OR Iatrogenic Disease/prevention and control OR Medical
Errors/prevention and control OR medication errors/prevention and control OR Medical
Errors/adverse effects OR Safety Management OR Cross Infection/prevention and control
OR infection control
AND
Physician-Patient Relations OR Nurse-Patient Relations OR Patient Participation OR
Patient Education as Topic OR Patient-Centered Care OR patient reporting OR patientempowering OR patient empowerment OR patient partnership OR patient activation or
patient self-effectiveness or patient involvement
NUMBER OF RESULTS: 1499
=====================================================================
SEARCH #10:
DATABASE SEARCHED & TIME PERIOD COVERED:
CINAHL 2000-10/24/2011
SEARCH STRATEGY:
(TI (Physician n3 Patient n3 relation*) OR (Nurse n3 Patient n3 relation*) OR Patient n3
Participat* OR Patient n3 Education OR Patient-Centered Care OR patient reporting OR
patient n2 empower* OR patient n3 partner* OR patient activation OR patient selfeffectiveness OR patient n3 involv*
OR AB (Physician n3 Patient n3 relation*) OR (Nurse n3 Patient n3 relation*) OR Patient n3
Participat* OR Patient n3 Education OR Patient-Centered Care OR patient reporting OR
patient n2 empower* OR patient n3 partner* OR patient activation OR patient selfeffectiveness OR patient n3 involv*
OR MW (Physician n3 Patient n3 relation*) OR (Nurse n3 Patient n3 relation*) OR Patient n3
Participat* OR Patient n3 Education OR Patient-Centered Care OR patient reporting OR
patient n2 empower* OR patient n3 partner* OR patient activation OR patient selfeffectiveness OR patient n3 involv*)
AND
adverse events OR iatrogenic disease OR medical errors OR medication errors OR
medication error OR medical error OR (cross infection AND prevent*) OR safety
management OR infection control
NUMBER OF RESULTS: 1283
=====================================================================
SEARCH #11:
C-70
C-71
MeSH descriptor Safety Management, this term only OR drug toxicity OR MeSH descriptor
Medical Errors explode all trees OR MeSH descriptor Infection explode all trees with qualifier:
PC (Prevention & Control)
NUMBER OF RESULTS: 81 (Cochrane Reviews [5] | Other Reviews [4] | Clinical Trials
[71] | Methods Studies [0] | Technology Assessments [0] | Economic Evaluations [1] |
Cochrane Groups [0])
SECTION B. Methods
PICOTS
Elements
Population
Intervention
Comparator
Outcomes
Timing
Settings
Patients in inpatient healthcare settings (adult and pediatric) and their family members
Any intervention to encourage patient involvement in safety, including reporting
adverse outcomes or errors
Usual practice
Effectiveness of the intervention
Before and after the intervention
Hospitals
Inclusion/exclusion criteria:
- Only English-language studies from the US, UK, Canada, and Australia were included in the
present review, due to potentially significantly different cultural issues in patient engagement in
their health care outside of these countries, as well as potential differences in tools for promoting
engagement.
-Included studies were required to focus on hospital care settings (e.g., intensive care units)
patient engagement in safety in the home setting would be difficult to differentiate from patient
self-management of their medications and care, when providers are not present.
-Only systematic reviews focusing on effectiveness and prospective, controlled studies were
included.
C-73
C-74
Timing
Settings
Inclusion/exclusion criteria:
-Only English-language studies from the US, UK, Canada, and Australia were included in the
present review. While there are a growing number of studies that have translated Englishlanguage surveys of culture into other languages, there is still limited evidence that construct
validity of such measures is comparable across samples.
-Included studies were required to focus on in-patient units within hospital care settings (e.g.,
intensive care units). Studies in the operating room, radiology, and other non-inpatient units were
not included in this review.
- Included studies had to report on measures of culture over at least two points in time.
- Included studies were also required to use a psychometrically valid measure of safety culture
that is present in the peer-reviewed literature. As a guideline, studies that utilized one of the
following measures of patient safety culture were included: hospital survey on patient safety
culture (HSOPSC) / HSOPS, Hospital Survey on Patient Safety Patient Safety Climate/ patient
safety climate in health care organizations (PSCHO), Safety Climate Survey (SCS) Safety
Attitudes Questionnaire (SAQ) Hospital safety climate scale (HSC) Operating Room
Management Attitudes Questionnaire (ORMAQ) Hospital Survey on Patient Safety
Stanford/PSCI Culture Survey Safety Climate Scale MSSA, Medication Safety Self Assessment
HTSSCS, Hospital Transfusion Service Safety Culture Survey Manchester Patient Safety
Framework.
- Interventions had to target practicing health care professionals or para-professionals. Studies
examining PSPs aimed at medical or nursing students, or otherwise including only education
outside of a clinical setting, were not included.
- The stated purpose of the PSP described had to specifically include aims to improve culture.
C-75
-Studies specifically targeting patient safety culture were included. Studies of other types of
culture (e.g., general organizational culture) were not included.
-Only prospective studies were included.
C-76
*1-4The figure shows the flow of articles through the review process. With one exception we did not include any additional crosssectional studies of association. The one exception is detailed in the text.
References
1.
3.
2.
4.
C-77
C-78
if focusing on a specific clinical specialty or care setting. For example, the major subheading
search for Delayed Diagnosis in MEDLINE would likely capture studies with a primary focus
related to the current review, but may not capture a study primarily focusing on particular
clinical processes such as x-ray review for injuries. The extensive number of possible search
terminology combinations using specific clinical domains (e.g., radiology), care settings (e.g.,
critical care), and intervention types (e.g., double review) where diagnostic errors may occur
was beyond the scope of the current review. However, with our extensive reference review we
expect that our included studies represent at a minimum a reasonable probe of the literature for
certain clinical specialties, care settings, and intervention types likely to be of importance to
reducing diagnostic errors (e.g., evaluations of an additional reviewer of radiology reports added
to the diagnostic pathway; laboratory-focused interventions).
Table 1, Chapter 35. MEDLINE search strategy
The final search was performed on October 10, 2011. The search was conducted using PubMed MEDLINE
and was limited to English language publications.
The following two search strategies were used in conjunction with one another to identify articles with
diagnostic error reduction interventions published between 1980 and 2011.
Search Strategy A: (((Diagnostic Errors[Majr]) OR Delayed Diagnosis[Majr])) AND (interven*[tiab] OR
intervention studies[mh] OR model*[ti] OR strateg* OR improv*[ti] OR implementation*[tiab] OR practices
OR random* OR controlled clinical trial[pt] OR program[tiab] OR programs[tiab] OR programme[tiab] OR
programmes[tiab] OR (program evaluation[mh] AND treatment outcome[mh]) OR systematic review*)
Search Strategy B: (((Diagnostic Errors[Majr]) OR Delayed Diagnosis[Majr])) AND (Affect[Mesh] OR
Clinical Competence[Mesh] OR Communication[Mesh] OR Continuity of Patient Care[Mesh] OR
Decision Making[Mesh] OR Decision Making, Organizational[Mesh] OR Decision Support Systems,
Clinical[Mesh] OR Decision Support Techniques[Mesh] OR Human Engineering[Mesh] OR
Judgment[Mesh] OR Medical Informatics[Mesh] OR Medical Records Systems, Computerized[Mesh]
OR Mental Recall[Mesh] OR Organizational Culture[Mesh] OR Patient Access to Records[Mesh] OR
Patient Participation[Mesh] OR Feedback[Mesh] OR Forms and Records Control/standards[Mesh] OR
Guidelines as Topic[Mesh] OR Health Knowledge, Attitudes, Practice[Mesh] OR Health Literacy[Mesh]
OR Health Records, Personal[Mesh] OR Physician-Patient Relations[Mesh] OR Physicians Practice
Patterns[Mesh] OR Problem Solving[Mesh] OR Professional-Patient Relations[Mesh] OR Reminder
Systems[Mesh] OR Systems Analysis[Mesh] OR Time Factors[Mesh] OR Truth Disclosure[Mesh] OR
Knowledge Bases[Mesh] OR cognitive error OR bias OR metacognition)
2
Search Strategy B replicates that used by Singh et al , though for a wider date range. Citations from their
search dates (2000-2009) were removed from both Search A and B, so as not to duplicate their work.
References
1.
2.
C-79
SECTION B. Methods
Figure 2, Chapter 35. Diagnostic errors systematic review flow chart
Date limits
1996 September 21, 2011
1996 September 21, 2011
Searched September 22, 2011
Searched September 13, 2011
Platform/provider
OVID SP
OVID SP
PubMed
AHRQ
C-80
Controlled Vocabulary
ae.fs.
Adverse outcome/
exp Cross infection/
Hospital infection/
Iatrogenic disease/
exp Medical errors/
Chart review
Concurrent review/
Documentation/
Drug utilization review/
Medical audit/
Medical records/
Medical record review/
Patient safety/
Safety/
Safety management
Patient safety
C-81
Keywords
Administration
Adverse events
Diagnostic
Error$
Iatrogenic
Medical
Medication
Nosocomial
Chart review
Case finding
Computerized surveillance
Patient safety
Patient Safety Organization
PSO
Concept
Reporting Systems
Controlled Vocabulary
Medline/Embase
Keywords
Trigger tools
Error reporting
Government reporting
Institutional reporting
Reporting
Medline/Embase
Electronic medical record/
Hospital information systems/
PSNet
Computerized adverse event detection
Electronic health records
Automated
Automatic
Computer-based detection
Data mining
Electronic adj2 screen$
Surveillance
Trigger tool$
Embase/Medline/Premedline
English language, human, remove overlap
Set
Number
1
Concept
Adverse events
2
3
4
5
6
Combine sets
Direct observation
Chart review
8
9
10
11
12
13
Combine sets
Combine sets trigger
tools
Search statement
Ae.fs. or exp cross infection/ or iatrogenic disease/ or exp
medical errors/ or adverse outcome/ or hospital infection/ or
iatrogenic disease/
Iatrogenic or nosocomial or (hospital adj acquired) or ((medical
or medication or diagnostic or administration) adj2 error$)
1 or 2
Direct observation or (executive adj walk$)
Chart review or chart$.ti. or case finding
Concurrent review/ or documentation or drug utilization review/
or medical audit/ or medical records/ or medical record review/
or utilization review/
5 or 6
(electronic medical record/ or hospital information systems/ )
and (data mining or automated or automatic or surveillance)
Electronic adj2 screen$
Computer-based detection
Trigger tool$
8 or 9 or 10 or 11
3 and (4 or 7 or 12)
C-82
Concept
#
downloaded
Search statement
Browsed categories:
Computerized adverse event detection Governmental reporting
Institutional reporting
Non-governmental reporting
Patient safety indicators
Total Downloaded
Total Retrieved
250
176
Total Included
SECTION B. Methods
Inclusion/Exclusion Criteria
Healthcare organizations have been using a wide array of methods to detect patient safety
problems. These methods include incident reporting, direct observation of patient care, chart
review, analysis of malpractice claims, patient complaints and reports to risk management,
executive walk rounds, trigger-tool use, patient interviews, morbidity and mortality conferences,
autopsy, and clinical surveillance. Many of these methods (e.g., trigger tools) can be further
categorized by the targeted problems (e.g., medication-related medical errors or iatrogenic
infections), tools, algorithms, and data source used. Given the limited time frame for this review,
we focus this chapter on general approaches to detecting patient safety problems that involve
using multiple methods (e.g., incident reporting, executive walk rounds, clinical surveillance,
chart review, and trigger tools) to collect data. We primarily reviewed studies that compared the
utilities of different methods, because we believe that understanding the strengths and
weaknesses of different methods is most relevant to decision makers who need to form an
effective strategy for monitoring patient safety problems for their organizations. Comparison
studies that used one method (e.g., chart review) as a gold standard to validate another method
(e.g., incident reporting) were not included for this chapter, because, in essence, these studies
still focused on one individual method (i.e., the method being validated).
Only full published studies were considered for review (meeting abstracts were excluded). Only
English-language publications were eligible for inclusion. For the effectiveness and harms of the
PSP, we considered including studies of any design (e.g., systematic reviews, randomized
controlled trials, non-randomized controlled trials, prospective and retrospective observational
studies, surveys) that may provide relevant data. For the implementation and context of the PSP,
we primarily abstracted data from the effectiveness or safety studies being reviewed.
Risk of Bias and Strength of Evidence
We did not assess the risk of bias of the included studies or the overall strength of evidence. The
body of evidence consists of one systematic review and several studies that compared the types
and numbers of patient safety problems identified using different methods. No adequately
C-83
validated tool is available for assessing this type of comparison studies or for assessing the
overall strength of evidence that mixes such comparison studies with a systematic review.
C-84
C-85
=====================================================================
DATABASE SEARCHED & TIME PERIOD COVERED:
PsycINFO 1990-8/15/2011
SEARCH STRATEGY:
transitional care OR readmission OR readmit* OR patient discharg* OR ( (length of stay OR
length-of-stay) AND discharge plan* ) OR emergency service,hospital OR emergency
service* OR emergency room*
AND
discharg* OR home visits OR home visit* OR house calls OR house call*
AND
home OR house
Search modes - Phrase Searching (Boolean)
OR
transitional care
OR
readmit* OR readmission* OR rehospitali* OR re-hospitali*
AND
home OR house
NUMBER OF RESULTS: 617
NUMBER AFTER REMOVAL OF DUPLICATES: 407
=====================================================================
DATABASE SEARCHED & TIME PERIOD COVERED:
Embase 1990-9/12/2011
LANGUAGE:
English
SEARCH STRATEGIES:
SEARCH #1 (S. Rennke Strategy)
transitional care OR care NEAR/3 transition* OR discharge NEAR/3 plan* OR patient
NEAR/3 discharg* AND [humans]/lim AND [english]/lim AND [1990-2012]/py 8,919 View |
Edit
#1transitional AND care OR care AND transition* OR transition* AND in AND care OR
transition* OR discharge AND plan* OR patient AND discharg*
AND
C-86
C-87
=====================================================================
DATABASE SEARCHED & TIME PERIOD COVERED:
CINAHL 1990-9/12/2011
LANGUAGE:
English
SEARCH STRATEGY:
transitional care OR care transition* OR rehospital* OR discharge plan* OR patient care team
OR multidisciplinary team OR ( (emergency service, hospital OR emergency service* OR
emergency room*) AND readmi* ) OR medical error OR medical errors OR medical mistake*
OR medication errors OR adverse event*
AND
postdischarge OR post-discharge OR readmi* OR patient discharg* OR continuity of patient
care OR continuity of care OR home visits OR home visit OR house calls OR house
call OR aftercare OR ( (avoid* OR reduc*) AND (admit* OR admission ) )
AND
interven* OR strateg* OR implement* OR practices OR random* OR controlled clinical trial
OR controlled clinical trials OR program* OR systematic review* OR ( program evaluation
AND outcome* ) OR TI ( model* OR improv* )
NUMBER OF RESULTS: 1235
NUMBER OF RESULTS AFTER REMOVING DUPLICATES: 588
SECTION B. Methods
C-88
Figure 1, Chapter 37. Theoretical model for the effectiveness of patient safety practices for
transitional care
C-89
C-90
of 66 articles (20%) were reviewed. Thirteen of these studies reported T2 or T3 outcomes and
were included for review of empirical literature.
Other Literature Capture Methods
Secondary or chain-method literature prompted full-text review of an additional 15 empirical
articles, 13 of which are presented in the empirical review. This search method also resulted in
16 articles that were used to provide supplemental information about simulation. Experts
recommended an additional 22 articles, all of which are presented in the review. Many of these
articles included important theoretical work or important resources for those looking to
implement simulation. Experts also recommended important T1 studies for inclusion in the
review (n = 9).
Literature Totals
The empirical literature search resulted in 833 titles, of which 174 (21%) were reviewed in fulltext for inclusion in the review. The final reference list in Simulation and Patient Safety was
ultimately comprised of 45% (n = 40) literature directly captured by database searches, 32%
secondary literature (n = 27), and 25% literature recommended by practitioners with expertise in
simulation (n = 25). Eight additional articles served an explanatory function (e.g., clinical
rationale for placing a central venous catheter), and these articles were retrieved from free-text
searches in PubMed and Google Scholar.
SECTION B. Methods
The methodology for identifying empirical literature in this review involved three primary
mechanisms. In the first mechanism, structured search strategies for PubMed and the Cochrane
Library Databases provided the initial capture of simulation references. These searches were
limited to meta-analyses or systematic reviews, and to studies that were empirical in nature.
Theoretical pieces and commentary publications were not excluded in these search strategies, but
these publication types were not a focus of this mechanism to capture literature. The search
strategies were limited to general terminology (e.g., simulation) rather than specific terms that
might be required if one wished to perform a systematic review of simulation practices. Specific
simulation search terms might include the clinical specialty under investigation (e.g.,
anesthesiology), the procedure under investigation (e.g., laparoscopic cholecystectomy), the
purpose of the simulation (e.g., curriculum), or the fidelity and specific simulation exercise
(e.g., mannequin). Due to the brief nature of the current review, and to the extensive possible
combinations of these specific terms, this search mechanism identified literature through general
terms rather than exhaust these combinations. In the second mechanism, practitioners with
expertise in simulation were asked to provide recommendations on seminal work in simulation,
current key articles, empirical research on simulation and patient safety, areas of focus most
pertinent to implementing simulation, and guidance in terms of implementing simulation.
Secondary or chain-method capture of references provided the third mechanism to inform this
review. That is, reference lists in articles captured from the first two mechanisms provided
additional literature. Specifics and resulting citations are provided below on each of these search
mechanisms.
C-92
Empirical articles were held to a translational science paradigm for inclusion. Articles were
given priority if they reported outcomes from care provided to actual patients, or from actual care
system interventions. In terms of translational science, these are T2 or T3 simulation studies,
respectively. Due to the nature of simulation, selected studies that did not report outcomes from
care provided to actual patients (i.e., T1 or within the lab studies) were included if experts
recommended their inclusion to adequately represent the applications of simulation. All efforts
were made to remain inclusive across clinical specialties, no preference was assigned to specific
procedures or care practices. There is a section on central venous catheter placement that
provided an in-depth look at simulation to improve patient safety. This literature was selected
for in-depth presentation because (1) as a specific topic it had the greatest number of articles
captured in our review with outcomes reported at both the T2 and T3 level, and (2) this particular
line of research included analyses of costs for those looking to implement simulation.
C-93
SECTION B. Methods
Systematic reviews and articles were abstracted by health services researchers with expertise in
the topic area, and the results were narratively synthesized.
Clinical Risk
Management
Health Devices
Healthcare Risk
Control
Health Technology
Assessment
Information Service
Date limits
2005 November 9, 2011
2005 November 9, 2011
C-94
Platform/providers
www.pubmed.gov
www.ecri.org
Concept
Tubing
misconnections
Reviewed related
citations for PMIDs
Search statement
Equipment design[mh] OR equipment safety[mh]
#1 AND (tube* OR tubing OR catheter*)
#2 AND (connector* OR connection* OR misconnect*)
Misconnections[ti]
(luer*[tw] OR tubing[tw]) AND misconnection*
#2 OR #3 OR #4 OR #5
16610452
17090266
18491692
16739386
Concept
Tubing
Misconnections
Relevant UMDNS
codes
Combine sets
Search statement
Misconnection OR connectors
17-501 Intravenous Line Connectors
OR
11-726 Fittings/Adapters
OR
16-795 Fittings/Adapters, Pin-indexed
OR
11-729 Fittings/Adapters, Luer
1 OR 2
C-95
C-96
D-1
Evidence Tables for Chapter 3. High-Alert Drugs: Patient Safety Practices for Intravenous
Anticoagulants
Table 1, Chapter 3. Evidence table
Author, year
Description of
PSP
Study Design
Theory or Description of
Logic Model Organization
Contexts
Sample Size
1
Baird, 2001
Fanikos,
2
2007
Fraipont,
3
2003
Multi-component
A single protocol
for heparin
administration was
developed by a
team of doctors,
nurses and a
pharmacists.
Pre-post
Leadership :
Protocol
development
team
Received optimal
Not reported Not reported
bolus dose Results: 5
(8.6%) pre vs: 10
(90%) post
Statistics: NR
Mean time to
anticoagulation
Results: 34 hrs vs 63
+- 49 hours
Statistics: NR
Not reported Not reported
Results:
Anticoagulation
medication errors: 49
before; 48 after
Statistics: NS
D-2
Comments
Results post
implementation
only: Prevented 10fold overdose in 40
patients; 100-fold
overdose in 40
patients; and >100fold overdose in 10
patients; similar
results for under
doses; heparin was
#4 most common
drug generating
alerts
Author, year
Oyen, 2005
Description of
PSP
Study Design
Theory or Description of
Logic Model Organization
Sample Size
Multi-component
Computerized
Pre-post
Logic model
nomogram for
419 nomogram,
acute coronary
98 comparison
syndromes
Cardiovascular
services (88
beds) at a 1300bed teaching
hospital
Contexts
Implementation Ot described
tools: Dosing
based on US
organization
guidelines
538-bed
Organizational
community
characteristics :
teaching hospital multidisciplinary
- expanded to all team and
units
relationship with
BCMA and IID
vendors to
develop
interoperability
between
systems
Leadership :
Executive
sponsorship,
Direction and
support of
pharmacy and
therapeutics
committee
Implementation
tools : Nurse
education
D-3
preparation,
pilot, validation,
and expansion;
extensive
software design
and testing
before
introduction to
patient care
Telemetry drug
library monthly
compliance
Results: 56.5 pre to
72.1 post
Statistics: p<0.001
Number of telemetry
manual pump edits
Results: 56.9 to 14.7
Statistics: p<0.001
Comments
Complications not
reported;
discussion that on
prior paper
nomogram,
clinicians deviated
over 50% of the
time by adjusting
doses; program
provided feedback
and performed
calculations;
computerization
allowed
individualized
protocol for acute
coronary
syndromes
similar decrease in
medical-surgical
drug library results;
reduction in
monthly reported
intravenous
heparin errors (28
to 17, NS); cost:
24.8% reduction
(23.4 sec onds) in
the mean nursing
time for pump
programming; 90%
compliance
Author, year
Toth, 2002
Description of
PSP
Study Design
Theory or Description of
Logic Model Organization
Contexts
Sample Size
Multi-component
Weight-based
RCT
Not reported Neurology ward,
nomogram for
Canada
heparin dosing in 206 patients
TIA and/or stroke.
Zimmermann, Weight-based
7
2003
heparin nomogram
for patients with
acute coronary
syndromes
Pre-post
Weight-based
nomogram was
based on other
nomograms in
literature;
dosage based
on absolute
weight. Weight
and aPTT
determined later
adjustment in
infusion rate.
84 patients
weight-based,
89 patients in
non-weightbased
D-4
Comments
Doctor completed
nomogram; bolus
provided if
indicated.Initial
heparin found by
nomogram. Nurses
changed heparin
from aPTT results
by following
nomogram. Also,
significantly fewer
calls to house staff
and mistakes made
in nomogram
group. Time to
discontinue
heparin:4 02.8 vs.
4.63.8; P=0.33;
94% of staff
preferred use of
nomogram
Adherence to
nomograms was
good (not
described in detail)
References
1.
2.
3.
4.
5.
6.
7.
D-5
Evidence Tables for Chapter 7. Barrier Precautions, Patient Isolation, and Routine
Surveillance for the Prevention of Healthcare-Associated Infections: Brief Update
Review
Table 1, Chapter 7. Evidence table
Reference
McGinigle KL,
2008(33)
Study Population/Setting
Systematic literature review evaluated
studies assessing the use of active
surveillance to reduce MRSA-related
morbidity, mortality and costs
Systematic literature
review
Backman C, 2011(35)
Systematic literature
review
Systematic literature
review
D-6
Summary/Main Contribution
The investigators did not identify any randomized, controlled
trials. They reviewed 16 observational studies and 4 economic
analyses. Only 2 studies included control groups. None of the
studies were of good quality. Thirteen studies reported
decreases in the incidence of MRSA infections associated with
the use of active surveillance
Existing evidence may favor the use of active surveillance, but
the evidence was of poor quality and the investigators could not
make definitive recommendations.
32 articles were assessed, of which 53% used surveillance; 75%
implemented infection control precautions to prevent
transmission; 22% introduced environmental measures; 28%
used patient decolonization; 56% had an administrative measure
as an intervention; 63% had education and training of healthcare
personnel; and 25% had judicious use of antimicrobial agents.
Although the evidence of the relationship between infection
prevention and multidrug-resistant infection rates was weak; the
overall evidence supported use of multiple interventions to
reduce the rates of multidrug-resistant organisms
46 studies were evaluated, of which 18 assessed isolation, 9
assessed cohorting nurses, and 19 assessed other isolation
policies. Few were planned prospective studies and all but one
included multiple interventions. Investigators for most studies did
not consider potential confounders, implement measures to
prevent bias, or use appropriate statistical analysis.
The studies were limited by major methodological weaknesses
and inadequate reporting. Thus, Cooper et al, could not exclude
plausible alternative explanations for the decreased incidence of
MRSA acquisition. The investigators conducting the
metaanalysis did not identify any well-designed studies that
allowed them to assess the role of isolation measures alone.
Active Surveillance
Reference
Robicsek A, 2008(36)
Study Population/Setting
In a 3-hospital, 850-bed organization with
approximately 40,000 admissions each year,
a 3-phase quasi-experimental study
compared:
Phase 1: Baseline,
Phase 2: Universal surveillance for MRSA
among all patients admitted to the ICU and
contact isolation for patients who carried
MRSA,
Phase 3: Universal surveillance for MRSA
among all patients admitted to the hospital
and contact isolation plus decolonization for
patients who carried MRSA.
The investigators implemented an MRSA
bundle, including active surveillance and
contact isolation for MRSA in 153 acute care
Veterans Affairs hospitals nationwide to
decrease healthcare-associated MRSA.
21,754 surgical ICU patients admitted to a
Swiss teaching hospital were included in a
crossover study comparing rapid screening
on admission to detect MRSA colonization
plus standard infection control measures vs
standard infection control measures alone.
Active Surveillance
Jain R, 2011(24)
Active Surveillance
Harbarth S, 2008(37)
Active Surveillance
Huskins WC,
2011(13)
Active Surveillance
D-7
Summary/Main Contribution
The absolute change between baseline and ICU surveillance
was -1.5 infections per 10,000 patient-days (p = 0.15), and the
absolute change between baseline and universal surveillance of
all patients admitted to the hospital was -5.0 infections per
10,000 patient-days (p < 0.01)
The investigators concluded that universal surveillance for MRSA
of all patients on admission was associated with a reduction in
MRSA infections during admission and within 30 days after
discharge.
Active Surveillance
Reference
Price CS, 2003(71)
Isolation of high-risk
patients
Matsushima A,
2011(52)
Universal Glove
Bearman G, 2007
(49)
Universal Glove
Bearman G, 2010
(50)
Study Population/Setting
This quasi-experimental study compared
hospital A, which did not routinely screen for
VRE colonization, to hospital B, which
actively screened high-risk patients for VRE
and placed VRE colonized or infected
patients under contact isolation. High-risk
patients were those admitted to the
hematology-oncology, transplant, or
intensive care units
This single ICU, quasi-experimental study to
assessed an intervention where all intubated
patients were placed under pre-emptive
contact precautions. In the first phase of the
study (415 patients), active surveillance for
MRSA was performed at ICU admission and
weekly with contact precautions for MRSA
positive patients. In the second phase of the
study (1280 patients), active surveillance
and contact precautions for MRSA remained,
however all intubated patients were also
placed on contact precautions.
6 month, single ICU study in which phase 1
(3 months) consisted of VRE and MRSA
surveillance cultures on admission and every
4 days with contact precautions for patients
colonized or infected with VRE or MRSA;
and phase 2 (3 months) consisted of
universal gloving only. In phase 1 there were
1090 patient-days and in phase 2 there were
1377 patient-days.
12 month single ICU study in which phase 1
(6 months) consisted of VRE and MRSA
surveillance cultures on admission and every
4 days with contact precautions for patients
colonized or infected with VRE or MRSA;
and phase 2 (6 months) consisted of
universal gloving and staff education. In
phase 1 there were 3,486 patient days and
in phase 2 there were 2,946 patient days.
D-8
Summary/Main Contribution
When the analysis was corrected for patient-days, the rate of
VRE bacteremia was 2.1-fold higher in hospital A compared to
hospital B. The majority of VRE isolates in hospital A were
clonally related.
Reference
Wright MO, 2004 (50)
Study Population/Setting
A single ICU, quasi-experimental study in
which phase 1 assessed active surveillance
and contact precautions for MRSA and VRE
while phase 2 included active surveillance
for MRSA and VRE but also implemented a
bundle to stop a multidrug-resistant
Acinetobacter baumannii outbreak. The
bundle included contact isolation for all
patients in the ICU regardless of culture
positivity, supervised terminal cleaning,
education sessions, and ban on artificial
fingernails.
D-9
Summary/Main Contribution
The intervention bundle controlled the outbreak of multidrugresistant Acinetobacter baumannii. In additions the bundle in
phase 2 led to a decrease in MRSA acquisition from 14% to 10%
(p=0.5), and VRE acquisition from 21% to 9% (p=0.05)
References
1.
2.
3.
Xu JQ, Kochanek KD, Murphy SL, TejadaVera B. Deaths: Final data for 2007.
National vital statistics reports. Hyattsville,
MD: National Center for Health Statistics;
2010.
4.
5.
6.
7.
8.
9.
D-10
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
29.
20.
30.
31.
21.
22.
32.
33.
34.
35.
36.
23.
24.
25.
26.
27.
28.
D-11
37.
38.
39.
40.
41.
42.
43.
44.
D-12
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
61.
62.
63.
55.
56.
64.
57.
65.
58.
66.
67.
68.
59.
60.
D-13
69.
70.
71.
D-14
Evidence Tables for Chapter 9. Reducing Unnecessary Urinary Catheter Use and
Other Strategies To Prevent Catheter-Associated Urinary Tract Infection: Brief
Update Review
Table 1, Chapter 9. Characteristics of studies with interventions to avoid unnecessary urinary catheter use
Study
(Country)
Study Design
Population,
Total N
Interventions to avoid
unnecessary catheter
PLACEMENT
None
Other
Interventions
Apisarnthanarak et
1
al, 2007
(Thailand)
Pre-Post
All Inpatients,
N=2412 patients
None
Bruminhent et al,
2
2010 (USA)
Pre-Post
Med-Surg: Ward +
ICU, N=400 patients
None
None
Nonrandomized
crossover trial
Medical (non-ICU),
N=70 patients
Computerized UC order
required selection of an
appropriate UC indication
UC care
education
Crouzet et al,
4
2007 (France)
Pre-Post
All Inpatients,
N=234 patients
None
Dumigan et al,
5
1998 (USA)
Pre-Post
ICU: Med-Surg,
N=27103 patientdays
Pre-Post
ICU: Medical,
N=337 patients
Pre-Post with
concurrent
controls
Med-Surg (non-ICU)
N=3736 intervention
patient-days, and
4041 control patientdays
D-15
None
UC care
education
None
None
Study
(Country)
Study Design
Population,
Total N
Pre-Post
Pre-Post
ED,
N=322 patients had
UCs placed, of 2517
ED patients in
sample
Statewide, N=163
inpatient units in 71
hospitals
ICU: Med-Surg,
N=not provided
10
Gokula et al,
11
2007 (USA)
Pre-Post
Other
Interventions
None
None
Education intervention to
promote adherence to
appropriate UC indications
Urinary retention protocol,
including use of bladder
scanner
None
None
None
Pre-Post
13
Pre-Post
ICU: Med-Surg,
N=6297 patients
None
Pre-Post
ICU: Med-Surg,
N=13471
catheter-days
Pre-Post
All Inpatients,
N=112,140
patient-days
14
ED,
N=200 patients with
UCs placed in ED
Medical (not ICU),
N=not provided
12
Pre-Post
Interventions to avoid
unnecessary catheter
PLACEMENT
Institutional guidelines for
appropriate UC placement, ED
physician education regarding
UC utilization
15
16
RCT
Medical (non-ICU),
N=692 patients
None
None
None
Bladder Bundle:
UC care steps,
selected use of
silver-alloy UC.
Bundle: UC care
education,
dedicated UC
nurse.
None
None
D-16
None
Study
(Country)
Study Design
Population,
Total N
Murphy et al,
17
2007 (USA)
Pre-Post
Not explained,
N=Not provided
Patrizzi et al,
18
2009 (USA)
Pre-Post
ED,
N=Not provided
Pre-Post
ICU: Med-Surg,
N=207 patients
Robinson et al,
20
2007 (USA)
Pre-Post
Med-Surg (non-ICU),
N=69 patients
Rothfield et al,
21
2010 (USA)
Pre-Post
19
22
Pre-Post with
concurrent
nonequivalent
controls
Schultz et al,
23
2011 (USA)
Pre-Post
Seguin et al,
24
2010 (France)
Pre-Post
ICU: Surgical,
N=1271 patients
Interventions to avoid
unnecessary catheter
PLACEMENT
None
Other
Interventions
UC care
education
Computerized ED UC order
with indications, UC alternatives
promoted, urinary retention
protocol with bladder scanner
use
Developed criteria for
appropriate UC placement in
ICU, implemented with
educational interventions
regarding UC indications, and
urinary retention protocol
None
None
UC care
education
None
None
D-17
None
None
None
Study
(Country)
Study Design
Population,
Total N
Stephan et al,
25
2006
(Switzerland)
Pre-Post with
concurrent
nonequivalent
controls
Surgery: Ward+ICU
Intervention:
Orthopedic, N=539
Control:
Abdominal, N=489
Pre-Post
Medical (non-ICU),
N = 245 patients
26
Voss, 2009
(USA)
28
Weitzel, 2008
(USA)
29
Wenger, 2010
(USA)
30
Pre-Post
All Inpatients, in 5
hospitals.
N=2943 patients
Interventions to avoid
unnecessary catheter
PLACEMENT
UC placement restrictions,
urinary retention protocol
Other
Interventions
UC care
education
UC care
education
Pre-Post
Medical (non-ICU),
N=187 patients age
65 or older
None
Pre-Post
Medical (unclear if
ICU), N=50 patients
None
Pre-Post
All Inpatients,
N=Not provided
None
D-18
Specially trained
UC nurse
None
None
UC care
education,
silver-alloy UC
References
1.
2.
3.
Apisarnthanarak A, Thongphubeth K,
Sirinvaravong S, et al. Effectiveness of
multifaceted hospitalwide quality
improvement programs featuring an
intervention to remove unnecessary urinary
catheters at a tertiary care center in
Thailand. Infect Control Hosp Epidemiol.
2007;28(7):791-8. PMID 17564980.
Bruminhent J, Keegan M, Lakhani A, et al.
Effectiveness of a simple intervention for
prevention of catheter-associated urinary
tract infections in a community teaching
hospital. Am J Infect Control.
2010;38(9):689-93. PMID 21034979.
Cornia PB, Amory JK, Fraser S, et al.
Computer-based order entry decreases
duration of indwelling urinary
catheterization in hospitalized patients. Am J
Med. 2003;114(5):404-7. PMID 12714131.
10.
11.
12.
13.
4.
5.
14.
6.
15.
7.
16.
17.
8.
9.
D-19
18.
19.
20.
21.
22.
24.
26.
27.
23.
25.
29.
30.
Evidence Tables for Chapter 10. Prevention of Central LineAssociated Bloodstream Infections: Brief Update Review
This brief review had no additional evidence tables.
Evidence Tables for Chapter 13. Preoperative Checklists and Anesthesia Checklists
Table 1, Chapter 13. Studies of the World Health Organization surgical safety checklist at other locations
Author/Year
Description of PSP
Study
Design
Description of Organization
Safety Context
Before and
after study,
comparing
pre-training
period to
post-training
Helmio 2011
Before and
after study
Finland, otorhinolaryngology-head
and neck surgery ORs. 747
operations in the two month study
periods combined. All subgroups
of otorhinolaryngology-head and
neck surgery were included.
Conley 2011
Case series
Sewell 2011
Bell 2010
4,5
D-21
Author/Year
Sparkes 2010
Royal Bolton
7
2010
Vats 2010
Kearns 2011
Description of PSP
Study
Design
Description of Organization
Safety Context
Case series
NR
Case series
Case series
D-22
Author/Year
Norton 2010
10
Description of PSP
Study
Design
Description of Organization
Safety Context
At this hospital they had been building a
quality infrastructure for five years prior, and
had already implemented the Universal
Protocol.
Table 2, Chapter 13. Implementation findings in studies of the World Health Organization surgical safety checklist at other locations
Author/Year
1
Sewell 2011
Training
Health Outcomes
D-23
Author/Year
Helmio 2011
Training
Training involved a
One-month implementation
presentation from an
period in Sept 2009
(443 operations).
outside expert and three
45 minute lectures. Specific
guidelines were in the OR,
and short instructions on the
back of the checklist.
D-24
Health Outcomes
Author/Year
Conley 2011
Bell 2010
4,5
Training
Health Outcomes
NR
Duration of rollout:
<2 months at three
hospitals, >6 months at two
hospitals.
NR
D-25
Author/Year
Sparkes 2010
Training
Extensive educational
support and training
Before checklist
NR
introduction: Although all
found the checklist to be
useful, many senior
clinicians felt that such
communication already took
place informally, and that
more paperwork would not
add to safety. Audit of 250
cases in Feb 2010 found
that team briefings occurred
in 77% and time outs in
86%.
Drop-in educational
sessions which involve
120 participants
The importance of
communicating with and
involving people beyond this
core group was recognised
straight away. Essentially
it is all about changing the
culture, which can be a long
process, but its well worth
it.
D-26
Health Outcomes
Author/Year
Vats 2010
Kearns 2011
Training
Health Outcomes
D-27
Author/Year
Norton 2010
10
Training
Health Outcomes
D-28
Table 3, Chapter 13. Studies of wrong-site-surgery checklists implementing the universal protocol
Author/Year
Description of PSP
Study Design
Garnerin et al.
11
2008
Verification protocol
for checking patient
identity and the site of
surgery
Case series
Theory or Logic
Model
the prevention
of wrong patients
and wrong site
surgery, not to
mention
accountability,
demanded an
intervention aimed
at improving the
way both patient
identity and site of
surgery checks
were performed,
while acquiring the
ability to identify
and correct
deficiencies
D-29
Description of
Organization
Swiss
anaesthesiology
service located
within a 1200 bed
university hospital
Safety Context
Implementation Details
Prior to introduction
of the checklist, all
patients were
required to wear ID
bracelets, and the
operative site had to
be signed by the
surgeon.
Anesthesiologists
were made aware
that they were being
monitored.
Author/Year
Description of PSP
Study Design
Nilsson et al.
12
2010
Questionnaire
after
implementation
Theory or Logic
Model
None explicitly
stated
D-30
Description of
Organization
Two Swedish
hospitals, bed
sizes not reported
Safety Context
Implementation Details
In the autumn of
2007, there were
two incidents of
wrong-side surgery
at these hospitals,
and a root-causes
analysis suggested
that a time-out
procedure might
help. The checklist
was pre-approved
by the heads of the
operating and
anesthesia
departments.
Author/Year
Description of PSP
Study Design
Owers et al.
13
2010
Case series
Anonymous
11
2007
Theory or Logic
Model
None explicitly
stated
Description of
Organization
English childrens
hospital, bed size
not reported
Hospital in North
Carolina, bed size
not reported
D-31
Safety Context
A preoperative
checklist already
existed at this
facility; they added a
correct site surgery
component
Implementation Details
Study
Design
Description of
Organization
Opinions, Knowledge
and Behavior
Health Outcomes
Thomassen et al.
14
2010
Case Study
Anaesthesia and
intensive care
department of a
1,100-bed tertiary
teaching hospital
Developed 26-item
checklist after review of
adverse events, PubMed
review of literature, and
expert panel discussions.
Modified Delphi
technique used.
Checklist used on
502 patients.
Emphasized avoiding
checklist fatigue.
Process was supervised
by participating senior
clinician; researchers
were also present.
85 checklists identified
one or more missing
items (17%).
Median checklist
completion time was
88.5 seconds; did not
substantially increase
pre-induction time.
Thomassen et al.
15
2010
Case Study
Anaesthesia and
intensive care
department of a
1,100-bed tertiary
teaching hospital
Checklist improved
confidence in unfamiliar
contexts. It revealed
insufficient equipment
standardization.
N/A
D-32
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
D-33
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
D-34
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
D-35
Discipline
involved in
intervention
Nursing
Ward Staff
Nursing
References
1
Ang et al, 2011 *
Study Design
RCT
Setting
8 medical wards in an acute
care hospital in Singapore
Barker et al,
2
2009
Before-and-after
study
Barry et al,
3
2001
Before-and-after
study
Yes/Local
Ward Staff
Multi
Before-and-after
study
No/No
Ward Staff
Nursing
Cluster RCT
Yes/No
Research
Staff/Ward
Staff
Nursing and
Physiotherapy
Brandis, 1999
Cumming et al,
5
2008
Participants
1822 newly admitted patients who
were age 21 or older, and scored 5
or above on fall risk model were
randomized.
271,095 patients admitted over 3
years before, and 6 years after
intervention
Individualized/
Use of Risk
Score
Yes/Local
D-36
Yes/Local
Individualized/
Use of Risk
Score
Yes/Local
Assessment/
Intervention
Performed
By
Research
Staff/ Ward
Staff
Discipline
involved in
intervention
Multi
References
Dykes et al,
6
2010 *
Study Design
Cluster RCT
Setting
8 medical units in 4 urban
United States hospitals
Participants
All patients admitted or transferred
to units over 6 month study period
Fonda et al,
7
2006
Before-and-after
study
Yes/Local
Ward Staff
Multi
GrenierSennelier et al,
8
2002
Before-and-after
study
Yes/Local
Ward Staff
Nursing
Haines et al,
9
2004
RCT
Yes/No
Ward Staff/
Research
Staff
Physiotherpay
and
Occupational
Therapy
Healey et al,
10
2004
Cluster RCT
Yes/No
Ward Staff
Multi
Cluster RCT
Yes/Local
Ward Staff
Nursing
Krauss et al,
12
2008
Before-and-after
study with
contemporaneous
cohort
Yes/No
Ward Staff
Nursing
11
D-37
References
Mitchell et al,
13
1996 *
Study Design
Before-and-after
Oliver et al,
14
2002
Before-and-after
study measure
Schwendimann
15
et al, 2006
Individualized/
Use of Risk
Score
Yes/Local
Assessment/
Intervention
Performed
By
Ward Staff
Discipline
involved in
intervention
Nursing
Setting
The intervention took place in a
32 bed medical ward serving
both acute and subacute
patients with high acuity needs,
which was compared to fall
rates in the entire 225 bed
acute care teaching hospital in
Australia prior to the pilot.
An elderly medical unit within an
acute hospital in England
Participants
All admissions in the hospital for 6
months prior to the pilot compared
to 6 months in the pilot ward after
implementation.
Yes/STRATIFY
Ward Staff
Multi
Before-and-and
after study
Yes/Local
Ward Staff
Multi
Stenvall et al,
16
2007
RCT
Orthogeriatric ward
(intervention) and orthopedic
ward and geriatric ward
(control) in a Swedish acute
hospital
Yes/No
Research and
ward staff
Multi
Uden et al,
17
1999
Before-andafterstudy
A geriatric department in an
acute hospital in Sweden
Yes/No
Ward Staff
Nursing
There were apparently no ward capacity issues as there is no mention of any patients not being admitted to the ward to which they were randomized.
D-38
References
Van der Helm et
318
al, 2006
Study Design
Before-and-after
study
Vassallo et al,
19
2004
Cohort Study
Before-and-after
study
Williams et al,
21
2007 *
Before-and-after
study
Individualized/
Use of Risk
Score
No/Local
Assessment/
Intervention
Performed
By
Ward Staff
Discipline
involved in
intervention
Nursing
Setting
One internal medicine ward and
one neurology ward within an
acute hospital in the
Netherlands
Participants
All admitted patients (2670) during
a 6-month before and 18-month
after period
Yes/Downton
Ward Staff
Multi
Yes/STRATIFY
Ward Staff
Multi
Note that the investigators describe the before period as a pilot study, but actually appear to be describing the falls rate and practice before the intervention,
that is, a baseline rather than the piloting of the intervention.
4
Although the investigators refer to the study as quasi-randomized and Oliver et al (2007)1 refer to it as a cluster RCT, it appears the intervention ward was
selected (not randomized) on the basis of being the ward where the researchers worked, and the quasi-randomization relates only to the fact that patients would
be allocated from a waiting list to whichever ward was the first to have an empty bed. The study also refers to matching patients, but this appears to be comparison
of the cohorts for differences rather than matching at individual patient level.
5
A separate publication (von Renteln-Kruse and Krause, 2004) describes a review of reported falls from January 2000 to December 2002 when 5946 patients
were admitted of whom 1015 were fallers and who had 1596 falls. This suggests that the proportion of fallers had been reducing substantially year-on-year
even before the intervention was introduced (ie, 17% [1015/5946] of patients fell before intervention in 20002002, 14% [611/4272] of patients fell before intervention
in 20032004, and 11% [330/2982] of patients fell after intervention in 2005-early 2006).
D-39
Table 2, Chapter 19. Stenvall et al. 2007, Main content of the postoperative program and differences between the two groups:
20
teamwork
Teamwork
Intervention group
Team included registered nurses (RN), licensed
practical nurses (LPN), physiotherapists (PT),
occupational therapists (OT), dietician, and
geriatricians.
Close cooperation between orthopedic surgeons and
geriatricians in the medical care of the patients
Control group
No corresponding teamwork at the
orthopedic unit.
D-40
Description of Fall
program
A new tool, the
ADAPT Fall
Assessment Tool,
was developed,
piloted, and
implemented as a
redesign for the
existing fall
prevention
program. The tool
automatically
calculates a fall risk
score from nurse
shift assessments
and produces a
score and
categorical
recommendation.
The 4 categories
were disorientation,
activity,
postmedication,
and toileting
precautions, and
each had a
corresponding
protocol and
suggested
interventions of
care.
Study Design
Theory or Logic
Model?
Descriptive with Redesign process
summative
looked for current
evaluation
recommendations in
the literature for fall
N=6402 inpatient risk factors. This
and observation included 4 authors.
records reviewed The tool of one of
from all adult
these authors,
medical-surgical Hendrich, was used
to validate the ADAPT
units, all
tool.
intensive care,
rehabilitation,
skilled nursing,
and psychiatric
units.
Description of
Organization
The Methodist
Healthcare System
(MHS) of San Antonio
used the Meditech
Clinical Documentation
Module for electronic
health records. This
system includes 7
inpatient facilities that
deliver full pediatric,
adult, rehabilitation,
maternal-child, and
psychiatric services.
D-41
Comments
Fall assessment
documentation
compliance on
admission and
daily increased to
100% for all units
in all hospitals.
Fall rates
decreased from
3.41 to 3.21 per
1000 adjusted
patient days.
Injuries per 100
falls decreased
from 1.44 to 0.95.
Author/
Year
Time series
design
No sample size
given
Theory or Logic
Model?
a literature search
looking at best
practices and
reviewing existing fall
risk assessment
tools
Description of
Organization
Additionally, care
coordination rounds
had an
interdisciplinary
team meet to
D-42
Comments
Author/
Year
Theory or Logic
Model?
Description of
Organization
Dempsey,
24
2004
Pre-post study of
implementation in
1995-1996,
follow up
assessment in
2001.
A regional teaching
hospital
In Australia.
Gutierrez,
25
2008
A specific specialty
adult focused
environment
(SAFE) unit as a
part of the definitive
observation unit.
The SAFE unit had
3 rooms with 2
beds each, staffed
by 2 RNs and 1
technical partner.
Fall protocol order
sets, post fall order
Time series
design
Total number of
patients not
responding
A literature review
was performed to
identify potentially
promising
interventions. Values
of physicians, and
nursing staff were
solicited to assess the
potential intervention
components.
Scripps
Mercy Hospital in San
Diego California
No other information
provided
D-43
1.
A possible reason
for the increase in
falls was
increased
reporting and not
an increase in
falls.
Comments
After an initial
reduction in falls, in
1995-1996,
beginning in 1998
falls reporting
began to increase
until they exceeded
pre 1995 levels.
The researcher
concluded that
falling compliance
associated with
increased
occupancy was
partly causative for
the decline in
effectiveness of the
program.
This project found
a lower rate of falls
(p-37/1000 patient
days) 3-6 months
after the
intervention
compared to the 9
months prior (3.0,
4.18, and 4.87
falls/1000 patient
days).
Author/
Year
Theory or Logic
Model?
Description of
Organization
D-44
Comments
Author/
Year
Kolin et al.,
26
2010
Theory or Logic
Model?
Description of
Organization
D-45
Comments
Author/
Year
McCollam,
27
1995
Theory or Logic
Model?
Description of
Organization
Descriptive
quantitative
a careful review of
Veterans Affairs
research-based falls Medical Center,
literaturefound only Portland, Oregon
Data provided on one falls assessment
fall rates,
instrument that met
compliance and [our criteria]. The
tool reliability
identified scale is the
MFS.
Patients scoring 45
or above received
nursing
implementations.
D-46
Additional
themes
Comments
Author/
Year
Neily, 2005
OConnell,
29
2001
28
Theory or Logic
Model?
Description of
Organization
Collaborative
breakthrough series
(BTS); the
intervention
includes signs to
identify high risk
patients, toileting
interventions, use
of hip pads,
environmental
rounds, staff
education, and
post-fall
assessment.
Pre-post study
with summary
evaluation
exploring the
influence of
context on
effectiveness.
Number of
patients not
reported.
The intervention
implementation was
based on the
collaborative
breakthrough series.
32 Veterans Affairs
facilities (a mix of
acute and long term
care facilities). State
veterans homes and
one private long term
care facility.
No other patient
data provided
D-47
Additional
themes
Comments
The primary
effectiveness
assessment of the
intervention was a
decrease in major
injury rate of 62%.
Methodological
barriers: Initial
attempt to design
evaluation as
RCT, then
controlled beforeand-after, left
evaluation team
discouraged that
pre-post study
was the only
feasible design.
In this study, no
statistically
significant benefit
of the program was
observed.
Author/
Year
Theory or Logic
Model?
Description of
Organization
University Medical
Center at Princeton
Plan-Do-Study-Act
(PDSA) performance
improvement model
was used throughout
the implementation
process.
Visual identifiers
were used in the
general
intervention, with a
daily list of at-risk
patients, arm
bands, and door
signs.
A postfall protocol
was developed and
introduced.
D-48
Comments
Author/
Year
Theory or Logic
Model?
Description of
Organization
Additional
themes
Comments
Nurses found
filling out falls
incident report
forms
troublesome.
In this study, no
statistically
significant benefit
of the program was
observed.
SeminGoossens,
31
2003
A guideline
developed by an
internal project
team of 11, with 4
nurses from each
ward, that focused
on identifying
patients, at
increased risk on
the basis of 3 main
risk factors and
then for patients at
increased risk
doing one or more
of: moving bed to
lowest position;
raising side rails;
noting the
increased risk in
Longitudinal time
series study
sample N=2670
patients. No
other patient data
provided.
D-49
Author/
Year
Weinberg et
32
al., 2011
Time series
design.
All beds were
included in
analysis, 714
beds in hospital.
Theory or Logic
Model?
and feedback on fall
rates. An
organizational plan to
use stickers to identify
high risk patients was
abandoned because
nurses judged them
stigmatizing with little
evidence of
effectiveness.
Description of
Organization
A previous fall
prevention program
failed. The belief was
that the
implementation was
simplistic.
Staten Island
University Hospital has
two campuses and 714
beds. Services include
medical/surgical.
pediatric, maternity,
behavioral sciences
and physical
rehabilitation.
D-50
Additional
themes
Comments
Author/
Year
Theory or Logic
Model?
documentation and
the role of a just
culture, were also
utilized.
Description of
Organization
D-51
Additional
themes
Comments
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
D-52
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
D-53
26.
27.
28.
29.
30.
31.
32.
Study Design
Patient
Population
Slor et al.
3
2011
The
Netherlands
Secondary
analysis of
RCT
526 patients
Adults aged
70 years or older
undergoing
acute or elective
hip surgery,
without delirium
at admission
(or profound
dementia
precluding
communication)
Burkhart et al.
4
2010
Switzerland
Cohort study
(post-hoc
analysis of
RCT)
113 patients
Adults aged
65 years or older
undergoing
cardiac surgery
with cardiopulmonary
bypass (CPB);
patients with
Mini-Mental
State Exam
(MMSE) score
<15/30 were
excluded
Description
of
Organization
Academic
hospital
(915 beds)
Academic
hospital
Diagnosis of
Delirium
Diagnostic
and Statistical
Manual of
Mental
Disorders,
Fourth Edition
(DSM-IV)
criteria and
Confusion
Assessment
Measure
(CAM)
CAM
D-54
Type of Analysis
and factors adjusted
for
Univariate analyses
followed by
multivariable logistic
regression; factors
controlled for include
age, APACHE II
score, MMSE score,
Snellen test score,
benzodiazepines,
anticholinergics,
opioids, type of
anesthesia
Univariate and
multivariable logistic
regression with
stepwise backward
elimination; factors
adjusted for include
C-reactive protein
(CRP), intraoperative
fentanyl, and duration
of mechanical
ventilation
Risk Factors
Modifiable risk
factors
None
Multivariable logistic
regression analyses:
Maximum value of
C-reactive protein
measured post-op:
OR: 1.1
(95% CI: 1.01-1.16)
P = 0.02
Fentanyl intraoperatively:
OR: 4.9
(95% CI: 1.72-13.8)
P = 0.003
Duration of mechanical
ventilation:
OR: 1.1
(95% CI: 1.04-1.21)
P = 0.004
Fentanyl amount,
duration of
mechanical
ventilation
Overall
risk of
bias
High
High
Author/Year/
Country
Study Design
Patient
Population
Hudetz et al.
5
2010
USA
Prospective
cohort study
40 patients
Kazmierski et
6
al. 2010
Poland
Prospective
cohort study
563 patients
Adult males
aged 55 years or
older scheduled
for elective
CABG and/or
valve
replacement/
repair
procedures with
CPB. Patients
with prior
documented
cognitive deficits
or vascular
dementia were
excluded.
Adult patients
admitted for
cardiac surgery
with
cardiopulmonary
bypass;
patients with
preop dementia
were excluded.
Description
of
Organization
Veterans
Affairs (VA)
medical center
Diagnosis of
Delirium
Academic
hospital
DSM-IV
criteria
Intensive
Care Delirium
Screening
Checklist
(ICDSC)
D-55
Type of Analysis
and factors adjusted
for
Univariate and
multiple logistic
regression; factors
adjusted for include
psychosocial
variables
(dispositional
optimism, perceived
social support,
perceived stress level,
and depression)
Risk Factors
Modifiable risk
factors
Incidence of post-op
delirium within 5 days
of surgery was reduced
by:
dispositional optimism:
OR: 0.57
(95% CI: 0.35-0.92)
p<0.02
None
Univariate analyses
followed by
multivariate backward
stepwise logistic
regression; factors
adjusted for include
age, MMSE score,
major depression,
anemia,
atrial fibrillation (AF),
intubation time, and
pO2 level.
Cognitive
impairment,
depression,
anemia, and AF
could be treated
prior to surgery
Overall
risk of
bias
Moderate
Moderate
Author/Year/
Country
Study Design
Patient
Population
Koebrugge et
7
al. 2010
The
Netherlands
Retrospective
cohort study
107 patients
Patients aged
65 years or older
undergoing
aortoiliac
surgery; patients
with Alzheimers
disease or
dementia were
excluded.
Retrospective
cohort study
26,057,988
hospitalizations
Hospitalizations
recorded in the
National
Inpatient Sample
(NIS) for
DRG categories
pneumonia,
orthopedic
surgery of the
lower extremity,
congestive heart
failure, and
urinary tract/
kidney infections
Description
of
Organization
Suburban
teaching
hospital
Diagnosis of
Delirium
NIS database
from
1998-2005
ICD-9 codes
for delirium
with
dementia,
drug-induced
delirium, and
nondementia,
non-drug
(NDND)
delirium
DSM-IV
criteria
D-56
Type of Analysis
and factors adjusted
for
Univariate and
multivariate step
forward logistic
regression; factors
adjusted for include
age and urgency of
surgery (emergency
vs. elective)
Multivariate stepwise
forward logistic
regression; factors
adjusted for include
age, gender,
logarithm base e,
length of stay, payor,
DRG, cerebrovascular
disease, dementia,
adverse drug effect,
sodium imbalance,
volume depletion,
anemia, atrial
fibrillation, respiratory
intervention, and
diabetes mellitus
Risk Factors
Modifiable risk
factors
Post-op delirium:
Age 70 years:
OR: 7.7
(95% CI: 1.9-30.4)
P<0.01
Emergency (vs. elective)
surgery:
OR: 5.3
(95% CI: 1.3-21.2)
P<0.01
Dementia-associated
delirium:
Age, logarithm base e,
length of stay,
cerebrovascular disease,
dementia, adverse drug
effect, sodium
imbalance, volume
depletion, atrial
fibrillation were all
significant risk factors for
delirium.
Female gender,
Medicaid as payor,
congestive heart failure
DRG, pneumonia DRG,
anemia, and diabetes
were associated with
significantly lower risk of
delirium.
Drug-induced delirium:
Age, logarithm base e,
length of stay,
cerebrovascular disease,
orthopedic DRG,
dementia, adverse drug
effect, were all significant
risk factors for delirium.
Female gender,
Medicaid as payor,
congestive heart failure
None
Sodium
imbalance,
volume depletion,
atrial fibrillation,
and anemia
Overall
risk of
bias
High
High
Author/Year/
Country
Study Design
Retrospective
cohort study
1,968,527
hospitalizations
Patient
Population
Acute care
hospitalizations
(for pneumonia,
lower extremity
orthopedic
surgery,
congestive heart
failure [CHF],
and kidney/
urinary tract
infection [UTI]) of
patients aged
18 years or older
in New York
Description
of
Organization
De-identified
inpatient data
obtained from
the New York
State Dept of
Health
Statewide
Planning for
Research
Cooperative
System
(SPARCS)
database
Diagnosis of
Delirium
ICD-9 codes
used to
identify
delirium
cases;
original
diagnostic
criteria not
reported
D-57
Type of Analysis
and factors adjusted
for
Forward stepwise
logistic regression;
factors adjusted for
include comorbidities,
DRG categories,
adverse drug effects
(ADEs), dementia,
mechanical
ventilation/ ventilator
assistance, gender,
age (in decade),
year of discharge,
Caucasian ethnicity,
Medicaid reimburse-
Risk Factors
Modifiable risk
factors
Overall
risk of
bias
None
High
Author/Year/
Country
Study Design
Patient
Population
Description
of
Organization
Diagnosis of
Delirium
State
(1998-2007).
Type of Analysis
and factors adjusted
for
ment, and elective
admission status
Risk Factors
Medicaid:
OR: 0.74
(95% CI: 0.66-0.82)
CHF DRG:
OR: 0.76
(95% CI: 0.64-0.89)
Lower extremity
orthopedic surgery
DRGs:
OR: 7.36
(95% CI: 6.38-8.50)
Any ADE:
OR: 22.19
(95% CI: 20.72-23.76)
Dementia:
OR: 1.26
(95% CI: 1.12-1.41)
Respiratory intervention:
OR: 1.96
(95% CI: 1.62-2.36)
Cerebrovascular
disease:
OR: 1.18
(95% CI: 1.01-1.39)
Atrial fibrillation:
OR: 1.24
(95% CI: 1.15-1.34)
Diabetes mellitus:
OR: 1.14
(95% CI: 1.06-1.23)
Volume depletion:
OR: 1.41
(95% CI: 1.28-1.57)
Anemia:
OR: 1.15
(95% CI: 1.05-1.25)
Hyponatremia:
OR: 1.42
(95% CI: 1.25-1.60)
D-58
Modifiable risk
factors
Overall
risk of
bias
Author/Year/
Country
Study Design
Patient
Population
Radtke et al.
10
2010
Germany
Cohort study
910 patients
Patients received
elective general
anesthesia and
were observed in
recovery room
and hospital
ward on first
postoperative
day
Prospective
cohort study
44 patients
Patients aged
65 or older who
spoke and
understood
English; Patients
with prevalent
delirium or
moderate to
severe cognitive
dysfunction were
excluded.
Rigney 2010
USA
11
Description
of
Organization
Academic
hospital
Diagnosis of
Delirium
Academic
hospital
(365 beds)
CAM
Nursing
delirium
screening
scale
(Nu-DESC)
Type of Analysis
and factors adjusted
for
Univariate and
multivariate logistic
regression with
delirium as the
response. Regression
analyses were
supplemented with a
feature selection
process using
backward elimination.
Factors adjusted for
include age, gender,
duration of surgery,
site, intraop opioids,
anesthetic, preop
fasting (solids and
fluids)
Univariate and
bivariate analyses
followed by logistic
regression; factors
adjusted for include
total allostatic load
(AL) scores,
primary mediators
score, secondary
outcomes score, and
individual AL
parameters
D-59
Risk Factors
Modifiable risk
factors
Multiple logistic
regression analyses:
Longer preoperative fluid
fasting time (>6 hr) was
the only significant risk
factor for delirium in both
the recovery room
(OR: 2.69,
95% CI: 1.38-5.24) and
the ward (OR: 10.57,
95% CI: 1.42-78.62).
Older age (OR: 1.02,
95% CI: 1.01-1.03) and
surgical site
(intraabdominal or
intrathoracic vs.
other sites) (OR: 1.83,
95% CI: 1.09-3.07) were
significant risk factors in
the recovery room.
Intraoperative opioid
choice (fentanyl vs.
remifentanil) was a
significant risk factor in
the ward (OR: 2.27,
95% CI: 1.01-5.06).
Primary mediators score
was the only significant
factor predicting delirium.
Preoperative fluid
fasting time,
choice of
intraoperative
opioid
None
Overall
risk of
bias
Moderate
Moderate
Author/Year/
Country
Study Design
Patient
Population
Sieber et al.
12
2010
USA
Double-blind
randomized
controlled trial
(RCT)
114 patients
Patients aged
65 or older
undergoing hip
fracture repair
under spinal
anesthesia with
propofol
sedation;
patients with
mental
barriers that
would preclude
data collection
were excluded.
Bo et al.
13
2009
Italy
Prospective
cohort study
252 patients
Patients aged
70 years
admitted from
emergency dept
(ED) to an acute
geriatric ward
(AGW) or an
acute general
medical ward
(AGMW);
patients with
delirium during
ED stay or at
ward entry were
excluded.
Description
of
Organization
Academic
medical center
Diagnosis of
Delirium
Academic
hospital
CAM
CAM
Type of Analysis
and factors adjusted
for
Univariate and
multivariate
regression; factors
adjusted for include
deep sedation,
dementia, units of
packed erythrocytes
transfused, and
admission to the ICU
Univariate analyses,
then multivariate
forward stepwise
modeling of variables
associated with
incident delirium;
factors adjusted for
include APACHE II
score, SPMSQ score,
stressful events,
AGW hospitalization
(vs. AGMW
hospitalization)
D-60
Risk Factors
Modifiable risk
factors
Multivariate regression
significant risk factors:
Deep sedation:
OR: 2.69
(95% CI: 1.04-6.93)
p = 0.04
preoperative dementia:
OR: 3.97
(95% CI: 1.54-10.2)
p = 0.004),
units of packed
erythrocytes transfused:
OR: 1.62
(95% CI: 1.10-2.38)
p = 0.01), and
admission to the ICU:
OR: 3.69
(95% CI: 1.17-11.7)
p = 0.02).
Risk of incident
delirium:
APACHE II:
RR: 1.30
(95% CI: 1.11-1.51)
P = 0.001
SPMSQ:
RR: 2.06
(95% CI: 1.62-2.64)
P<0.001
Stressful events:
RR: 3.36
(95% CI: 2.86-5.44)
P = 0.001
AGW hospitalization:
RR: 0.04
(95% CI: 0.01-0.21)
P<0.001
Sedation
More patients
can be admitted
to AGW vs.
AGMW, some
stressful events
might be reduced
Overall
risk of
bias
Moderate
Moderate
Author/Year/
Country
Study Design
Patient
Population
Greene et al.
14
2009
USA
Prospective
cohort study
100 patients
Patients aged
50 years or older
admitted for
major elective
noncardiac
surgery with
at least a 2-day
postop stay
Hattori et al.
15
2009
Japan
Prospective
cohort study
160 patients
Patients aged
75 years
admitted for
abdominal
surgery,
vascular surgery,
or orthopedic
surgery (all
non-emergency);
patients with
severe dementia
were excluded.
Description
of
Organization
Academic
medical center
Diagnosis of
Delirium
4 hospitals
(1 academic),
bed size
ranged from
300 to 887
NEECHAM
Confusion
Scale
CAM
D-61
Type of Analysis
and factors adjusted
for
Bivariate analyses
then multivariate
analysis: factors
adjusted for include
Geriatric Depression
Score-Short Form,
Trails B time,
Digit Symbol Test,
and Symbol Search
Test
Univariate and
multivariate analyses;
factors adjusted for
include age, gender,
department,
anesthesia, MMSE,
and preop NEECHAM
score
Risk Factors
Modifiable risk
factors
Geriatric Depression
Score-Short Form:
OR per unit: 1.53
(95% CI: 1.22-2.05)
P = 0.0001);
Trails B time:
OR: 1.02
(95% CI: 1.01-1.04)
Depression
Overall
risk of
bias
Moderate
Risk of postop
delirium:
Age >80 years:
OR: 3.14
(95% CI: 1.35-7.26)
Male:
OR: 2.86
(95% CI: 1.09-7.47)
Preop MMSE <25:
OR: 3.96
(95% CI: 1.52-10.39)
Preop NEECHAM <27:
OR: 5.33
(95% CI: 1.84-15.31)
None
Moderate
Author/Year/
Country
Study Design
Patient
Population
Katznelson et
16
al. 2009
Canada
Prospective
cohort study
1,059 patients
Patients
undergoing
cardiac surgery
with CPB
Description
of
Organization
Academic
hospital
Diagnosis of
Delirium
CAM-ICU
D-62
Type of Analysis
and factors adjusted
for
Univariate analysis
then multivariate
logistic regression
with backward and
stepwise selection;
factors adjusted for
include older age,
gender,
preop depression,
preop renal
dysfunction,
hypertension,
peripheral vascular
disease, New York
Heart Association
(NYHA) class >2,
preop anemia,
diabetes,
preop history of
cerebrovascular
accident/TIA,
prolonged CPB,
intraop anemia and
hyperglycemia,
complex cardiac
surgery, perioperative
intraortic balloon
pump support, and
massive blood
transfusion
Risk Factors
Modifiable risk
factors
Risk of postop
delirium:
Red blood cell
transfusion (>5 units):
OR: 3.29
(95% CI: 2.09-5.19)
Perioperative intraaortic
balloon pump support:
OR: 3.84
(95% CI: 1.72-8.56)
Preop depression:
OR: 3.06
(95% CI: 1.36-6.90)
Preop creatinine
>150 mM:
OR: 2.96
(95% CI: 1.9-4.63)
Age 60 years:
OR: 2.47
(95% CI: 1.43-4.23)
Combined CABG and
valvular surgery:
OR: 1.86
(95% CI: 1.16-2.98)
Preop administration of
statins:
OR: 0.54
(95% CI: 0.35-0.84)
Preop
administration of
statins, preop
depression,
preop creatinine
Overall
risk of
bias
High
Author/Year/
Country
Study Design
Patient
Population
Maldonado et
17
al. 2009
USA
RCT
118 patients
were
randomized to
three different
sedatives
Patients aged
18-90 years
admitted to the
ICU following
elective cardiac
surgery. Patients
with prior
diagnosis of
dementia were
excluded.
Pisani et al.
18
2009
USA
Prospective
cohort study
304 patients
Patients aged
60 years
admitted to ICU
Description
of
Organization
Academic
medical center
and a VA
medical center
Diagnosis of
Delirium
Academic
hospital
(900 beds,
with
14-bed ICU)
CAM-ICU
DSM-IV
criteria
applied by a
neuropsychiatrist
Type of Analysis
and factors adjusted
for
Univariate followed by
multiple logistic
regression; factors
adjusted for include
age, gender,
ASA class, baseline
MMSE score,
Midazolam (vs.
Dexmedetomidine),
and Propofol (vs.
Dexmedetomidine)
Bivariate analyses,
then multivariate
forward selection
regression of
variables associated
with delirium
(P<0.20); factors
adjusted for include
benzodiazepine or
opioid use,
Haloperidol use,
steroid use,
ADL impairment,
history of depression,
dementia,
ICU diagnosis of
respiratory disease,
APACHE II score
(minus Glasgow
Coma Scale), Alanine
aminotransferase
level, intubated during
ICU stay, restraint use
during ICU stay
D-63
Risk Factors
Modifiable risk
factors
Post-op sedation:
Midazolam vs.
Dexmedetomidine:
OR: 28.6
(95% CI: 3.7-262.5)
p = 0.01
propofol vs.
dexmedetomidine:
OR: 29.6
(95% CI: 4.8-280.6)
p = 0.01
Age (increasing
10 years):
OR: 1.3
(95% CI: 1.1-1.5)
p = 0.01
Benzodiazepine or
opioid use:
Rate Ratio: 1.64
(95% CI: 1.27-2.10)
Dementia:
Rate Ratio: 1.19
(95% CI: 1.07-1.33)
Haloperidol:
Rate Ratio: 1.35
(95% CI: 1.21-1.50)
APACHE II score:
Rate Ratio:1.01
(95% CI: 1.00-1.02)
Other models showed:
Benzodiazepines or
opioids are a significant
risk for delirium when
dementia is absent, but
not when it is present.
Haloperidol is a
significant risk for
delirium when dementia
is absent, but not when it
is present.
Post-op sedation
Overall
risk of
bias
High
Medication use
Moderate
Author/Year/
Country
Study Design
Patient
Population
Rudolph et al.
19
2009
USA
Prospective
cohort study
122 patients
(derivation
set),
109 patients
(validation set)
Patients aged
60 years who
underwent
cardiac surgery
under general
anesthesia;
patients with
delirium prior to
surgery were
excluded.
Description
of
Organization
Two academic
medical
centers and a
VA hospital
Diagnosis of
Delirium
CAM
D-64
Type of Analysis
and factors adjusted
for
Multivariate modeling
with bootstrap
resampling was used
to develop a
prediction rule.
Risk Factors
Modifiable risk
factors
Depression,
cognitive
impairment,
abnormal
albumin
Overall
risk of
bias
Moderate
Author/Year/
Country
Study Design
Patient
Population
Smith et al.
20
2009
USA
Retrospective
cohort study
998 patients
Adults aged
18 years
undergoing
non-cardiac
surgery, with a
minimum of
2 days inpatient
stay. Patients
with history of
dementia or
MMSE score 23
were excluded.
Description
of
Organization
Academic
hospital
Diagnosis of
Delirium
Retrospective
chart review
and/or CAM
D-65
Type of Analysis
and factors adjusted
for
General linear
modeling and logistic
regression with all
covariates entered
simultaneously.
Factors adjusted for
include age, years of
education, Charlson
comorbidity scale,
alcohol consumption
(drinks per week),
pain, and depressive
symptoms
Risk Factors
Modifiable risk
factors
Depressive
symptoms are
modifiable with
treatment
Overall
risk of
bias
High
Author/Year/
Country
Study Design
Patient
Population
Van Rompaey
21
et al. 2009
Belgium
Prospective
cohort study
523 patients
Patients aged
18 years were
in the ICU for
at least 24 hours.
Description
of
Organization
One academic
hospital,
one private
hospital, and
two community
hospitals
Diagnosis of
Delirium
Neelon and
Champagne
Confusion
Scale
D-66
Type of Analysis
and factors adjusted
for
Univariate logistic
regression followed
by multivariate
forward conditional
regression analysis;
factors adjusted for
include daily alcohol
use, cognitive
impairment,
admission for internal
medicine,
psychoactive
medication,
endotracheal tube or
tracheostomy,
more than 3
perfusions, isolation,
no visible daylight,
and no visit
Risk Factors
Modifiable risk
factors
Alcohol intake,
psychoactive
medication dose,
isolation,
daylight, allowing
visitors
Overall
risk of
bias
Moderate
Author/Year/
Country
Study Design
Patient
Population
Description
of
Organization
Academic
hospital
Diagnosis of
Delirium
Vidan et al.
22
2009
Spain
Controlled
clinical trial
542 patients
Patients aged
70 years
admitted to the
geriatric acute
care unit and two
internal medicine
wards. Patients
had to be free of
delirium yet have
risk factors for
delirium at time
of admission.
Voyer et al.
23
2009
Canada
Crosssectional study
155 patients
Patients aged
65 years with a
prior diagnosis of
dementia
CAM
Bivariate analyses
then multivariate
regression; factors
adjusted for include
age, severity of
dementia, and
risk factor scores
Koster et al.
24
2008
The
Netherlands
Prospective
cohort study
112 patients
Patients aged
45 years who
underwent
elective cardiac
surgery (with or
without CPB).
Patients with
preop delirium
were excluded.
Hospital
DSM-IV
criteria
Univariate and
multivariate analysis;
factors adjusted for
include age, type of
operation, anxiety
score, disturbed
sodium/potassium,
diabetes mellitus,
use of CPB, and
EuroSCORE
CAM
D-67
Type of Analysis
and factors adjusted
for
Logistic regression
with adjustment for
confounders;
these included age
(per decade),
dementia, baseline
ADL independence,
in-hospital stay
(per day),
intervention group
Risk Factors
Modifiable risk
factors
Dementia:
OR: 2.14
(95% CI: 1.15-3.99)
P = 0.02
Baseline ADL
independence:
OR: 0.78
(95% CI: 0.69-0.89)
P = 0.001
In-hospital stay (per
day): OR: 1.02
(95% CI: 1.00-1.05)
P = 0.05
Intervention group:
OR: 0.43
(95% CI: 0.24-0.77)
P = 0.005
Severity of dementia:
OR: 1.04
(95% CI: 1.02-1.06)
risk factor scores:
OR: 1.67
(95% CI: 1.11-2.51)
Risk factor scores based
on number of
predisposing factors for
each patient.
EuroSCORE:
OR: 1.12
(95% CI: 1.05-1.19)
P = 0.001
Electrolytes disturbance:
OR: 3.29
(95% CI: 1.16-9.34)
P = 0.025
Intervention
Overall
risk of
bias
Moderate
Dehydration,
fever, number of
medications,
depression were
modifiable factors
associated with
higher risk scores
Moderate
Electrolytes
disturbance
High
Author/Year/
Country
Study Design
Patient
Population
Lin et al.
25
2008
Taiwan
Prospective
cohort study
151 patients
Mechanicallyventilated adult
patients admitted
to ICU; delirium
assessed for first
5 days; history of
dementia was an
exclusion
criterion
Oh et al.
26
2008
Korea
Retrospective
cohort study
224
Description
of
Organization
Academic
medical center
Diagnosis of
Delirium
Academic
medical center
MMSE and
CAM
CAM-ICU
D-68
Type of Analysis
and factors adjusted
for
Univariate analyses,
then multivariate
stepwise regression
using selected
variables (P<0.1);
factors adjusted for
include diabetes
mellitus, sepsis, and
hypoalbuminemia
Univariate analyses
followed by
multivariate
regression of
significant factors;
factors adjusted for
include prior
dementia/ delirium,
abnormal preop
serum glucose,
diabetes, local or
regional anesthesia,
duration of surgery,
recovery room stay,
VAS score (>6.8), and
analgesics usage
Risk Factors
Modifiable risk
factors
Sepsis:
OR: 3.65
(95% CI: 1.03-12.90)
Hypoalbuminemia:
OR: 5.94
(95% CI: 1.23-28.77)
Note: Medications were
not associated
with delirium in
univariate
analyses.
Multivariate model risk
factors:
Previous
dementia/delirium:
OR: 630.4
(95% CI: 289.2-852.4)
P<0.0001
Pre-existent diabetes:
OR: 1.47
(95% CI: 1.17-2.45)
P = 0.012
Local or regional
anesthesia:
OR: 2.21
(95% CI: 1.34-3.47)
P<0.001
VAS score (>6.8):
OR: 1.99
(95% CI: 1.45-4.16)
P<0.001
Analgesics usage:
OR: 1.38
(95% CI: 1.06-2.14)
P = 0.038
None
Preop serum
glucose, type of
anesthesia,
analgesics usage
Overall
risk of
bias
Moderate
High
Author/Year/
Country
Study Design
Patient
Population
Redelmeier et
27
al. 2008
Canada
Retrospective
cohort study
284,158
patients
Description
of
Organization
Database
representing
all Ontario
hospitals
Diagnosis of
Delirium
ICD codes
used to
identify cases
Type of Analysis
and factors adjusted
for
Multivariable logistic
regression; factors
adjusted for include
age, sex,
neuropsychiatric drug,
type of surgery,
duration of surgery
Risk Factors
Modifiable risk
factors
Neuropsychiatric
drug use
Musculoskeletal surgery:
OR: 1.19
(95% CI: 1.08-1.31)
Lower urologic and
gynecologic:
OR: 0.55
(95% CI: 0.48-0.62)
Breast and skin surgery:
OR: 0.46
(95% CI: 0.36-0.59)
External head and neck
surgery:
OR: 0.39
D-69
Overall
risk of
bias
High
Author/Year/
Country
Study Design
Patient
Population
Description
of
Organization
Diagnosis of
Delirium
Type of Analysis
and factors adjusted
for
Inouye et al.
28
2007
USA
Prospective
cohort study
491 patients
(development
cohort)
469 patients
(validation
cohort)
Patients
aged 70 years
admitted to
6 general
medicine units
at an academic
hospital
Academic
medical center
CAM
Bivariable analyses
then multivariate
model; factors
adjusted for include
dementia, vision
impairment, activities
of daily living
impairment, Charlson
score, and restraint
use during delirium
Ely et al.
29
2007
USA
Prospective
cohort study
53 patients
Patients aged
18 years
admitted to the
ICU for >24 hrs
Community
teaching
hospital
(541 beds)
CAM-ICU
(intensive
care unit)
Ordinal logistic
regression
(dependent variable
was delirium days);
factors adjusted for
include APOE4, age,
APACHE II score,
coma days,
sepsis/ARDS/
pneumonia, and
Lorazepam total dose
D-70
Risk Factors
Modifiable risk
factors
Overall
risk of
bias
Restraint use,
vision
impairment,
functional
impairment
Moderate
None
High
Author/Year/
Country
Study Design
Patient
Population
Leung et al.
30
2007
USA
Prospective
cohort study
203 patients
Patients aged
65 years
scheduled for
major noncardiac
surgery requiring
anesthesia
Ouimet et al.
31
2007
Canada
Prospective
cohort study
203 patients
Patients age
18 years
admitted for
more than 24 hr
to an ICU
Description
of
Organization
Academic
medical center
Diagnosis of
Delirium
Academic
hospital
Intensive care
delirium
screening
checklist
(ICDSC)
CAM
D-71
Type of Analysis
and factors adjusted
for
Univariate analysis
then multivariate
logistic regression
with the most
promising factors
(APOE, age, history
of CNS disorders,
education, pain,
ADLs, alcohol intake,
cognitive status,
GDS score)
Univariate then
multivariate stepwise
logistic regression on
selected variables;
factors adjusted for
included age,
hypertension,
tobacco consumption,
alcohol consumption,
APACHE II score,
epidural catheter use,
opiate dose,
benzodiazepine dose,
propofol dose,
indomethacin dose,
coma, anxiety, and
pain
Risk Factors
Modifiable risk
factors
None
Overall
risk of
bias
High
Moderate
Author/Year/
Country
Study Design
Patient
Population
Pisani et al.
32
2007
USA
Prospective
cohort study
304 patients
Patients
60 years old
admitted to ICU
for at least
24 hrs
Rudolph et al.
33
2007
USA
Prospective
cohort study
1,218 patients
Patients aged
60 years
undergoing
noncardiac
surgery. Patients
with dementia
were excluded.
Description
of
Organization
Academic
hospital
(900 beds,
with
14-bed ICU)
Diagnosis of
Delirium
13 hospitals in
8 countries
(Denmark,
France,
Germany, the
UK, Greece,
the
Netherlands,
Spain, and the
USA)
DSM-III
criteria
CAM-ICU
D-72
Type of Analysis
and factors adjusted
for
Univariate analysis
then multivariate
modeling; factors
adjusted for include
alcohol, Medicaid
status, race, history of
depression,
medication use,
dementia, APACHE II
score, admitting
diagnosis, admitting
laboratory variables,
and admitting
physiologic variables
Bivariate analyses,
then stepwise
backward and forward
proportional hazard
regression models
using the most
promising variables;
factors adjusted for
included age,
gender (male),
cognitive
performance,
tobacco exposure,
diabetes, prior
myocardial infarction
(MI), and vascular
surgery
Risk Factors
Modifiable risk
factors
Dementia by
IQCODE >3.3:
OR: 6.3
(95% CI: 2.9-13.8)
Benzodiazepines
before ICU admission:
OR: 3.4
(95% CI: 1.6-7.0)
Creatinine >2 mg/dL:
OR: 2.1
(95% CI: 1.1-4.0)
Arterial pH <7.35:
OR: 2.1
(95% CI: 1.1-3.9)
Vascular risk factors
(tobacco exposure and
vascular surgery):
Rate Ratio: 3.2
(95% CI: 2.1-4.9)
Mildly impaired cognitive
performance:
Rate Ratio: 2.2
(95% CI: 1.4-2.7)
Age (per year):
Rate Ratio: 1.1
(95% CI: 1.0-1.1)
Benzodiazepine
use,
creatinine level,
and arterial pH
are modifiable
Cognitive deficit
might be
treatable prior to
surgery
Overall
risk of
bias
Moderate
Moderate
Author/Year/
Country
Study Design
Patient
Population
Description
of
Organization
Academic
hospital
Diagnosis of
Delirium
VelizReissmuller et
34
al. 2007
Sweden
Prospective
cohort study
107 patients
Patients aged
60 years
scheduled for
CABG,
valve surgery or
combined
procedures;
none had
dementia
Beaussier et
35
al. 2006
France
Double-blind
RCT
59 patients
Patients aged
>70 years
undergoing
surgical
resection of
cancer of the left
colon or rectum;
patients with
preoperative
mental
dysfunction were
excluded.
Academic
hospital
CAM
Furlaneto and
Garcez-Leme
36
2006
Brazil
Prospective
cohort study
103 patients
Patients aged
65 years
admitted to the
geriatric
orthopedic ward
for hip fracture
(almost all
underwent
surgery)
Academic
medical center
CAM
CAM
D-73
Type of Analysis
and factors adjusted
for
Univariate analysis
then logistic
regression of
significant variables;
factors adjusted for
include age,
alcohol consumption,
memory complaints,
CABG-valve vs.
CABG, valve vs.
CABG, MMSE score
Comparison of
randomized group
outcomes,
no adjustment for
other factors. General
anesthesia for colon
resection; pre-op
intrathecal morphine
(0.3 mg) + postop
patient-controlled
(PCA) intravenous
morphine vs.
PCA alone
Univariate regression
prior to logistic
regression modeling;
factors adjusted for
include mental
assessment factors
(MMSE,
clock drawing,
blessed), ADL and
length of hospital stay
Risk Factors
Modifiable risk
factors
Overall
risk of
bias
High
Memory complaints:
OR: 3.37
(95% CI: 1.0-11.5)
Valve vs. CABG:
OR: 3.90
(95% CI: 1.0-15.8)
MMSE score
(28 preop):
OR: 11.3
(95% CI: 2.7-47.7)
Cognitive deficit
may be treatable
prior to surgery
No significant difference
in delirium incidence was
found between the two
groups.
None, since
neither
intervention
showed a
difference
Moderate
Cognitive deficit:
OR: 3.04
(95% CI: 1.24-7.41)
Cognitive deficit
may be treatable
Moderate
Author/Year/
Country
Study Design
Patient
Population
Goldenberg et
37
al. 2006
USA
Prospective
cohort study
77 patients
Patients aged
>65 years
admitted for hip
surgery; patients
with existing
delirium were
excluded
Kazmierski et
38
al. 2006
Poland
Prospective
cohort study
260 patients
All patients
received cardiac
surgery; patients
with preop
delirium or
dementia were
excluded
Description
of
Organization
Community
teaching
hospital
Diagnosis of
Delirium
Academic
hospital
DSM-IV
criteria
CAM
D-74
Type of Analysis
and factors adjusted
for
Univariate logistic
analysis identified
12 factors as
predictors; these
were included in a
multivariate logistic
regression analysis
(age, morbidity index,
Hct, Alb, MMSE
score, set test score,
ADL score, dementia,
skilled nursing facility
(SNF) residence,
multiple medications,
CNS medications and
abnormal laboratory
values)
Univariate analyses,
then significant
variables added to
multivariate
regression model
(backward stepwise
procedure); factors
adjusted for include
MMSE score, AF,
peripheral vascular
disease, major
depression,
cerebrovascular
disease, and age
Risk Factors
Modifiable risk
factors
Multiple
medications and
cognitive
impairment, but
there may not be
time before
surgery to modify
these factors
Depression,
cognitive
impairment,
AF can be
treated prior to
surgery
Overall
risk of
bias
Moderate
High
Author/Year/
Country
Study Design
Patient
Population
Leung et al.
39
2006
USA
Blind RCT
228 patients
Patients aged
65 years
undergoing
non-cardiac
surgery requiring
general
anesthesia,
expected to
remain in the
hospital 48 hr
Pandharipande
40
et al. 2006
USA
Prospective
cohort study
198 patients
All adult
mechanicallyventilated
patients admitted
to ICU; patients
with preop
neurological
diseases that
would confound
delirium
diagnosis were
excluded.
Description
of
Organization
Academic
hospital
Diagnosis of
Delirium
Academic
medical center
CAM-ICU and
Richmond
Agitation
Sedation
Scale (RASS)
CAM
D-75
Type of Analysis
and factors adjusted
for
Bivariate analyses
then multivariate
logistic regression
analysis with
variables associated
with delirium
(P0.20); factors
adjusted for include
age, anesthetic type
(N2O vs. oxygen),
dependence on
performing 1 IADL,
Postop analgesia
(PCA vs.
oral opioids),
benzodiazepine use
on POD 1 or POD 2
Multivariable analysis
of sedative and
analgesic medications
as risk factors for
delirium in a Markov
model; factors
adjusted for include
age, gender, visual
and hearing deficits,
dementia, depression,
severity of illness,
sepsis, neurologic
disease, hematocrit,
daily serum glucose
level, lorazepam,
midazolam, fentanyl,
morphine, and
propofol
Risk Factors
Modifiable risk
factors
Age:
OR: 1.07
(95% CI: 1.02-1.26)
Dependence on
performing 1 IADL:
OR: 1.54
(95% CI: 1.01-2.35)
Postop analgesia
(PCA vs. oral opioids:
OR: 3.75
(95% CI: 1.27-11.01)
Benzodiazepine use on
POD 1 or POD 2:
OR: 2.29
(95% CI: 1.21-4.36)
Postop
analgesia,
benzodiazepine
use
Use of lorazepam
(alternative
medications can
be substituted)
Overall
risk of
bias
Moderate
Moderate
Author/Year/
Country
Study Design
Patient
Population
Ranhoff et al.
41
2006
Italy
Prospective
cohort study
401 patients
Patients
60 years of age
admitted to a
sub-intensive
care unit for
elderly patients
(SICU)
Sheng et al.
42
2006
Australia
Prospective
cohort study
156 patients
Stroke patients
aged 65 years
recruited over
1 year
Description
of
Organization
General
hospital
Diagnosis of
Delirium
Academic
teaching
hospital
(450 beds)
DSM-IV
criteria
CAM
Type of Analysis
and factors adjusted
for
Bivariate analysis
then multiple logistic
regression of
variables with p<0.05
in bivariate analysis;
factors adjusted for
include heavy alcohol
use, fitted bladder
catheter, number of
drugs, visual
problems, Acute
Physiology Score
(APS), Age,
S-albumin, dementia
Binary logistic
regression then
multiple logistic
regression analyses
using significant
variables;
factors adjusted for
include age,
dementia prestroke,
hemorrhagic stroke,
metabolic factor,
able to lift both arms,
Glasgow coma scale
score <15, neglect,
dysphasia,
vision field loss,
urinary tract infection,
urinary incontinence,
fecal incontinence,
systolic blood
pressure, diastolic
blood pressure, and
one or more
metabolic factors
D-76
Risk Factors
Modifiable risk
factors
Use of bladder
catheters and no.
of drugs
None
Overall
risk of
bias
Moderate
Moderate
Author/Year/
Country
Study Design
Patient
Population
Korevaar et al.
43
2005
The
Netherlands
Prospective
cohort study
126 patients
All patients
>65 years and
acutely admitted
Shulman et al.
44
2005
Canada
Retrospective
cohort study
10,230
patients
All patients
>65 years who
were newly
dispensed 1 of 3
drugs: lithium,
valproate, or
benztropine
Description
of
Organization
Academic
medical center
Diagnosis of
Delirium
4 administrative databases
covering all
hospitals in
Ontario
Not reported
CAM
D-77
Type of Analysis
and factors adjusted
for
Univariate and
multivariate logistic
regression analysis;
factors adjusted for
include cognitive
impairment,
Katz ADL, Urea, and
leucocytes
Cox proportional
hazards regression,
adjusted for lithium,
valproate,
benztropine, age,
sex, comorbidity,
visual impairment,
and hearing
impairment
Risk Factors
Modifiable risk
factors
Cognitive
impairment
Benztropine use
Overall
risk of
bias
Moderate
High
Author/Year/
Country
Study Design
Patient
Population
Yildizeli et al.
45
2005
Turkey
Retrospective
cohort study
432 patients
Patients aged
18 years
admitted for
major elective or
urgent thoracic
surgery
Description
of
Organization
Academic
hospital
Diagnosis of
Delirium
DSM-IV
criteria
D-78
Type of Analysis
and factors adjusted
for
Univariate analyses,
then multivariate
stepwise logistic
regression; factors
adjusted for include
age, gender,
chronic disease,
alcohol abuse,
psychiatric problems,
diabetes,
cerebrovascular
disease,
chemotherapy usage,
operation due to
malignancy,
urgent operation,
respiratory
insufficiency,
markedly abnormal
serum chemistry
values, operation
time, length of
hospital stays,
length of intensive
care unit stays,
sleep deprivation,
hypertension,
infection, blood
transfusion, use of
various drugs,
immobilization
Risk Factors
Modifiable risk
factors
Markedly abnormal
serum chemistry values:
OR: 3.01
p = 0.038
Sleep deprivation:
OR: 5.64
p = 0.05
Age:
OR: 1.04
p = 0.03
Operation time:
OR: 1.29
p = 0.04
Sleep
deprivation,
abnormal serum
chemistry
Overall
risk of
bias
High
Author/Year/
Country
Study Design
Patient
Population
Bucerius et al.
46
2004
Germany
Retrospective
cohort study
16,184
patients
All patients
receiving cardiac
surgery
Description
of
Organization
Academic
hospital
Diagnosis of
Delirium
Physician
diagnosis
based on
American
Psychiatric
Association
(APA)
guidelines
Type of Analysis
and factors adjusted
for
Univariate analyses,
then significant
variables added to
multivariate
regression model
(backward stepwise
procedure); factors
adjusted for include
age, beating-heart
surgery, atrial
fibrillation,
cerebrovascular
disease, diabetes,
peripheral vascular
disease, LVEF,
preop cardiogenic
shock,
urgent operation,
operating time,
intraop hemofiltration,
and RBC transfusion
Risk Factors
Modifiable risk
factors
Type of surgery
(if patient is
candidate for
beating-heart
surgery); AF can
be treated prior
to surgery
Intraop hemofiltration:
OR: 1.26
(95% CI: 1.06-1.49)
RBC transfusion
2000 ml:
OR: 3.15
(95% CI: 2.71-3.65)
Lower risk of delirium:
Beating-heart surgery:
OR: 0.47
(95% CI: 0.32-0.69)
YOUNGER AGE:
Age <50 years:
D-79
Overall
risk of
bias
High
Author/Year/
Country
Study Design
Patient
Population
Description
of
Organization
Diagnosis of
Delirium
Type of Analysis
and factors adjusted
for
Caeiro et al.
47
2004
Portugal
Prospective
cohort study
218 patients
Consecutive
acute stroke
patients admitted
to stroke unit
Academic
hospital with
12-bed stroke
unit
Delirium
Rating Scale
(DRS) score
10 and
fulfilled
DSM-IV-TR
criteria
Univariate and
multivariate analysis
with stepwise logistic
regression
Santos et al.
48
2004
Brazil
Prospective
cohort study
220 patients
Patients aged
60 years
admitted for
nonemergency
CABG; patients
with severe
cognitive deficits
were excluded.
Academic
tertiary referral
hospital
DSM-IV
criteria
3 multivariate
analyses:
(1) preop variables;
(2) preop and intraop
variables;
(3) preop, intraop, and
postop variables);
factors adjusted for
include age,
blood urea,
cardiothoracic index,
hypertension,
smoking, blood
replacement, AF,
pneumonia,
blood balance
nd
2 postop day
D-80
Risk Factors
OR: 0.22
(95% CI: 0.15-0.31)
Age 50 and <60 years:
OR: 0.34
(95% CI: 0.27-0.43)
Age 60 and <70 years:
OR: 0.6
(95% CI: 0.52-0.68)
Non-neuroleptics
anticholinergics (ACH)
during hospitalization,
medical complications,
ACH taken before stroke,
and intracerebral
hemorrhage (ICH) all
remained in the final
regression model.
Age:
OR: 1.1
(95% CI: 1.01-1.19)
Blood urea:
OR: 1.03
(95% CI: 1.01-1.05)
Cardiothoracic index:
OR: 3.38
(95% CI: 1.39-8.25)
Hypertension:
OR: 3.55
(95% CI: 1.25-10.14)
Smoking:
OR: 4.19
(95% CI: 1.35-13.05)
AF:
OR: 2.62
(95% CI: 1.05-6.58)
U:
OR: 6.36
(95% CI: 1.24-32.71)
Modifiable risk
factors
Overall
risk of
bias
Use of ACH
medications
High
Blood urea,
hypertension and
AF are potentially
modifiable
prior to
nonemergency
surgery
Moderate
Author/Year/
Country
Study Design
Patient
Population
Bohner et al.
49
2003
Germany
Prospective
cohort study
153 patients
Patients
undergoing
elective arterial
surgery with an
expected time of
90 minutes
Description
of
Organization
Academic
hospital
Diagnosis of
Delirium
DSM-IV
criteria plus
DRS score
12 points
Type of Analysis
and factors adjusted
for
Univariate then
stepwise multivariate
analysis, which
adjusted for age,
depression,
major amputation,
supraortic occlusive
disease, body length,
cognitive impairment
(MMSE), colloid
infusion, minimal
potassium level,
hypercholesterinemia
Risk Factors
Modifiable risk
factors
Intraop colloid
infusion,
intraop minimal
potassium;
cognitive
impairment can
be treated prior
to surgery
Intraop minimal
potassium <3.4 mmol/L:
OR: 3.18
P = 0.021
D-81
Overall
risk of
bias
High
Author/Year/
Country
Study Design
Patient
Population
Centorrino et
50
al. 2003
USA
Retrospective
cohort study
139 patients
Consecutive
adult
hospitalized
patients given
clozapine
Description
of
Organization
Academic
hospital
Diagnosis of
Delirium
Investigator
consensus
based on
signs and
symptoms in
medical chart,
and rated by
consensus on
a 3-point
severity scale
(mild,
moderate,
severe)
D-82
Type of Analysis
and factors adjusted
for
Bivariate analysis
followed by
multivariate logistic
regression of factors
with associations with
delirium (p0.10);
factors adjusted for
include anticholinergic
meds, clinical
responder, age,
hospitalized 20 days,
antipsychotic meds,
CNS agent,
anticonvulsants,
any mood stabilizer,
clozapine dose
>250 mg/day,
tricyclic antidepressan
ts, benzodiazepines,
serotonin reuptake
inhibitors, women,
lithium, any
antidepressant
Risk Factors
Modifiable risk
factors
Anticholinergic
exposure
Overall
risk of
bias
High
Author/Year/
Country
Study Design
Patient
Population
Morrison et al.
51
2003
USA
Prospective
cohort study
541 patients
Patients
admitted for
hip fracture
without evidence
of delirium
Zakriya et al.
52
2002
USA
Prospective
cohort study
168 patients
Patients
admitted for
hip fracture
service
(age 50-98);
patients with preexisting delirium
or dementia
were excluded.
Description
of
Organization
4 metropolitan
hospitals
Diagnosis of
Delirium
Academic
hospital
CAM
CAM
Type of Analysis
and factors adjusted
for
Univariate analyses
then multivariate
logistic regression;
factors adjusted for
included age, gender,
residence,
cognitive impairment,
FIM score,
RAND score,
abnormal BP,
abnormal heart
rhythm, chest pain,
heart failure,
medical complication,
morphine, meperidine
Univariate analyses
then multiple logistic
regression of
variables with P0.1
from univariate;
factors adjusted for
include normal white
blood cell count,
abnormal serum
sodium, ASA class,
history of congestive
heart failure,
history of AF,
history of peripheral
vascular disease
D-83
Risk Factors
Modifiable risk
factors
Morphine dose,
meperidine use;
cognitive
impairment can
be treated prior
to surgery
Abnormal serum
sodium and white
blood cell count
Overall
risk of
bias
Moderate
High
Author/Year/
Country
Study Design
Patient
Population
Agostini et al.
53
2001
USA
Prospective
cohort study
426 patients
Andersson et
54
al. 2001
Sweden
Prospective
cohort study
457 patients
Patients aged
70 years with
no baseline
delirium admitted
to general
medical service
(non-ICU);
profound
dementia
precluding verbal
communication
was an exclusion
criterion.
Patients aged
65 years
referred for
orthopedic
surgery (hip
fracture or
elective
coxarthros or
gonarthros
surgery)
Description
of
Organization
Academic
hospital
(900 beds)
Diagnosis of
Delirium
Hospital
Modified
Organic Brain
Syndrome
(OBS) Score;
also
considered
DSM-IV
criteria
CAM
D-84
Type of Analysis
and factors adjusted
for
Logistic regression
model adjusted for
baseline delirium risk,
gender, and age
Risk Factors
Modifiable risk
factors
Diphenhydramine:
OR: 2.3
(95% CI: 1.4-3.6)
Diphenhydramine
Multiple regression,
stepwise model;
factors adjusted for
included gender, age,
vision, hearing,
reason for hospital
admission, number of
other diseases,
postop complications,
bladder catheter,
preop medical
treatment, anesthesia
time and method,
loss of blood during
surgery, time from
admission to surgery,
surgery time,
time of admission,
marital status,
cohabitation,
type of housing
Risk of developing
delirium:
Four or more physical
diseases:
Exp (B): 15.94
(95% CI: 4.60-55.31)
Reason for admission:
Exp (B): 4.74
(95% CI: 1.76-12.80)
Impaired vision:
Exp (B): 4.52
(95% CI: 2.27-8.98)
Preop medical treatment:
Exp (B): 2.66
(95% CI: 1.26-5.62)
Anesthesia time:
Exp (B): 1.82
(95% CI: 1.31-2.53)
OBS-score on
admission:
Exp (B): 1.28
(95% CI: 1.06-1.54)
Age:
Exp (B): 1.10
(95% CI: 1.04-1.15)
Impaired vision,
anesthesia time,
possibly preop
medical
treatment;
cognitive
impairment can
be treated prior
to surgery
Overall
risk of
bias
Moderate
Moderate
Author/Year/
Country
Study Design
Patient
Population
Dubois et al.
55
2001
Canada
Prospective
cohort study
418 patients
Consecutive
patients aged
18 years
admitted for
>24 hrs to the
ICU
Description
of
Organization
Academic
hospital with
16-bed
medical and
surgical ICU
Diagnosis of
Delirium
Intensive care
delirium
screening
checklist
D-85
Type of Analysis
and factors adjusted
for
Univariate analyses
then multivariate
analysis using the 5
best factors
(morphine, use of
epidural, smoking
history, bilirubin level,
hypertension)
Univariate nonsignificant factors:
COPD, alcohol abuse,
sodium level, glucose
level, lorazepam,
rooms without
windows, rooms with
windows
Risk Factors
Modifiable risk
factors
Risk of developing
delirium:
Hypertension:
OR: 2.6
(95% CI: 1.14-5.72)
Bilirubin level (% days
abnormal):
OR: 1.2
(95% CI: 1.03-1.40)
Use of Epidural:
OR: 3.5
(95% CI: 1.20-10.39)
Morphine (mean daily
dose):
0.01-7.1 mg:
OR: 7.8 (1.76-34.4)
7.2-18.6 mg:
OR: 9.2 (2.17-39.0)
18.7-331.6 mg:
OR: 6.0 (1.41-25.4)
Hypertension,
bilirubin level,
use of epidural,
morphine dose
Overall
risk of
bias
High
Author/Year/
Country
Study Design
Patient
Population
McCusker et
56
al. 2001
Canada
Retrospective
cohort study
444 patients
(326 with
delirium,
118 without)
Patients
65 years
admitted from
ED to
medical services;
59.5% had
dementia.
Christe et al.
57
2000
Switzerland
Double-blind
RCT
65 patients
Consecutive
geriatric
inpatients
requiring upper
gastrointestinal
endoscopy
Description
of
Organization
Primary acute
care general
hospital
Diagnosis of
Delirium
Academic
geriatric
hospital
(304 beds)
MMSE
decrease of
3 points or
more
CAM
D-86
Type of Analysis
and factors adjusted
for
Multivariable analyses
of variance;
factors adjusted for
include age,
delirium index score,
comorbidity, length of
follow-up, dementia,
study group,
prevalent delirium,
visual or hearing
impairment,
number of room
changes,
hospital unit,
in isolation,
stimulation level,
not in the same room,
in a single room,
physical restraint,
medical restraint,
surroundings not welllit, surroundings
noisy/quiet, radio/TV
on, clock/watch
absent,
calendar absent,
no personal
possessions,
not wearing glasses,
not using hearing
aids, family absent
Univariate analyses
then multivariate
stepwise forward and
backward logistic
regressions; factors
adjusted for were not
stated
Risk Factors
Modifiable risk
factors
Room changes,
physical and
medical restraint,
glasses
None
Overall
risk of
bias
Moderate
High
Prospective
controlled
before-after
(CBA) study
199 patients
Theory or
Logic Model
Not reported
Contexts
Implementation
Details
Outcomes: Benefits
External:
None mentioned
Organizational
Characteristics:
6 community
hospitals (part of
Summa Health
System), over 2,000
licensed beds. Acute
Care for Elders
(ACE) unit had prior
experience using
delirium prevention
guidelines.
Teamwork:
Multidisciplinary
delirium workgroup
with physicians and
ACE nurses, director
of hospital quality.
Nurse quality
management and
leadership, clinical
informatics nurses,
geriatric pharmacy,
and geriatric
medicine fellows.
Leadership:
3 of the authors led
the pilot in the ACE
unit.
Culture:
Statement that
Summa Health
System maintains a
strong commitment
to patient safety and
quality
Implementation tools:
First obtained
stakeholder
agreement, then
multidisciplinary
workgroup devised
strategy and carried
out the pilot project.
It involved education
of ACE unit staff on
delirium screening,
prevention and
treatment protocols
that were then
implemented.
Delirium incidence
decreased from 8.8%
in pre-implementation
group to 7.2% in
implementation group
(not statistically
significant).
Mean length of stay
decreased from
7.6 days to 4 days
(difference 3.6 days,
95% CI: 0.66 to
6.49 days).
D-87
Outcomes:
Harms
Influence of
Contexts on
Outcomes
Overall
Risk of
Bias
No harms
Not reported High
reported for
intervention.
Deaths,
ICI transfers, and
30-day
readmissions
all decreased in
intervention
group.
Black et al.
59
2011
Northern
Ireland
Nursefacilitated
family
participation
Chen et al.
60
2011
Taiwan
Modified
Hospital Elder
Life Program
(HELP);
modified to
include
3 shared risk
Theory or
Logic Model
Contexts
Implementation
Details
Outcomes: Benefits
Outcomes:
Harms
Influence of
Contexts on
Outcomes
Overall
Risk of
Bias
Incident delirium:
Not reported
Intervention: 25/87
(29%)
Control: 64/83 (77%)
OR = 0.12
(95% CI: 0.06-0.24)
P<0.0001
Authors also state
there were no
significant differences
in mean scores
between groups.
Delirium at
discharge:
HELP:
0/102 (0%)
Control:
12/77 (15.6%)
OR = 0.03
D-88
Not reported
Theory or
Logic Model
Contexts
factors
(functional,
nutritional, and
cognitive
status)
targeted by
3 modified
HELP
protocols (early
mobilization,
nutritional
assistance,
and therapeutic
cognitive
activities
gastrointestinal ward
for elective
surgery, with
expected
length of stay
>6 days
functional
decline.
The authors
earlier work
suggests that 3
key elements
are the most
relevant for
surgical patients
and those were
used in this
study.
Hospital Elder
Life Program
(HELP)
CBA study
(historical
control)
Thousands of
patients (aged
70 years)
from 2002 to
2008
HELP provides
skilled interdisciplinary staff
and trained
volunteers to
conduct
intervention
protocols
targeted toward
6 delirium
risk factors:
orientation,
therapeutic
activities, early
mobilization,
vision and
hearing
protocols,
oral volume
repletion, and
sleep
enhancement; it
has been
gastrointestinal ward)
Leadership:
Researchers
designed program
and led the study
Teamwork:
Not reported
Culture:
Not reported
Implementation tools:
A full-time trained
HELP nurse, blinded
to the study
hypothesis and
not an outcomes
assessor,
implemented the
program.
External:
None mentioned
Organizational
Characteristics:
Community teaching
hospital (500 beds)
Leadership:
The project director
was primarily
responsible for
implementation
Teamwork:
Interdisciplinary staff
and trained
volunteers
Culture:
Not reported
Implementation tools:
Volunteers were
trained by staff; some
received additional
training by speech
therapists and
Rubin et al.
61
2011
USA
Implementation
Details
Outcomes: Benefits
Outcomes:
Harms
Influence of
Contexts on
Outcomes
Overall
Risk of
Bias
D-89
Delirium rate:
Pre-HELP (2001):
41%
HELP (2002): 26%
HELP (2005): 16%
HELP (2008): 18%
Nurse satisfaction:
Nurses and
nurses aides reported
benefit and
satisfaction with HELP
and agreed with a
questionnaire item that
their job was more
satisfying due to
HELP.
Not reported
Theory or
Logic Model
shown to be
effective for
delirium
prevention.
Inouye et al.
62
63
2003 , 1999
USA
Contexts
Implementation
Details
director solicited
input from each
nursing unit director;
the project director
worked with the
Chief Nursing
Director to identify
subsequent units to
target; as patient
volume increased,
paid HELP staff and
volunteers were
added; one Elder
Life Specialist
became the
lead volunteer
coordinator; weekly
meetings of staff
were held to
maintain quality and
document
modifications to the
original HELP
protocols
Delirium has
External:
With oversight by a
HELP for
Prospective
None mentioned
geriatric nurse
prevention of matched CBA been
associated with Organizational
specialist and
delirium in
study
Characteristics:
geriatrician,
elderly patients 852 patients several risk
at least
Urban teaching
factors;
the Elder Life
70 years old the HELP
hospital (900 beds) specialists
Teamwork:
admitted to
targets 6 of
implemented
Interdisciplinary team 6 interventions:
general
these risk
including a geriatric orientation,
medicine floor factors
nurse-specialist,
(later study
(cognitive
therapeutic
two Elder Life
included
impairment,
activities, mobility,
specialists, a certified sleep, hearing or
422 patients sleep
from the HELP deprivation,
therapeuticvision, and
arm of the
recreation specialist, volume repletion (for
immobility,
dehydration); they
study)
a physical therapy
visual
were assisted by
consultant, a
impairment,
hearing
geriatrician, and
trained volunteers;
Outcomes: Benefits
Outcomes:
Harms
Influence of
Contexts on
Outcomes
Overall
Risk of
Bias
physical therapists
D-90
In the earlier
Not reported
publication, incident
delirium was
significantly lower in
the intervention group
vs. the usual care
group (9.9% vs. 15%,
OR: 0.60
(95% CI: 0.390.92); P
= 0.02
Theory or
Logic Model
Contexts
Implementation
Details
Outcomes: Benefits
Outcomes:
Harms
Influence of
Contexts on
Outcomes
Overall
Risk of
Bias
Patients underwent
pre-hospital care,
nurse assessment
immediately
after admission,
orthopedic surgeon
assessment 30 min
before referral to Xray department,
then transfer to
orthopedic ward,
then surgery for hip
fracture with general
or spinal anesthesia.
Delirium was
assessed by
researchers within 4
hrs of admission
and 8 hr after the
end of anesthesia.
D-91
Post-op delirium:
Intervention:
28/131 (21.4%)
Control:
44/132 (33.3%)
OR = 0.54 (0.31-0.95)
P = 0.03
Any
complications:
Intervention:
66/131 (50.4%)
Control:
70/132 (53.0%)
P = 0.67
Needham et
65
al. 2010
USA
delirium
screening
using Organic
Brain
Syndrome
(OBS) scale,
avoid polypharmacy, and
perioperative/
anesthetic
period protocol
Structured
quality
improvement
(QI) model with
components
including:
understanding
the problem
within the
larger
healthcare
system,
creating a
multidisciplinary
improvement
team, enlisting
all
stakeholders to
identify barriers
to change and
appropriate
solutions, and
creating a
change in
practice
through
engagement,
education,
execution, and
CBA study
(historical
control)
57 patients
with acute
respiratory
failure
Theory or
Logic Model
Contexts
Implementation
Details
Outcomes: Benefits
The QI model
was based on a
4 Es model
(engage,
educate,
execute, and
evaluate).
Previous
studies have
shown that
early physical
medicine and
rehabilitation
(PM&R) in the
ICU provides
benefits for
critically ill
patients, and
the QI model
applied this
evidence to
patients in the
medical ICU
(MICU).
External:
None mentioned
Organizational
Characteristics:
Academic hospital
with 16-bed MICU
Leadership:
The lead author was
the project leader.
Researchers were in
charge
Teamwork:
A multidisciplinary QI
team with
representatives from
each relevant
clinician group in the
MICU and PM&R
Culture:
Not reported
Implementation tools:
Education and
training of nurses,
physical therapists,
occupational
therapists, and
respiratory therapists
to obtain specific
skills related to rehab
of mechanically
Standardized MICU
admission modified
to change default
activity from
bed rest to
as tolerated;
change in sedation
practice from
continuous
intravenous
infusions to
as needed bolus
doses; establishing
guidelines for PT
and OT
consultation;
developing safetyrelated guidelines
for PM&R-related
consultation;
including a full-time
PT and OT and a
part-time rehab
assistant; consulting
a physiatrist; and
increasing
consultations to
neurologists for
MICU patients with
severe or prolonged
Incident delirium:
QI period:
125/482 (28%) MICU
patient days
Pre-QI period: 107/312
(36%) MICU patient
days
P = 0.003
D-92
Outcomes:
Harms
Unexpected
events:
QI period:
4 cases of rectal
or feeding tube
removal, without
any significant
complications
Pre-QI period:
No unexpected
events
P>0.99
Influence of
Contexts on
Outcomes
Overall
Risk of
Bias
Vidan et al.
22
2009
Spain
Harari et al.
66
2007
U.K.
evaluation
Education
measures and
specific actions
in 7 risk areas
(orientation,
sensory
impairment,
sleep,
mobilization,
hydration,
nutrition, drug
use), with daily
monitoring of
adherence
Proactive care
of older people
undergoing
surgery
(POPS); multidisciplinary
preoperative
comprehensive
geriatric
assessment
(CGA) service
with postoperative
Controlled
clinical trial
542 patients
aged
70 years
admitted to a
geriatric acute
care unit and
two internal
medicine
wards
CBA study
(historical
control)
108 patients
aged
65 years
undergoing
elective
surgery
Theory or
Logic Model
Authors discuss
the HELP
program
as inspiration,
but the new
protocol was
designed to be
implemented in
daily practice
without extra
staff (unlike
HELP).
The authors
hypothesized
that
preoperative
CGA
incorporating
prediction of
adverse
outcomes
combined with
targeted
interventions,
would reduce
Contexts
Implementation
Details
ventilated patients.
External:
None mentioned
Organizational
Characteristics:
Academic hospital
Leadership:
A specialist geriatric
nurse coordinated
the intervention and
monitored
adherence.
Teamwork:
A multidisciplinary QI
team including
geriatricians,
residents and nurses
who worked in the
geriatric ward.
muscle weakness.
Intervention
implemented within
first 24 hours of
admission to
geriatric ward by
geriatricians,
residents, and
nurses. A specialist
geriatric nurse
coordinated the
intervention and
monitored
adherence.
Culture:
Not reported
Implementation tools:
Educational program
aimed at changing
the patient care
approach of geriatric
ward staff.
External:
None mentioned
Organizational
Characteristics:
Urban teaching
hospital
Leadership:
Not reported
Teamwork:
A multidisciplinary QI
team including a
consultant
geriatrician,
The multidisciplinary
QI team
implemented POPS.
Most patients
received pre-op
home visits from
occupational
therapist and
physiotherapy.
Social worker
provided inputs if
needed. Patients
were educated in
D-93
Outcomes: Benefits
Outcomes:
Harms
Influence of
Contexts on
Outcomes
Overall
Risk of
Bias
New delirium
episodes:
Intervention:
20/170 (11.7%)
Usual care:
69/372 (18.5%)
OR = 0.59
(95% CI: 0.34-1.00)
P = 0.05
Not reported
Post-op delirium:
POPS:
3/54 (5.6%)
Pre-POPS:
10/54 (18.5%)
OR = 0.26 (0.07-1.00)
P = 0.036
Only reported
complications
were related to
surgery,
not POPS
Lundstrom et Postoperative
67
al. 2007
multifactorial
Sweden
intervention to
reduce delirium
and improve
outcomes in
patients with
femoral neck
fractures
Theory or
Logic Model
Contexts
post-operative
complications
and hence
length of stay
(LOS) in
older people
undergoing
elective
surgery.
This strategy
did not target
delirium alone,
but any factor
that might
contribute to
complications or
longer LOS.
nurse specialist in
older people,
occupational
therapist,
physiotherapist and
social worker
Culture:
Not reported
Implementation tools:
Geriatrician and
nurse provided staff
education in post-op
early detection and
treatment of medical
complications, early
mobilization, pain
management, bowelbladder function,
nutrition and
discharge planning.
Implementation
Details
optimizing post-op
recovery. The
geriatrician and
nurse reviewed
patients in
surgical wards and
provided
staff education in
post-op early
detection and
treatment of medical
complications, early
mobilization, pain
management,
bowel-bladder
function, nutrition
and discharge
planning. Follow-up
therapy home visits
were provided to
those with functional
difficulties, and
outpatient clinical
review in those with
ongoing medical
problems.
External:
After education,
None mentioned
all team members
Organizational
(except dietician)
Characteristics:
assessed each
Academic hospital
patient, usually
with 24-bed geriatric within 24 hours after
unit (used only for
admission; team
intervention group)
planning of
Leadership:
individual rehab
Not reported
performed twice a
Teamwork:
week; assessment
A multidisciplinary
of patients with
team including RNs, delirium for
LPNs, registered
precipitating factors;
physiotherapists,
prevention and
D-94
Outcomes: Benefits
Outcomes:
Harms
Post-op delirium:
Not reported
Intervention:
56/102 (54.9%)
Control:
73/97 (75.3%)
OR = 0.40 (0.22-0.73)
p = 0.003
Days with post-op
delirium:
Intervention:
5.0 7.1 days
Control:
10.213.3 days,
p = 0.009
Influence of
Contexts on
Outcomes
Overall
Risk of
Bias
Lundstrom et Education
68
al. 2005
program and
Sweden
reorganization
of nursing and
medical care
Quasi-RCT
400 patients
aged
70 years
admitted to
two wards
(intervention
and usual
care)
Theory or
Logic Model
Not reported
Contexts
Implementation
Details
registered
occupational
therapists, a dietician
and geriatricians
Culture:
Not reported
Implementation tools:
All nursing and
medical staff
members attended a
4-day course in
caring, rehabilitation,
teamwork, and
medical knowledge.
treatment of
complications
(infection, anemia,
embolism);
assessment of
bowel/bladder
function; treatment
of sleep apnea;
prevention and
treatment of
decubitus ulcers;
prevention and
treatment of post-op
pain; ensure oxygen
saturation during
first post-op day;
measure blood
pressure for first 2
post-op days;
ensure adequate
nutrition;
mobilization within
first post-op day;
and secondary
prevention of falls
and fractures.
All nursing and
medical staff
members attended a
2-day course
focusing on
dementia and
delirium in geriatric
patients. Staff were
also trained in the
caregiver-patient
interaction. Nursing
care was
reorganized to
support
individualized care,
External:
None mentioned
Organizational
Characteristics:
Academic hospital
Leadership:
Two of the authors
were the leaders of
the program
Teamwork:
A multidisciplinary
team including all
staff in the
intervention ward
Culture:
D-95
Outcomes: Benefits
Outcomes:
Harms
Influence of
Contexts on
Outcomes
Overall
Risk of
Bias
Prevalent delirium
No harms related Not reported High
within 24 hrs of
to intervention
admission:
were reported
Intervention:
63/200 (31.5%)
Control:
62/200 (31%)
P = 0.91
Delirium on Day 7:
Intervention:
19/63 (30.2%)
Control:
37/62 (59.7%)
OR = 0.29 (0.14-0.61)
P = 0.001
Tabet et al.
Educational
69
70
2005 ; 2006 package for
U.K.
medical and
nursing staff to
reduce
incidence of
delirium in
hospitalized
elderly
patients; a
control ward
did not receive
the educational
package and
performed
usual practice.
CBA study
(concurrent
control)
250 patients
aged
70 years
admitted to
two acute
admission
wards
Theory or
Logic Model
Contexts
Implementation
Details
Not reported
Implementation tools:
All nursing and
medical staff
members attended a
2-day course
focusing on dementia
and delirium in
geriatric patients.
Staff were also
trained in the
caregiver-patient
interaction.
External:
None mentioned
Organizational
Characteristics:
Two acute admission
wards in an inner-city
teaching hospital
Teamwork:
Geriatric psychiatrist
educated staff,
who altered their
practice based on the
education.
Leadership:
The lead investigator
(a geriatric
psychiatrist)
supervised the
project.
Culture:
Not reported
Implementation tools:
Education program
on delirium for staff
with follow-up
sessions delivered by
geriatric psychiatrist
The educational
package, which
highlighted delirium
risk factors, was
delivered on site
and at various times
to ensure all staff
were involved.
There was an initial
1 hr formal
presentation, written
information on
guidelines for
delirium prevention
and management,
and regular followup meetings to
reinforce learning;
researchers did not
intervene in day-today management or
provide specific
advice pertaining to
specific patients.
D-96
Outcomes: Benefits
Point prevalence of
delirium:
Intervention ward:
12/122 (9.8%)
Usual care ward:
25/128 (19.5%)
OR: 0.45
(95% CI: 0.21-0.94,
P<0.05)
Outcomes:
Harms
Not reported
Influence of
Contexts on
Outcomes
Overall
Risk of
Bias
The
High
educational
package was
found to more
effectively
prevent
delirium in
men
(OR: 0.17,
95% CI: 0.050.65) than in
women (OR:
1.04, 95% CI:
0.38-2.81).
Delirium
education for
hospital staff
plus
recommendations by
geriatric
registrar for up
to 10 possible
targeted
intervention
strategies to
prevent
delirium after
hip fracture
CBA study
(historical
control)
99 patients
aged
>50 years with
hip fracture
admitted to a
general
orthopedic unit
Theory or
Logic Model
Contexts
Implementation
Details
This strategy
had been
successfully
used at a U.S.
hospital in a
previouslypublished study;
targeted
recommendations include
regulation of
bladder and
bowel function,
early detection/
treatment of
major
complications,
correction of
fluid and
electrolyte
imbalance,
discontinuation
of unnecessary
medications,
provision of
oxygen, severe
pain treatment,
agitated
delirium
treatment, use
of appropriate
environmental
stimuli,
adequate
nutritional
intake, and
early
mobilization and
rehabilitation.
External:
None mentioned
Organizational
Characteristics:
Urban teaching
hospital (460 beds)
Teamwork:
Multidisciplinary
committee with
medical, nursing, and
allied health
members of the
orthopedic, geriatric,
and anesthetic depts.
Leadership:
The lead investigator
supervised the
project.
Culture:
Not reported
Implementation tools:
The leader educated
frontline staff
(interns, ward
nurses, and allied
health staff) on
delirium every
10 weeks
The lead
investigator
educated staff,
supervised data
collection and
assessed patients;
the project team met
fortnightly to
supervise the
program; the
intervention was
implemented over a
3-month period; the
major barrier was a
high turnover of
nursing staff that
was partly overcome
by the nurse
manager of the
orthopedic unit
ensuring that all
nursing staff
attended the
tutorials and
received education
about the use of the
CAM.
D-97
Outcomes: Benefits
Incident delirium:
Intervention:
9/71 (12.7%)
Pre-intervention:
10/28 (37.5%)
OR = 0.26
(95% CI: 0.09-0.74)
P = 0.012
Outcomes:
Harms
None reported
Influence of
Overall
Contexts on
Risk of
Outcomes
Bias
Not reported High
Proactive
geriatrics
consultation
with target
recommendations based
on a structured
protocol for
patients after
hip fracture
(target
recommendations same
as in Wong et
al. 2005)
Long-term care
Lapane et al. Pharmacist-led
73
2011
Geriatric Risk
USA
Assessment
MedGuide
(GRAM)
reports and
automated
monitoring
plans focusing
on medication
monitoring
phase to
prevent
potential
Theory or
Logic Model
Contexts
Implementation
Details
Single-blind
RCT
126 patients
aged
65 years
admitted
emergently for
surgical repair
of hip fracture
Not clearly
stated, other
than that
geriatrics
consultation is
easily
implementable
and that a
targeted,
proactive
strategy with
intervention on
defined
outcomes has
shown
effectiveness,
although it is
not clear
whether it has
shown prior
effectiveness in
delirium
prevention.
External:
None mentioned
Organizational
Characteristics:
Academic tertiary
medical center
Teamwork:
Geriatrician and
orthopedics team
worked together
Leadership:
Not reported
Culture:
Not reported
Implementation tools:
Not reported
A geriatrician
evaluated patients
preoperatively or
within 24 hours
postop, performed
daily visits for
duration of
hospitalization and
made targeted
recommendations.
The orthopedics
team (surgeons and
nurses)
implemented the
recommendations
(adherence rate:
77%).
The usual care
group received
management by the
orthopedics team,
including internal
medicine or geriatric
consults on a
reactive rather than
proactive basis.
Post-op delirium:
None reported
Consult:
20/62 (32%)
Usual care:
32/64 (50%)
P = 0.04
However, when
adjusted for baseline
imbalances the effect
size was no longer
statistically significant:
OR: 0.6
(95% CI: 0.3-1.3)
No significant
between-group
difference in days of
delirium per episode
Quasi-RCT:
3,202 patients
(2003)
3,321 patients
(2004)
25 nursing
homes were
randomized to
receive
intervention or
control
GRAM was
designed to
assist
healthcare
professionals
with expertise in
geriatric
pharmacotherapy in
problem
identification
when evaluating
complex
medication
External:
None mentioned
Organizational
Characteristics:
25 nursing homes
(each with 50 or
more geriatric beds).
All nursing homes
had stable contracts
with Omnicare and
had few short stay
residents
Teamwork:
Pharmacists shared
After training in
2003, GRAM
database for falls
and delirium was
integrated in
January 2004 into
the pharmacies
commercial
pharmacy software
system for the
intervention homes.
Reports were
generated on
medications that
Potential delirium
indicator:
In home 2003/04:
Adjusted hazard ratio:
0.93 (0.80-1.09)
New admits 2004:
Adjusted hazard ratio:
0.42 (0.35-0.52)
D-98
Outcomes: Benefits
Outcomes:
Harms
No significant
difference
between groups
for potential
adverse-event
related hospitalization, falls, or
death
Influence of
Overall
Contexts on
Risk of
Outcomes
Bias
Consultation Moderate
showed a
trend toward
being more
effective
among
patients
without
prefracture
dementia or
ADL
impairment,
but the
differences
were not
statistically
significant
Theory or
Logic Model
Contexts
Implementation
Details
adverse drug
events (falls
and delirium) in
nursing homes
regimens of
older adults to
identify, resolve,
and prevent
medicationrelated
problems, aid in
evaluation of
medications as
a cause or
aggravating
factor
contributing to
an older adults
physical,
cognitive, or
functional
decline, and
facilitate
incorporation of
medication
monitoring
information into
the older adults
plan of care.
contribute to falls
and delirium, as well
as medication
monitoring care
plans and flow
records. Facility
nurses received
reports within 24 h
of admission for new
admissions;
consultant
pharmacists did onsite reviews of drug
regimens for each
resident once every
30 days.
D-99
Outcomes: Benefits
Outcomes:
Harms
Influence of
Contexts on
Outcomes
Overall
Risk of
Bias
Description
of PSP
Larsen et al.
75
2010
USA
Atypical
antipsychotic
Perioperative
olanzapine
(5 mg orally
before and
after surgery)
or placebo to
prevent
postop
delirium in
elderly
patients after
joint
replacement
surgery
Study
Design
and
patients
Theory or
Logic Model
Description of
Organization
Implementation
Details
Outcomes:
Benefits
Outcomes:
Harms
Influence of
Contexts on
Outcomes
Overall
Risk of
Bias
Doubleblind RCT
145
patients
aged
65 years
admitted to
internal
medicine
service
The article
cites a theory
that delirium
may be related
to abnormal
tryptophan
metabolism,
which can be
regulated by
melatonin
supplementation
Olanzapine is
an
antipsychotic
with some
prior evidence
of efficacy for
delirium
treatment and
prevention.
Internal
medicine
service in a
tertiary care
center
Study medication
was administered (in
double-blind fashion)
daily between
1,800 and 2,400 h
depending upon
patient availability
and medication
administration
schedules for up to
14 days
Incidence of
delirium:
Melatonin:
2/56 (3.6%)
Placebo:
10/52 (19.2%)
RR = 0.19
(95% CI: 0.04-0.81)
P<0.02
2/61 patients
on melatonin
had side
effects of
nightmares or
hallucinations
Not
applicable
Moderate
Academic
hospital
Perioperative
olanzapine (5 mg
orally) or placebo was
administered before
and after surgery by
nurses not involved in
ongoing care of the
patients.
Incidence of
delirium:
Olanzapine:
28 (14.3%)
Placebo:
82 (40.2%)
RR = 0.36
(95% CI: 0.24-0.52)
P<0.0001
The difference was
also significant in
separate subgroups
(knee replacement,
hip replacement)
Severity of
delirium was
greater in the
olanzapine
group
(DRS-R-98
score:
16.44 vs.
14.5,
p = 0.02), and
lasted longer
(2.2 vs.
1.6 days,
p = 0.02).
Medical
complications
did not differ
significantly
between
groups.
Not
applicable
Moderate
Doubleblind RCT
400
patients
aged
65 years
undergoing
elective
knee or hip
replacement
surgery
D-100
Author/Year
Description
of PSP
Prakanrattana
and
Prapaitrakool
76
2007
Thailand
Atypical
antipsychotic
Risperidone
(1 mg) or
placebo taken
orally
(sublingually)
a single time
following
cardiac
surgery
Light propofol
sedation
during
hip repair
surgery
Sieber et al.
12
2010
USA
Study
Design
and
patients
Doubleblind RCT
126
patients
aged
>40 years
undergoing
elective
cardiac
surgery
Doubleblind RCT
114
patients
aged
65 years
undergoing hip
fracture
repair
Maldonado et
17
al. 2009
USA
Different
types of
post-op
sedation
after cardiac
surgery
RCT
118
patients
aged
18 years
undergoing
elective
cardiac
valve
surgery
Shehabi et al.
77
2009
Australia
Sedation
Dexmedetomidine vs.
Doubleblind RCT
306
Theory or
Logic Model
Description of
Organization
Implementation
Details
Outcomes:
Benefits
Outcomes:
Harms
Influence of
Contexts on
Outcomes
Overall
Risk of
Bias
Risperidone is
an
antipsychotic
with some
previous
evidence of
efficacy for
treatment of
delirium
Academic
hospital
Risperidone (1 mg
orally) or placebo was
given by nurses when
patients began to
wake in the ICU
Post-op delirium:
Risperidone:
7/63 (11.1%)
Placebo:
20/63 (31.7%)
RR = 0.35
(95% CI: 0.16-0.77)
P = 0.009
None
reported
(post-op
complications
did not differ
significantly
between
groups)
Not
applicable
Low
The authors
hypothesized
that minimizing
sedation depth
during spinal
anesthesia for
hip fracture
repair in
elderly
patients could
decrease the
occurrence of
postop
delirium
The authors
hypothesized
that dexmedetomidine
may be
associated
with a lower
incidence of
delirium due to
its
pharmacologic
properties
Dexmedetomidine is a
selective and
Academic
medical center
Implemented by
anesthesiologists
during surgery.
Post-op delirium:
Light sedation:
11/57 (19%)
Deep sedation:
23/57 (40%)
RR = 0.48
(95% CI: 0.26-0.89)
P = 0.02
Complication
rates were
similar in
both groups.
Light
sedation:
26/57 (46%)
Deep
sedation:
30/57 (53%
p = 0.57
Not
applicable
Moderate
Academic
medical center
Implemented in the
ICU following cardiac
surgery. Patients
were randomized to
three different
sedatives.
Not reported
Not
applicable
High
Two tertiary
referral
academic
Post-op delirium
(Intention-to-treat):
Dexmedetomidine:
4/40 (10%)
Propofol:
16/36 (44%)
Midazolam:
17/40 (44%)
p<0.001
Per protocol analysis
also significantly
different (p<0.001)
Incident Delirium:
Dexmedetomidine:
13/152 (8.6%)
Bradycardia
occurred
more often in
Not
applicable
Moderate
D-101
Author/Year
Description
of PSP
morphine,
effect on
prevalence of
delirium in
patients
at least
60 years old
after cardiac
surgery
Study
Design
and
patients
patients
aged
60 years
undergoing
cardiac
surgery
Theory or
Logic Model
Description of
Organization
Implementation
Details
Outcomes:
Benefits
Outcomes:
Harms
potent
2 adrenergic
receptor
agonist.
In theory,
its specificity
may provide
an advantage
for delirium
prevention
compared to
other
post-surgical
sedatives or
analgesics
hospitals
Morphine:
22/147 (15%)
Rate Ratio: 0.57
(95% CI: 0.26-1.1),
P = 0.09
Dex group
(16.5%) than
in the
Morphine
group (6.1%)
P = 0.006
Systolic
hypotension
occurred
more often in
Morphine
group
(38.1%)
compared to
Dex group
(23%)
P = 0.006
D-102
Duration of
delirium, median:
Dexmedetomidine:
2 days
Morphine: 5 days
(95% CI: 1.1-6.7)
P = 0.03
Influence of
Contexts on
Outcomes
Overall
Risk of
Bias
Author/Year
Description
of PSP
Hudetz et al.
78
2009
USA
Anesthetic
(NMDA
receptor
antagonist)
Ketamine
during
anesthetic
induction in
older patients
undergoing
cardiac
surgery with
CPB.
Mouzopoulos
79
et al. 2009
Greece
Local
anesthetic
Fascia iliac
block
prophylaxis
(via
Bupivacaine)
for
hip fracture
patients
Study
Design
and
patients
RCT
58 patients
aged
55 years
undergoing
cardiac
surgery
with CPB.
RCT
207
patients
aged
70 years
admitted
for hip
fracture
Theory or
Logic Model
Description of
Organization
Implementation
Details
Outcomes:
Benefits
Outcomes:
Harms
Influence of
Contexts on
Outcomes
Overall
Risk of
Bias
Citing prior
evidence that
ketamine may
have neuroprotective
effects, the
authors
hypothesized
that a single
dose of
ketamine
during
anesthetic
induction
would
attenuate
postop
delirium in
older patients
undergoing
cardiac
surgery with
CPB.
The authors
cite prior
studies
suggesting
that hip
fracture
patients are at
increased risk
of delirium due
to severe pain;
therefore, a
fascia iliac
block might
prevent
delirium by
preventing
severe pain.
Veterans
Affairs medical
center
Post-op delirium:
Ketamine:
1/29 (3.4%)
Placebo:
9/29 (31%)
RR = 0.11
(95% CI: 0.02-0.81)
P = 0.01
Not reported
Not
applicable
Moderate
Hospital (type
not reported)
(980 beds)
Bupivacaine was
injected on admission
(in blinded fashion)
and repeated daily
every 24 h until
delirium occurrence
or hip surgery; 24 hr
after surgery it was
re-administered and
repeated daily until
delirium occurrence
or discharge.
Incident delirium:
Prophylaxis:
10.8% (11/102)
Placebo:
23.8% (25/105)
OR = 0.45
(95% CI: 0.23-0.87)
No
complications
other than
3 local
hematomas
at injection
site which
resolved
spontaneously
Not
applicable
Moderate
D-103
Author/Year
Description
of PSP
Gamberini et
80
al. 2009
Switzerland
Acetylcholinesterase
inhibitor
Rivastigmine
administered
every 8 hrs
from night
before
surgery until
6th postop
day in a highrisk group of
elderly
patients
undergoing
elective
cardiac
surgery with
CPB
Study
Design
and
patients
Doubleblind RCT
120
patients
aged
65 years
undergoing
elective
cardiac
surgery
with CPB
Theory or
Logic Model
Description of
Organization
Implementation
Details
Outcomes:
Benefits
Outcomes:
Harms
Influence of
Contexts on
Outcomes
Overall
Risk of
Bias
Based on prior
studies
suggesting
cholinesterase
inhibitors can
successfully
treat delirium,
the authors
hypothesized
that short-term
administration
of oral
rivastigmine
would reduce
the incidence
of postop
delirium in a
high-risk group
of elderly
patients
undergoing
elective
cardiac
surgery with
CPB
Academic
hospital
Rivastigmine
administered every
8 hrs as a colorless
odorless solution from
night before surgery
th
until 6 postop day
Incident delirium
as assessed by
CAM:
Rivastigmine:
18/56 (32%)
Placebo:
17/57 (30%)
RR = 1.12
(95% CI: 0.50-2.48)
P = 0.80
No significant
betweengroup
difference for
any adverse
events.
Not
applicable
Low
D-104
Author/Year
Description
of PSP
Liptzin et al.
81
2005
USA
Acetylcholinesterase
inhibitor
Donepezil
(given at
5 mg/day) or
placebo for
14 days
preop and
14 days
postop in
patients
undergoing
total joint
replacement
(knee or hip)
Study
Design
and
patients
Doubleblind RCT
80 patients
aged
50 years
undergoing knee
or joint
replacement
Theory or
Logic Model
Description of
Organization
Implementation
Details
Outcomes:
Benefits
Outcomes:
Harms
Influence of
Contexts on
Outcomes
Overall
Risk of
Bias
Donepezil is a
cholinesterase
inhibitor
(disruption in
cholinergic
transmission is
thought to be
in causal
pathway of
delirium)
Academic
medical center
Post-op delirium:
Donepezil:
8/39 (20.5%)
Control:
7/41 (17.1%)
Rate Ratio = 1.2
(95% CI: 0.6-2.6)
P = 0.69
Not reported
Not
applicable
Moderate
D-105
Author/Year
Description
of PSP
McCaffrey et
82
al. 2006
USA
Music therapy
(musical
selection with
bedside CD
turned on
1-3 times/day
+ standard
postop care
from
anesthesia
awakening
time until
discharged)
for patients
undergoing
hip or knee
surgery
Study
Design
and
patients
RCT
(music
therapy +
usual care
vs.
usual care
alone)
124
patients
aged
65 years
undergoing
elective
hip or knee
surgery
Theory or
Logic Model
Description of
Organization
Implementation
Details
Outcomes:
Benefits
Outcomes:
Harms
Influence of
Contexts on
Outcomes
Overall
Risk of
Bias
Prior studies
have shown
evidence that
music can
improve
cognition and
calm agitated
patients
Large tertiary
care center
Nurses blinded to
room designation
made room
assignments. Various
CDs were available in
the music therapy
rooms. Music was
played when patients
were awakening from
anesthesia. CD was
set to play for 1 hour
4 times daily. Also,
nurses were asked to
turn on the music
each time they
entered the room,
and family members
and patients were
instructed how to use
the CD player.
Research assistants
checked that CD
players were working
and that the music
and timing of play
suited patient
preferences.
Patients who
experienced acute
confusion:
Music therapy:
2/62 (3.2%)
Usual care:
36/62 (58.1%)
RR = 0.06
(95% CI: 0.01-0.22)
P<0.0001
None
reported
Not
applicable
High
D-106
Author/Year
Description
of PSP
McCaffrey and
83
Locsin 2004
USA
Music therapy
(musical
selection with
bedside CD
turned on
1-3 times/day
+ standard
postop care
from
anesthesia
awakening
time until
discharged)
for patients
undergoing
elective hip
and knee
surgery
Study
Design
and
patients
RCT
(music
therapy +
usual care
vs.
usual care
alone)
66 patients
aged
65 years
undergoing
elective
hip or knee
surgery
Theory or
Logic Model
Description of
Organization
Implementation
Details
Outcomes:
Benefits
Outcomes:
Harms
Influence of
Contexts on
Outcomes
Overall
Risk of
Bias
Prior studies
have shown
evidence that
music can
improve
cognition and
calm agitated
patients
Large tertiary
care center
Nurses blinded to
room designation
made room
assignments. Various
CDs were available in
the music therapy
rooms. Music was
played when patients
were awakening from
anesthesia. CD was
set to play for 1 hour
3 times daily. Also,
nurses were asked to
turn on the music
each time they
entered the room,
and family members
and patients were
instructed how to use
the CD player.
Research assistants
checked that CD
players were working
and that the music
and timing of play
suited patient
preferences.
Significantly fewer
patients in the music
therapy group had
episodes of
confusion and
delirium (F = 19.568,
P = 0.001)
None
reported
Not
applicable
High
D-107
Author/Year
Description
of PSP
Kalisvaart et
84
al. 2005
The
Netherlands
Typical
antipsychotic
Haloperidol or
placebo
(0.5 mg
3 times daily)
was started
on admission
and
continued
until 3 days
postop to
prevent
delirium after
hip surgery
Aizawa et al.
85
2002
Japan
Benzodiazepines
Diazepam +
flunitrazepam
drip infusion
and pethidine
drip infusion
for first
3 days (day
of operation
and first 2
postop days)
in patients
undergoing
gastrointestinal
surgery
Study
Design
and
patients
Doubleblind RCT
430
patients
aged
70 years
undergoing hip
surgery
RCT
(deliriumfree
protocol
[DFP] vs.
non-DFP)
40 patients
aged
>70 years
undergoing
gastrointestinal
surgery
Theory or
Logic Model
Description of
Organization
Implementation
Details
Outcomes:
Benefits
Outcomes:
Harms
Influence of
Contexts on
Outcomes
Overall
Risk of
Bias
Haloperidol is
a dopamine
antagonist
which can
enhance
acetylcholine
release. Since
delirium is
highly
associated
with
cholinergic
deficiency,
the authors
hypothesized
that
haloperidol
may have an
indirect
beneficial
effect on
delirium.
Sleep-wake
cycle disorders
have been
reported to be
associated
with postop
delirium, so
medications
that target
sleep cycle
disorders
might prevent
delirium
Teaching
hospital
Haloperidol (0.5 mg
3 times daily) or
placebo was started
on admission and
continued until 3 days
after surgery.
Experienced geriatric
nurses and
geriatricians provided
proactive geriatric
consultation (based
on a structured
multimodal protocol)
to all patients.
Post-op delirium:
Haloperidol:
32/212 (15.1%)
Placebo:
36/218 (16.5%)
RR = 0.91 (95% CI
0.59-1.42)
P = 0.69
Duration of
delirium (days):
Haloperidol: 5.44.9
Placebo: 11.87.5
P<0.001
No drugrelated side
effects were
reported
Not
applicable
Moderate
A city hospital,
no other
details
provided
DFP was
reported to
cause
morning
lethargy in
8/20 patients
(40%). No
other side
effects were
reported.
Not
applicable
High
D-108
Author/Year
Description
of PSP
Long-term care
Mentes and
Hydration
86
Culp 2003
(individually
USA
calculated
fluid intake
goal) over an
8-week
period in
nursing home
residents
aged
65 years
Moretti et al.
87
2004
Italy
Rivastigmine
(3-6 mg/day)
for 2 years in
patients with
vascular
dementia
Study
Design
and
patients
Theory or
Logic Model
Description of
Organization
Implementation
Details
Outcomes:
Benefits
Outcomes:
Harms
Influence of
Contexts on
Outcomes
Overall
Risk of
Bias
QuasiRCT
(randomization by
coin toss
of different
participating
facilities)
49 participants aged
65 years
Prior studies
have shown
that chronic
underhydration may
lead to
delirium and
other adverse
events
2 Veterans
Administration
(VA),
2 community
nursing homes
Episodes of acute
confusion:
Treatment:
0/25 (0%)
Control:
2/24 (8.2%)
P = not significant
None
reported
Not clear,
but the
possibility
was raised
that control
group staff
might have
altered their
standard
hydration
practices
due to
awareness
of research
staff data
collection.
High
RCT
(Rivastigmine vs.
cardioaspirin)
246
patients
aged 6885 years
with
vascular
dementia
Delirium in
patients with
vascular
dementia
might be due
to lack of
acetylcholine
in the brain.
Rivastigmine
is an anticholinesterase
inhibitor
Academic
hospital
Patients with
episodes of
delirium during
follow-up:
Rivastigmine:
46/115 (40%)
Cardioaspirin:
71/115 (62%)
RR = 0.65
(95% CI: 0.50-0.85)
P<0.001
Mean duration of
delirium:
Rivastigmine:
4 1.71 days
Cardioaspirin:
7.86 2.73 days
P<0.01
Not reported
Not
applicable
High
D-109
References
1.
8.
Lin RY, Heacock LC, Fogel JF. Druginduced, dementia-associated and nondementia, non-drug delirium hospitalizations
in the United States, 1998-2005: an analysis
of the national inpatient sample. Drugs
Aging 2010 Jan 1;27(1):51-61.
PMID: 20030432
2.
9.
10.
11.
12.
13.
14.
15.
16.
3.
4.
5.
6.
7.
D-110
17.
27.
18.
28.
19.
29.
20.
30.
21.
31.
32.
33.
34.
35.
22.
23.
24.
25.
26.
D-111
36.
46.
37.
47.
38.
48.
49.
39.
40.
50.
41.
51.
42.
52.
53.
54.
43.
44.
45.
D-112
55.
56.
57.
58.
65.
66.
67.
68.
69.
70.
71.
72.
73.
59.
60.
61.
62.
63.
64.
D-113
74.
75.
76.
Prakanrattana U, Prapaitrakool S.
Efficacy of risperidone for prevention of
postoperative delirium in cardiac surgery.
Anaesth Intensive Care 2007 Oct;35(5):7149. PMID: 17933157
77.
78.
79.
80.
81.
82.
D-114
83.
84.
85.
86.
87.
88.
89.
90.
91.
92.
93.
94.
95.
96.
97.
98.
99.
100.
101.
D-115
102.
103.
104.
105.
106.
107.
108.
109.
110.
111.
112.
113.
114.
115.
116.
117.
118.
119.
120.
121.
122.
123.
124.
125.
126.
127.
128.
129.
130.
D-116
131.
132.
133.
Papaioannou A, Fraidakis O,
Michaloudis D, et al. The impact of the type
of anaesthesia on cognitive status and
delirium during the first postoperative days
in elderly patients. Eur J Anaesthesiol 2005
Jul;22(7):492-9. PMID: 16045136
134.
135.
136.
137.
138.
139.
140.
141.
D-117
Description of PSP
Study
Design
Lynch and
1a
Vickery 2010
Zero-tolerance
philosophy
Target safety
problem: PU
Key elements:
Multidisciplinary
team; educate
staff/resident/family,
streamline
documentation,
wound-care
workshops,
case studies,
setting goals,
identify and address
barriers
Pre-post
Theory or
Logic
Model
NS
Description
of
Organization
166-bed
acute
rehabilitation
Contexts
Implementation Details
Outcomes:
Benefits
External:
New CMS reimbursement
Organizational
Characteristics:NS
Teamwork, Leadership,
Culture:
After reviewing 2007 data
on PUs, the team was
dismayed at the number of
misidentification of PUs
at admission; skin
assessments incomplete
and inconclusive;
incorrect staging;
incorrect documentation
(e.g., document denuded
skin as PU); documentation
fragmented; definition of
thorough skin assessment;
inconsistent documentation
of interventions; incorrectly
transcribing interventions to
appropriate documentation.
Implementation tools:
Interdisciplinary team
Education
at orientation, annually,
and one-on-one,
via web
Documentation
streamlined to 1 form
Wound care workshops
for nurses
at orientation; after
2 months
Report cards
Length: 1 year
Process:
Multidisciplinary team
reviews current processes
of care and finds errors with
assessment and
documentation; education
of staff is quickly put in
place; staff is encouraged to
report HAPUs and view as
an opportunity to learn; rate
goals are set for hospital
and by unit; report cards are
posted so units can track
their progress.
Successes:
Due to streamlining
documentation, timely and
accurate completion of
documentation increases
from 60% to 90% in
90 days; patients on a
neurobehavioral stroke unit
did not develop PU
Barriers:
Patients dissatisfied with
off-loading boots
Addressing Barriers:
Trial initiated to evaluate
use of pillow; leads to
improved outcomes
Sustainability:
Quarterly newsletter
attached to paychecks
PU Rates:
Pre: 2.8%
Post: 0.48%
(-82.8%)
D-118
Influence of
Contexts on
Outcomes
NS
Author/Year
Description of PSP
Study
Design
Young et al.
2a
2010
Clinician-led
task force
leads prevention
initiatives
Target safety
problem: PU
Key elements:
Clinician-led
task force;
skin champions;
adoption of
Save our Skin logo;
education/training;
revise policies and
procedures based
on CPGs; integrate
new documentation
and assessment
tools
Pre-post
Theory or
Logic
Model
Shared
governance
approach
(decisions
made
at point of
care)
Description
of
Organization
540-bed
acute care
facility
(3 campuses)
in Indiana
Contexts
Implementation Details
Outcomes:
Benefits
External:NS
Organizational
Characteristics: NS
Teamwork, Leadership,
Culture:
A clinician-led skin care
team replaces
administration-led
Implementation tools:
Save our Skin (SOS)
logo adopted and
appears on educational
forms and t-shirts worn
during audits and
educational in-services
Laminated SOS logo
tool placed on doors of
patients at-risk
Revise online policies
and procedures; 1 new
policy remains
Adopt Braden Scale
(electronically)
Body map assessment
tool
Mandatory training
includes presentation
of case studies
Educational brochure
for residents and
families
Feedback on
educational
presentations
Updating of core
orientation, uploaded to
hospital Web site
Posters depicting PU
rates, examples of new
forms, and revised
policies/procedures
Length: 2 years
Process:
Members of clinician-led
task force include director of
Clinical Care and Oncology
Nursing Services, manager
of Wound Care Institute,
and nursing representatives
from 15 hospital units;
task force members appoint
skin champions from each
unit; champions invited to
join team; task force
members join
subcommittees of choice to
develop logo, policy and
procedures and other
program components;
after comparing practices to
CPGs 7 existing policies are
combined into 1; manager
of the Wound Care Institute
works with the Director of
Informatics on revising
online policies and
procedures; monthly quality
audits
Successes:
Revised policies reduced
from 7 to 1; documentation
of skin care reduced from
8 to 3
Barriers:
Time constraints,
insufficient computer
resources, competing goals
Addressing Barriers:
Clinicians were allocated
4 paid hours to carry out
responsibilities; web access
to library resources were
Incidence:
Pre:
Campus 1:
12.5%
Campus 2:
8.7%
Campus 3:
NR
Post:
Campus 1:
9.1%
Campus 2:
2.8%
Campus 3:
NR
D-119
Influence of
Contexts on
Outcomes
NS
Author/Year
Bales et al.
3
2009
Description of PSP
Implementation of
evidenced-based
prevention
strategies
Target safety
problem: PU
Key elements:
Increase hours of
certified wound,
ostomy and
continence nurses
(CWOCNs) to fulltime to provide
24-hour support to
staff, provide
mandatory
education and
resources for
staging and treating
wounds, increase
wound monitoring
and reporting
efforts, purchase
pressureredistribution beds,
add musical alarms
to remind nurses to
turn patients, and
identify at-risk
surgical patients.
Study
Design
Time
series
Theory or
Logic
Model
NS
Description
of
Organization
300-bed
community
hospital, USA
Contexts
Implementation Details
added to intranet;
Sustainability:
RNs and LPNs must
demonstrate
competency annually
Monthly updates
provided via intranet to
nursing personnel by
unit champions/ team
members; includes
product changes
Length: 1 year
Process:
24-hour support
provided by CWOCNs
Mandatory education
Strict oversight of
monitoring and
reporting
Periodical motivational
campaigns that
included staff and unit
incentives.
Successes:
Patients received
optimal care
Institution avoided the
cost of treating stage 3
or 4 ulcers
Barriers:
Staff motivation and lack of
proper reporting and
documentation
D-120
Addressing Barriers:
Monthly to quarterly
campaigns are launched to
maintain staff motivation.
Nursing units that had zerohospital acquired PUs are
recognized and awarded
Outcomes:
Benefits
Influence of
Contexts on
Outcomes
Hospital
acquired
prevalence
rates:
Pre: 4.20%
Post: 0%
The hospitals
managerial
style
encouraged
staff
involvement
in decisionmaking about
the process of
developing a
program and
the leadership
team gave
strong
support to the
program and
promoted it to
both other
leaders in the
team and
hospital staff.
Author/Year
Description of PSP
Study
Design
Theory or
Logic
Model
Description
of
Organization
Contexts
Implementation Details
during campaigns.
Sustainability:
Success requires
awareness of
key management skills and
priorities, such as
strong leadership,
involvement of staff in
decision-making and a
desire to foster
interdisciplinary
relationships.
D-121
Outcomes:
Benefits
Influence of
Contexts on
Outcomes
Author/Year
Description of PSP
Study
Design
Chicano &
Droishagen,
4
2009
Implement
strategies to lower
the incidence of
hospital-acquired
pressure ulcers
Target safety
problem: PU
Key elements:
Developed a
protocol for
assessment and
documentation of
wounds,
implemented
procedures for
CWOCN to work
with staff and
patients to initiate
appropriate
treatment,
implemented the
Braden Scale for
Pressure Sore Risk,
conducted a
literature review on
the use of
thromboembolic
device stockings
and compression
devices and revised
practice standards
for use of devices
based on findings of
review
Time
series
Theory or
Logic
Model
NS
Description
of
Organization
25-bed
intermediate
care unit in
the
United States
Contexts
Implementation Details
Outcomes:
Benefits
External:
Quarterly HAPU data
indicate increased
prevalence
Organizational
Characteristics: NS
Teamwork, Leadership,
Culture:
The care unit supported a
self-governing nursing
council.
Implementation tools:
Survey to identify practices
regarding skin assessment,
documentation, and
nursing intervention and
opportunities for education,
CWOCNs, and
Braden Scale
Length: 23 months
Process:
Phase 1 took 5 months and
involved developing
protocols and procedures to
assess and treat wounds;
Phase 2 took 3 months to
complete and involved
educating staff and
implementing the Braden
Scale,
Phase 3 took 15 months to
complete and involved a
literature review and
revision of practice
standards on use of
compression devices.
Successes:
Staff participation in survey,
continued adherence to
implemented prevention
practices, development of
educational materials, and
staff acceptance of shared
governance
Barriers:
Engaging staff as council
members in the planning
and implementation of the
project.
Addressing Barriers:
Updating staff of progress
and continual
encouragement to
participate from other
council members
Sustainability:
Commitment and diligence
from the quality improvement team and selfgovernance council.
Hospital
acquired
incidence
rate:
Pre:
6 occurrences
during a
12 month
period,
5 during
subsequent
5 months
D-122
Post:
1 occurrence
within
12 months
following
implementation, 0 at
latest
assessment
covering
2 month
period.
Influence of
Contexts on
Outcomes
Commitment
and diligence
from the
quality
improvement
team and
from the
members of
the staffs
selfgovernance
councils
played a
significant
factor in
achieving our
goal of
reducing
HAPU
prevalence in
our
intermediate
care unit.
Author/Year
Description of PSP
Study
Design
Walsh et al.
5a
2009
Implementation of
evidence-based
practices
Target Safety
problem: PU
Key elements:
revision of skin
management
program; use of
CPGs; educating
clinician/nurse;
multidisciplinary
team; add certified
WOCN to
management;
replace risk
assessment tool;
replace wound care
products
Time
series
Theory or
Logic
Model
NS
Description
of
Organization
1 acute care
facility in
northwest CT
Bed size: 371
Contexts
Implementation Details
Outcomes:
Benefits
External: CMS
Organizational
Characteristics:
Regional medical center
and community teaching
hospital; primary provider
for 350,000
Teamwork, Leadership,
Culture: NS
Implementation tools:
Add WOCN nurse to
team
Rely on EBPs
(AHRQ CPGs, IHI,
WOCN Society)
Rely on The National
Database of Nursing
Quality Indicators PU
presentation for reeducation on wound
etiology and staging
Clinician and staff
education (computerbased and classroom
presentations)
Replace risk
assessment tool with
Braden Scale
Update skin
management policies/
procedures
Assessed wound care
products
Multidisciplinary team
Alert system (POA
sticker)
Visual turning clocks,
laminated pocket cards
including CPG
information
Bed surface algorithms
Length: 18 months
Process:
Clinical education relies on
6 essential elements of
prevention; in 2007,
Braden scale risk
assessment tool replaces
current un-validated tool
form (not research based);
each unit assigns an
interdisciplinary team;
purchases of new beds,
stretchers and curtains
followed by new skin lotion
and incontinence care
products
Successes:
Reduction in prevalence;
increased focused
communication among
patient caregivers; buy-in
from clinicians improves
behavior
Barriers: NS
Addressing Barriers: NS
Sustainability:
Remains current regarding
initiatives for improved
patient safety, changes in
regulatory mandates, and
changes in EBP.
Prevalence:
Baseline:
12.8%
Postimplementation:
0.6%
D-123
Influence of
Contexts on
Outcomes
NS
Author/Year
Dibsie, L. 2008
Description of PSP
Implementation of
evidence-based
protocol and
practices for
preventing and
treating PUs
Target safety
problem: PU
Key elements:
Development of
protocol for
monitoring,
preventing, and
treating PUs based
on recent evidence,
standardization of
all products related
to prevention and
treatment of PUs,
and education for
nurses on the
protocol and use of
products.
Study
Design
Time
series
Theory or
Logic
Model
NS
Description
of
Organization
4 adult critical
care units
(54 beds
total) at
2 academic
medical
centers in the
United States
Contexts
Illustrative wound
reference guides with
recommended
treatment modalities
Standardize wound
care products
External:
Two significant events
occurred and there was an
overall lack of reporting and
communication of issues
related to skin breakdown
(specifics of events not
reported)
Organizational
Characteristics: NS
Teamwork, Leadership,
Culture: NS
Implementation tools:
Staff nurse skin committee
and skin champions
D-124
Implementation Details
Outcomes:
Benefits
Influence of
Contexts on
Outcomes
Surgical ICU
acquired:
Pre: 6.1%
Post: 6.1%
Facility-wide
overall rate:
Pre:4.2%
Post: 3.2%
The changes
in the climate
and practice
related to skin
care and
prevention of
breakdown
are the direct
result of
nursing taking
ownership of
their practice
with the
support of
nursing
leaders at all
levels.
Author/Year
Description of PSP
Study
Design
Theory or
Logic
Model
Description
of
Organization
Contexts
Implementation Details
D-125
Outcomes:
Benefits
Influence of
Contexts on
Outcomes
Author/Year
Description of PSP
Study
Design
McInerney, J.
7
2008
To implement
multiple strategies
for decreasing the
prevalence of
hospital-acquired
PUs
Target safety
problem: PU
Key elements:
Electronic medical
records, risk
assessment
measures, pressure
relief measures
(new equipment
and personnel
augmentation, and
interdisciplinary
team to develop
protocols.
Time
series
Theory or
Logic
Model
NS
Description
of
Organization
Two-hospital
system with
548 beds in
United States;
548 bed; nonprofit
Contexts
Implementation Details
Outcomes:
Benefits
External: NS
Organizational
Characteristics: NS
Teamwork, Leadership,
Culture: NS
Implementation tools:
WOCN oversaw
implementation of
strategies.
Length:
18 months to implement
program; follow-up reported
for over 5 years
Process:
Electronic medical
records (EMRs) to
assess and document
skin care needs.
Risk assessment
measures (e.g., Braden
Scale)
Automated consults
and orders through
EMRs
Pressure relief
measures
Staff education
Hiring of second
WOCN
Successes:
Reduction in PUs, cost
savings and elimination of
pain and suffering for the
patients
Barriers: NS
Addressing Barriers: NS
Sustainability: NS
Hospital
acquired
prevalence
rates:
Pre:
12.8% all
PUs;
6.7% PUs on
heel.
Post
(4.5 years
after
implementation):
1.9% all PUs;
1.1% PUs on
heel.
D-126
Influence of
Contexts on
Outcomes
With the
assistance of
the
automated
consults and
orders, the
addition of
another WOC
nurse, the
appropriate
equipment,
the interdisciplinary
task force,
continuing
education,
and
monitoring,
the hospital
system was
able to reduce
the hospitalacquired
pressure ulcer
prevalence
rate by 81%,
and the rate
for heel ulcers
alone was
reduced by
90%.
Estimated
annual cost
savings:
$11,466,000
Author/Year
Description of PSP
Study
Design
Ballard et al.
8
2007
Implementation of
multiple strategies
in an intensive care
unit (ICU) to reduce
the rate of PUs
Target safety
problem: PU
Key elements:
Strategies included:
restructured risk
assessment and
documentation,
translated numeric
data into easy-tounderstand graphs
of PU rates,
increased staff
awareness,
implemented
turn rounds,
increased
prevalence
assessments and
redesigned
skin team, used
evidence-based
practices for
monitoring and
treating PUs, and
created an access
database to track
weekly prevalence
Time
series
Theory or
Logic
Model
NS
Description
of
Organization
Two-unit ICU
with a total of
44 beds
located in
two separate
geographical
locations in
the
United States.
Contexts
Implementation Details
Outcomes:
Benefits
External:
Joining the National
Database of Nursing Quality
Indicators (NDNQI) and
realizing that ICU had high
prevalence of PUs.
Organizational
Characteristics: NS
Teamwork, Leadership,
Culture:
Primary nursing model
Implementation tools:
CWOCN conducted a
needs assessment, creation
of user friendly reports to
show rate of PUs, posted
data of PU rates for staff to
see, skin teams (consisted
of nursing staff who
performed weekly
prevalence assessments
and provided education),
and Access database
Length: 1 year
Process:
Conducted needs
assessment to identify
areas of weakness in
identifying, monitoring,
treating, and reporting PUs;
made revisions to protocol
based on results of needs
assessment; created user
friendly reports to display
PU rates; increased staff
awareness through
displaying PU rates and
providing education;
implemented turn rounds
every two hours;
redesigned skin team,
implemented evidencebased practices to assess
risk and monitor PUs, and
implemented Access
database to track PUs.
Successes:
Reduction in rate of PUs
and improved patient
outcomes.
Barriers: NS
Addressing Barriers: NS
Sustainability:
Staff commitment to
implementing strategies and
maintaining quality care.
Hospital
acquired
incidence
rate: NS
Hospital
acquired
prevalence
rates:
Pre: 34%
Post: 8.0%
D-127
Influence of
Contexts on
Outcomes
Utilizing
benchmark
data helped
the ICU focus
on pressure
ulcer
prevention,
which led to
improved
patient
outcomes.
Author/Year
Description of PSP
Study
Design
Catania et al.
9
2007
Design and
implementation of
the Pressure Ulcer
Prevention Protocol
Interventions
(PUPPI): a nursing
initiative to prevent
PUs
Target safety
problem: PU
Key elements:
The PUPPI involves
assessing risk and
nutritional status,
providing skin care,
documenting, and
giving referrals
as needed.
Time
series
Theory or
Logic
Model
NS
Description
of
Organization
All 5 inpatient
units in one
hospital in the
United States.
Units included
2 medical
units,
2 surgical
units, and
one critical
care unit.
Contexts
Implementation Details
Outcomes:
Benefits
External:
2 stage IV ulcer identified;
evidence from the NDNQI
survey that the prevalence
of PUs in the hospital in the
study exceeded the national
benchmark by close to
50%.
Organizational
Characteristics:
Dedicated cancer hospital
Teamwork, Leadership,
Culture: NS
Implementation tools:
Quality improvement team
that consisted of a quality
manager, nursing director,
certified nurse aids (CNSs),
nursing staff developmental
specialists, and an
enterostomal therapy nurse
to develop and implement
protocol.
Length:
6-months to implement;
follow-up data reported for
18 months
Process:
Initial efforts started in 2003
and involved having clinical
nurse specialists assess
patient risk using the
Braden Scale. These efforts
led to the development of a
quality-improvement team
in 2004 and the
development of the
PUPPIs. The PUPPI was
implemented in
September 2004 and
included a systematic
process for monitoring and
educating staff.
Successes:
Reduction in rates of PUs
Barriers: NS
Addressing Barriers: NS
Sustainability:
Proactive nursing staff who
have adopted initiatives in
protocol into their daily
routine.
Hospital
acquired
incidence
rate: NS
Hospital
acquired
prevalence
rates:
Pre:
11.11% all
ulcers;
6.67%
hospital
acquired
Post:
4.08% all
ulcers;
1.36%
hospital
acquired
D-128
Influence of
Contexts on
Outcomes
While the
unit CNSs
have
championed
this process
and continue
to monitor the
program, it
has been the
nursing staff
who have
embraced
evidencebased nursing
practice and
brought it to
the bedside
by adopting
the initiative
into daily
practice.
Author/Year
Description of PSP
Study
Design
LeMaster, K.
10
2007
Pressure Ulcer
Prevention Project
implemented on
targeted units
pulmonary unit and
oncology unit
Target safety
problem: PU
Key elements:
Turning at-risk
patients every
two hours
minimally, placing a
pressure-reducing
overlay on the bed
of every patient at
risk, and elevating
bony prominences
at risk
Time
series
Theory or
Logic
Model
NS
Description
of
Organization
Pulmonary
and oncology
unit of the
largest
hospital
campus
within one
healthy
system;
a 502-bed
hospital in the
United States.
Contexts
Implementation Details
Outcomes:
Benefits
External:
Selection of study units was
based on unit having a
higher hospital-acquired PU
rate than the NDNQI
database mean for similar
units and having higher
hour-of-care ratios than the
NDNQI mean for similar
units.
Organizational
Characteristics: NS
Teamwork, Leadership,
Culture: NS
Implementation tools:
Manual containing
information about wounds
and wound care,
instructions on the use of
the Braden Scale to assess
risk for developing PUs,
patient turn schedule, and
cues to use as reminders to
turn patients.
Length:
Summer 2004April 2005
Process:
The first phase of
implementation involved
assessing and establishing
baseline knowledge of unit
staff nurses for assessing
risk. Staff then identified
resources and studied the
manual. Nursing staff began
assessing and documenting
risk and implementing other
aspects of the protocol
(placing pressure-reducing
overlay on bed). A CNS
provided consultation and
oversight throughout
implementation period.
Successes:
Reduction of PUs in
targeted units and
successful duplication of
intervention in other medical
units.
Barriers:
Braden scores were not
documented at 100% per
policy. Patients were
missed because of failure in
communication between
two different electronic
documentation systems.
Addressing Barriers:
Barrier to be eliminated with
transition to a single,
universal electronic record
system within hospital.
Sustainability:
Manual and cues to help
maintain consistent and
complete practice patterns.
Hospital
acquired
incidence
rate: NS
Hospital
acquired
prevalence
rates:
Pre
Pulmonary
Unit: 9.0%
Post
Pulmonary
Unit: 0.0%
Pre Oncology
Unit: 12.0%
Post
Oncology
Unit: 0.0%
D-129
Influence of
Contexts on
Outcomes
NS
Author/Year
Description of PSP
Study
Design
Courtney et al.
11
2006
To develop and
implement Save
Our Skin program to
reduce the rate of
PUs
Target safety
problem: PU
Key elements:
Updating pressurerelieving mattress,
introducing skin
breakdown
prevention
protocols, clarifying
staff roles and
responsibilities
(introduced a skin
champion), and
improving
measurement a
communication of
PU performance
data
Time
series
Gibbons et al.
12
2006
To develop and
implement best
practice guidelines,
known as the
SKIN bundle
(Surfaces, Keep the
patient turning,
Incontinence
management,
Time
series
Theory or
Logic
Model
Used
procedures
of the
Six Sigma
method, a
datafocused,
decisionmaking
process
that utilizes
a five
phase
process
called
DMAIC:
Defining
the
problem,
Measuring
the
performance,
Analyzing
the
problem,
Improving
the
situation,
and
Initiating
change.
NS
Description
of
Organization
710-licensed
bed, multisite,
not-for-profit
facility that
serves a
37-county
area; is
Magnet
designated
Contexts
Implementation Details
Outcomes:
Benefits
External:
Results using the Nursing
Care Quality Initiative
guidelines that revealed
high prevalence of PUs
(13%) and lack of
documentation and
management. Revitalized
interest in treatment and
prevention shown by
American Nurses
Association and AHRQ in
developing new guidelines.
Organizational
Characteristics:
Magnet designated hospital
Teamwork, Leadership,
Culture: NS
Implementation tools:
Implemented guidelines for
the prevention and
management of PUs from
the Wound, Ostomy, and
Continence Nurses Society
and assessed risk using the
Braden Scale.
Hospital
acquired
incidence
rate:
Pre: 9.4%
Post:
st
1 quarter
implementation 3.1%;
last follow-up
1.8%
Hospital
acquired
prevalence
rates: NS
Large
528-bed
hospital, part
of nations
largest
Catholic and
non-profit
health system
External:
Development of the SKIN
bundle is part of the
Ascension Health Care
systems initiative to
reduce/eliminate
preventable hospital-related
injuries and deaths.
Organizational
Length:
Follow-up 3 years
Process:
Adopted Six Sigma
procedures, assessed
potential causes of high
incidence of PUs and lack
of staff coordination and
management of PUs, and
introduction of solutions:
staff training and
awareness, development
and implementation of
Skin Breakdown Prevention
protocol, replacement of
pressure mattresses,
designation of
Skin champion,
clarification of staff roles,
and implementing
monitoring procedures
Successes:
Reduced incidence of PUs
and cost savings
Barriers: NS
Addressing Barriers: NS
Sustainability:
Defining staff
responsibilities, monitoring
performance, using data to
inform staff performance,
and making data public
Length:
10 months to implement;
follow-up 2 years
Influence of
Contexts on
Outcomes
This project
refocused
efforts on
traditional
direct nursing
care and
problem
solving
procedures
from the
Six Sigma
method to
implement the
Save Our
Skin program.
Hospital
acquired
incidence
rate:
Pre: 5.7%
Post: 0.448
No new
Stage III or IV
HAPU
Of eight
priorities
identified for
action by
Ascension
Health;
St. Vincents
Medical
Center was
D-130
Process:
Began with engaging
leadership, forming an
interdisciplinary team, and
providing protected time to
Author/Year
Description of PSP
Nutrition) to prevent
PUs
Target safety
problem: PU
Key elements:
Developed a
synergistic group
of interventions that
includes appropriate
surface selection
(e.g., pressure
mattress), regular
turning of patients,
incontinence
management,
nutrition and
hydration, ongoing
monitoring and staff
training.
Study
Design
Theory or
Logic
Model
Description
of
Organization
Contexts
Implementation Details
Outcomes:
Benefits
Characteristics:
Faith-based, non-profit
hospital
Teamwork, Leadership,
Culture: NS
Implementation tools:
Regular assessment and
documentation on flow
sheets, skin risk alert
reminders, weekly team
meetings, and ongoing
performance monitoring and
reporting.
occurred
between
August 2004
and February
2006
D-131
Influence of
Contexts on
Outcomes
selected to
develop the
PU process.
The hospital
leadership
welcomed
the
opportunity to
develop this
nursing-driven
program as a
means of
establishing
pride in
professional
nursing
practice.
67 acute care
facilities in the
Ascension
health system
agreed to
implement the
SKIN bundle
plus common
measures of
quality and
performance.
Author/Year
Description of PSP
Study
Design
Hiser, et al.
13
2006
To implement a
team approach to
performance
improvement and
develop an
education plan for
clinical staff to
better prevent and
treat PUs
Target safety
problem:
PU
Key elements:
Education,
policy changes,
development of
evidence-based
protocols, cost
improvement
strategies,
implementing new
support surfaces,
and improved
reporting and
monitoring through
quarterly
prevalence studies
and improved risk
assessment using
the Braden scale.
Time
series
Theory or
Logic
Model
NS
Description
of
Organization
580-bed
regional
medical
facility in the
U.S.
Contexts
Implementation Details
Outcomes:
Benefits
External: NS
Organizational
Characteristics: NS
Teamwork, Leadership,
Culture: NS
Implementation tools:
Created a Wound Care
Team that consisted of
CWOCNs and an advanced
registered nurse practitioner
to implement changes and
educate staff; replaced the
Norton Scale with the
Braden scale to assess risk.
Length:
2 years follow-up.
Process:
Implementation started with
a review of the literature of
best practices for
prevention and treatment of
PUs.
Successes:
Reduced prevalence of PUs
and annual cost savings.
Barriers: NS
Addressing Barriers: NS
Sustainability:
Ongoing education and
newsletters reporting
progress and positive
feedback to staff.
Hospital
acquired
prevalence
rates:
Pre: 9.2%
Post: 6.6%
(measured at
2 years
follow-up)
D-132
Influence of
Contexts on
Outcomes
NS
Author/Year
Description of PSP
Study
Design
Lyder et al.
14
2004
To implement a
multihospital
collaboration to
increase the
identification of
patients at high risk
of PUs and to
promote the use of
preventive
measures among
hospitalized
Medicare patients
Target safety
problem: PU
Key elements:
Increase in the
following: tracking
of PUs,
performance and
documentation of
risk assessment,
use of prevention
protocol (includes
education and
oversight of staff),
scheduled
repositioning, use of
pressure-reducing
devices, nutritional
consults, and
accuracy of staging.
Pre-post
Theory or
Logic
Model
Plan-DoStudy-Act
(PDSA)
Description
of
Organization
17 hospitals
ranging from
200 to 800
beds with
9 located in
urban and
8 in rural
settings in the
U.S.
Contexts
Implementation Details
Outcomes:
Benefits
External:
In response to Centers for
Medicare & Medicaid
Services charge to improve
quality of care to Medicare
patients.
Organizational
Characteristics: NS
Teamwork, Leadership,
Culture:
Hospitals needed to
develop a team approach to
implementing changes.
Implementation tools:
Qualidigm, the Connecticut
QIO,
Computerized charting
system for tracking PUs,
creation of skin care task
force, and pocket-sized
wound staging card.
Length:
9 months implementation,
2 years follow-up.
Process:
The PDSA framework
involved 1) identifying
problem to be changed and
designing an intervention;
2) implementing this
intervention; 3) evaluating
the impact of the
intervention, and
implementing what was
learned from evaluation.
Successes:
Significant increases in
identifying high-risk
patients, repositioning bed
and chair bound patients,
use of nutritional consults,
and staging of acquired
Stage II or greater PUs.
Barriers:
View that PU prevention
was a nursing issue.
Addressing Barriers:
Re-educating various
disciplines about their role
in PU prevention
Sustainability:
Hospitals found that the
most sustainable
interventions were
institutionalized, such as
change in pressurerelieving mattress.
Interventions that depended
more on sufficient staff
such as turn schedules
were less sustainable.
Hospital
acquired
incidence
rate:
No
statistically
significant
change from
baseline to
follow-up
(20.6 to 20.8,
p = 0.90)
Hospital
acquired
prevalence
rates: NS
Decrease in
median length
of hospital
stay (8.0 days
to 7.0 days,
p = 0.05)
D-133
Influence of
Contexts on
Outcomes
Focusing
pressure ulcer
prediction and
prevention
programs on
the nursing
staff is limited
insofar as
effective
pressure ulcer
prevention
requires a
multidisciplina
ry effort. The
PDSA model
assists
hospitals in
working in
multidisciplina
ry teams and
places the
onus for
improvement
on the team
rather than on
a particular
discipline.
Author/Year
Stier et al.
15
2004
Description of PSP
Study
Design
Theory or
Logic
Model
NS
Description
of
Organization
A large notfor-profit
health care
system in the
U.S. with over
5,600 beds
and more
than 33,000
employees.
The system is
composed of
18 hospitals,
4 skilled
nursing
facilities,
1 certified
home health
agency, and
2 hospice
agencies.
The focus of
current study
is on
implementing
skin care
initiatives in
acute care
hospitals.
To implement a
Time
system wide
series
multidisciplinary
skin care initiative to
standardize care to
reduce the
incidence and
severity of PUs
Target safety
problem: PU
Key elements:
The project involved
standardization of
risk assessment
methods,
delineating
timeframes for
patient assessment,
and reassessment,
developing a
uniform skin care
formulary,
negotiating a
system-wide
contract for
therapeutic support
surfaces, and
providing staff
education.
a
Not included in the Soban 2009 review16
CMS:
Centers for Medicare and Medicaid Services
CNS:
Clinical nurse specialist
CPG:
Clinical practice guidelines
CWOCN:
Certified Wound, Ostomy, and Continence Nurse
DMAIC: Defining, measuring, analyzing, improving, initiating change
EBP:
Evidence-based practice
EMR:
Electronic medical record
HAPU: Hospital-acquired pressure ulcer
ICU:
Intensive care unit
IHI:
Institute for Healthcare Improvement
LPN:
Licensed practical nurse
Contexts
Implementation Details
Outcomes:
Benefits
External: NS
Organizational
Characteristics: NS
Teamwork, Leadership,
Culture: NS
Implementation tools:
Implemented Braden scale
to standardize risk
assessment
Hospital
acquired
incidence
rate:
Pre: 2.2%
Post: 1.3%
D-134
Influence of
Contexts on
Outcomes
A
standardized
approach to
patient
assessment/
reassessment
through the
use of
evidencebased
guidelines
and
educational
programs led
to a common
understanding
of pressure
ulcer
management,
improved
communicatio
n among care
providers, and
sustained
improvement
in patient
outcomes.
MICU:
NDNQI:
NR:
NS:
PDSA:
POA:
PSP:
PU:
PUPPI:
QIO:
RN:
SKIN:
SOS:
WOC:
WOCN:
D-135
Table 2, Chapter 21. Multi-component pressure ulcer prevention initiatives conducted in long-term care settings in the United States
Author/
Year
Description
of PSP
Study
Design
Horn et al.
17
2010
Real-Time
Time series
Program
(renamed
On-Time Quality
Improvement for
Long Term Care
[On-Time])
Target safety
problem: PU
Key elements:
CNA assist in
redesigning
documentation to
include core data
elements to help
identify high-risk
patients;
facilitators
provide feedback
on weekly clinical
decision-making
reports; staff
educated on QI
methods and
smooth
integration of
these CNA
documentation
and clinical
reports into dayto-day flow
Theory or Logic
Model
Description of
Organization
Based on best
practices from
AHRQ and AMDA
guidelines, and
findings from the
National Pressure
Ulcer Long-term
Care Study
(NPULS)
11 not-for-profit
facilities in 7 states
Bed size:
44432 beds
13 highest-risk
units per facility
participated
Contexts
Influence of
Contexts on
Outcomes
External: AHRQ
Length: 9 months
CMS HRPrU QM
Facility B which
funded
Process:
prior to
had the highest
Organizational
implementation
Facilitators work
reduction in PU
Characteristics: NS
(k = 7): 13.0%
with a
(-82.4%) was the
Teamwork,
CMS HRPrU QM
multidisciplinary
only facility that:
Leadership,
12 months after
team from each
had 100%
implementation
Culture: NS
facility.
participation of
Implementation tools: Redesigned CNA
(k = 7): 8.7%
residents
HRPrU QM %
documentation
CNA
(n = 75)
incorporating core change
documentation
Facility B was 1 of
(5 facilities using
data elements
processes and
3 facilities who
2 reports)
including nutrition
timely reports to
incorporated all 4
-25% to -82.4%
identify patients at and incontinence
clinical reports for
High Risk PrU QM care planning.
variables.
risk
% change
CNAs coached to
Two facilities with
A project leader
(2 facilities using 1 the lowest reduction
improve
(e.g., DON) and
report)
documentation.
in PUs did not
ongoing team
Sites fax scannable +8.3%, +14.3%
involve a
identified
Average number of multidisciplinary
forms to project
Educate staff on
in-house acquired team.
office.
QI methods and
PU (all stages) per
Clinical reports
use of
facility prereturned
within
documentation
implementation
forms and reports 24 hours and
vs postdisplayed.
Feedback includes implementation:
inconsistencies and 12.1 to 4.6
(62% reduction)
completeness of
Average number of
CNA
documentation per CNA
documentation
unit/unit over
forms reduced by
time/shift.
After reviewing with 53.2%.
CNAs, need for
additional
education noted.
Conference calls
(bi-weekly), allfacility meetings
(every 6 months)
and on-site
D-136
Implementation
Details
Outcomes:
Benefits
Author/
Year
Description
of PSP
Study
Design
Theory or Logic
Model
Description of
Organization
Contexts
Implementation
Details
meetings were
scheduled with
facilitators, project
leaders and
frontline staff.
Successes:
CNAs widely
accept revised
forms and increase
productivity.
Documentation
completeness rates
increase from
80%90% to
mid-90%.
Barriers:
EMR system used
by 1 facility could
only export data
elements and
create 1 report
Issues raised with
preparing the CNA
documentation
forms needing
the residents
study ID number
and
faxing forms for
report generation
Staff turnover
especially by DON
slowed project
momentum.
Addressing
Barriers:
Add new CNA
documentation
process into
orientation
programs
D-137
Outcomes:
Benefits
Influence of
Contexts on
Outcomes
Author/
Year
Rantz et al.
18
2010
Description
of PSP
Bedside EMR
(OEMR, Irvine,
CA) and
statewide on-site
clinical
consultation
services
(QIPMO
Quality
Improvement
Program for
Missouri)
Target safety
Study
Design
RCT 4-group
comparison
Group 1:
EMR plus
consult
Group 2:
EMR
Group 3:
Consult
Group 4:
Control
Theory or Logic
Model
NS
Description of
Organization
18 facilities in 3
U.S. states
Group 1:
4 facilities
Bed size range,
98240,
total 668
Group 2:
4 facilities
Bed size range,
105218,
total 635
Group 3:
Contexts
Implementation
Details
Outcomes:
Benefits
Phase in use of
documentation.
Develop a strong
multidisciplinary
team to lead
improvement
efforts and not
rely on one
project leader.
Sustainability:
HIT needed to
capture CNA
documentation and
generate reports.
Managing the
manual data
collection, faxing
forms to the project
office and creating
clinical reports for
distribution back to
the facilities on a
weekly basis could
not be maintained
over the long term
for many facilities.
Program expanded
throughout the U.S.
External: CMS funds Length: 2 years
Relative
OEMR hardware,
improvement in
Process:
software and
high risk pressure
ongoing tech support Project
sores (negative
Organizational
scores indicate
coordinator
Characteristics:
improvement)
works with
12 months
Mix of for-profit,
OEMR staff
not-for-profit, and
Staff works with Group 1: -53%
Group 2: -12%
governmental
QIPMO nurses
facilities
at least monthly Group 3: -5%
Teamwork,
Group 4: +435%
QIPMO nurses
Leadership, Culture:
encourage staff 24 months
NS
Group 1: -3%
to focus on
D-138
Influence of
Contexts on
Outcomes
Author/
Year
Description
of PSP
problem:
Comprehensive
Key elements:
Mandatory
OEMR training,
QIPMO nurses
Study
Design
Theory or Logic
Model
Description of
Organization
Contexts
5 facilities
Bed size range,
90123,
total 543
Group 4:
5 facilities
Bed size range,
120310,
total 890
Group 1, 3, 4 from
Missouri
Group 2:
Other States
Implementation tools:
clinical care and
improving care
Project
systems to be
coordinator
enabled by
assigned at
OEMR
intervention
Successes:
facility
Group 1, 2 and 3
QIPMO nurses
showed
improvements at
12 months;
Group 1 and 2
sustained at
24 months
Barriers: NS
Addressing
Barriers: NS
Sustainability:
Improvement
sustained during
Year 2 for Group 1
and 2
D-139
Implementation
Details
Outcomes:
Benefits
Group 2: -8%
Group 3: +59%
Group 4: +105%
Influence of
Contexts on
Outcomes
Cost increases
were most likely
attributable to the
cost of technology,
maintaining and
supporting the
technology, and ongoing staff training
to use the EMR and
not increase direct
care staffing or
turnover.
Author/
Year
Description
of PSP
Study
Design
Milne et al.
19
2009
LTACH care
Time series
process
improvement
Target safety
problem: PU
Key elements:
Nursing
association
consults; team
training; improve
assessment and
documentation
methods; EMR
revised; formal
and informal staff
education;
wound care
product reviews
Theory or Logic
Model
Description of
Organization
Contexts
Long-term acute
care facility in CT
Bed size, 108
External: NS
Organizational
Characteristics:
Above average PU
prevalence
Teamwork,
Leadership, Culture:
Faulty EMR
Inconsistent use
of EMR by
clinicians
Deficient risk
assessment
documentation
Implementation tools:
Training by
nursing
association
APN appointed
in-house leader
APN and nursing
supervisor
become WCC
Team clinicians
trained in
prevalence data
collection
EMR revised;
PUSH tool added
Staff educated via
formal clinical
rounds, interactive
sessions and oneon-one bedside
sessions
Immediate
feedback given on
training
D-140
Implementation
Details
Outcomes:
Benefits
Influence of
Contexts on
Outcomes
Length: 13 months Mean facilityData on PU
facility wide
acquired PU
prevention
Process:
prevalence:
implementation in a
LTACH is spare.
Roles for new
Pre: 41%
Two LTACH units
skin team
Post: 4.2%
members
Pulmonary-focused however were able
to reduce PUs to
defined
unit:
<3% due to
Team meets
Pre: 25%
increased
weekly to review Post: <3%
diligence by the
failure modes
SCI/trauma unit:
team.
and develop new Pre: 33.8%
The authors noted
care processes
Post: 2.9%
an increased
Revamping of
collaboration among
policies and
disciplines with
procedures after
regard to wound
review of CPGs
prevention and
Wound care
treatment as well as
product reviews
a tendency for early
Successes:
intervention when
PU reduced to <3%
wounds are newly
on two units due to
discovered.
increased
monitoring of
modified nasal
cannula (pulmonary
unit) and increased
attentiveness to
heel offloading,
support surfaces
and proper
positioning
(SCI/trauma unit);
of the 396 charts
reviewed, <1% had
missing data;
staging and wound
etiology were
consistently
determined by
wound team in
greater than 90% of
Author/
Year
Description
of PSP
Study
Design
Theory or Logic
Model
Description of
Organization
Contexts
Implementation
Details
cases (based on a
review of 45 patient
charts)
Barriers:
Rates climbed once
strict monitoring
was leveled off
Addressing
Barriers:
Increase in unit
presence, chart
monitoring,
feedback to staff,
and biweekly
prevalence rounds
Sustainability:
CWCN
certification of 2
team members
provide in-house
expertise
Monthly review
of documentation
and PU
prevention
interventions
Early
intervention
D-141
Outcomes:
Benefits
Influence of
Contexts on
Outcomes
Author/
Year
Description
of PSP
Study
Design
Tippet A.
20
2009
Physician
Time series
consultant leads
deficient nursing
home to zero
facility-acquired
PUs
Target safety
problem: PU
Key elements:
Physician wound
consultant,
multidisciplinary
team, education,
weekly informal
feedback, wound
care protocols
based on AHRQ
CPG, wound
coordinator
sustains program
Theory or Logic
Model
Description of
Organization
Contexts
Based on AHRQ
CPG
Midwest skilled
facility
Bed size: 151
External: G-level
citation (actual harm
deficiency) and state
survey deficiencies
Organizational
Characteristics: NS
Teamwork,
Leadership, Culture:
NS
Implementation tools:
Physician
consultant
Multi-disciplinary
team
Braden Scale,
AHRQ CPG
Incentive
programs
Informal feedback
Simplified wound
care formulary
Equipment
evaluation (Delphi
process used to
evaluate
products)
D-142
Implementation
Details
Outcomes:
Benefits
Influence of
Contexts on
Outcomes
Length: 6 years
Average preEstimated cost
Process:
initiative incidence: savings per PU/per
5.19%
month:
Physician
Average post$1,617
consultant
initiative incidence: Monthly savings:
educates staff
0.73%
$10,187
and conducts
Yearly savings:
yearly follow-up (p<0.0001)
4 year post>$122,000
training (all
initiative incidence:
mandatory)
0.06%
Team forms
goals and meets (p<0.0001)
weekly
Select members
conduct wound
rounds
Follow-up
training through
in services, and
yearly follow-up
Nursing
supervisors
conduct one-onone with staff
and weekly
informal
feedback
Preventive care
plans created
Protocols
discussed in
classes, become
part of routine
shift reporting
and charting
All nursing staff
made
accountable for
care and
reporting
Successes:
Goal of zero facility
Author/
Year
Description
of PSP
Study
Design
Theory or Logic
Model
Description of
Organization
Contexts
Implementation
Details
acquired ulcers
reached after
6 months
Facility citation free
Accolades from
surveyors for
wound program
Judged
competitions
between floors
promote teamwork
and buy-in
Barriers: NS
Addressing
Barriers: NS
Sustainability:
Wound care
coordinator position
established to
supervise, train,
provide clinical
support and track
wounds.
Permanent decline
after 6 months
through study end
D-143
Outcomes:
Benefits
Influence of
Contexts on
Outcomes
Author/
Year
Description
of PSP
Study
Design
Theory or Logic
Model
Rosen et al.
21
2006
Ability,
Incentives, and
Management
feedback (AIM
system)
Target safety
problem: PU
Key elements:
Staff ability
enhancement
(skin care
training, use of
penlights and
TAP card),
real-time
management
feedback,
financial
incentives
Longitudinal
NS
time series
study; four
12-week
periods
(baseline
assessment,
intervention,
and two postintervention
periods)
Description of
Organization
Contexts
Implementation
Details
Outcomes:
Benefits
Not-for-profit
nursing home in
U.S.
Bed size, 136
External: AHRQ
funded
Organizational
Characteristics:
Received multiple
Department of
Health citations due
to persistently high
PU rates
Teamwork,
Leadership, Culture:
Lack of management
to oversee earlier
processes
Implementation
tools:
Research team
contacts
administrators
responsible for
overseeing
implementation.
Mandatory
skin care
training (a
40-minute
computerbased,
interactive-video
education
program).
Penlights
Caregivers wear
plastic TAP
(turn and
position card) to
remind all
hospital
personnel the
direction
residents should
Length: 48 weeks
Process:
One skin care
nurse assessed
patients upon
admission or
notification by staff
of any skin
changes.
During the postintervention
periods, no weekly
reports were
provided to the
administrators, no
established targets
or goals were
established, and
there were no
financial incentives
offered to staff.
Only 3 of 29 new
hires completed
training.
Sustainability:
The intervention
was not sustained
over the two postintervention periods
however Rosen et
al. indicated that a
highly motivated
administrator could
have maintained
the 3 program
components.
Significant
reduction in
emergence of
stage 14 PUs
Pre-intervention:
28.3%
Intervention:
9.3%
(z[I] = 2.64,
p<0.001)
Total ulcers
Stage 1 and
beyond
Pre-intervention
(n = 134):
38% (28.3)
Intervention
(n = 107):
10% (9.3)
Post-Intervention I:
19% (17.7)
Post-Intervention II:
19% (17.7)
Total ulcers
Stage 2 and
beyond
Pre-intervention:
31% (23.1)
Intervention:
10% (9.3)
Post-Intervention I:
15% (14.0)
Post-Intervention II:
17% (15.9)
D-144
Influence of
Contexts on
Outcomes
With a mean cost of
$2700 of treating a
single stage II PU,
[26] reducing the
incidence of these
ulcers by
approximately 15
over 12 weeks
yields a potential
savings of more
than $40,000 while
distributing less
than $10,000 as
incentives. This
does not take into
consideration the
added savings in
fewer personal
injury lawsuits.
The primary
management
feedback tool was
adherence to the
mandated training
(not emergence of a
new PU). Additional
real-time feedback
was provided to
staff in the form of a
visual
thermometer of
PU occurrences
each week. All a
nonfinancial
incentive, it served
as a supplementary
motivating factor as
the incidence of
PUs was visually
perceived as
declining.
Author/
Year
Description
of PSP
Study
Design
Theory or Logic
Model
Description of
Organization
Contexts
D-145
be facing every
2 hours.
Administrators
receive a
weekly report of
staff that had
completed
training.
A graphic
thermometer
of PU incidence
was also
updated weekly
and displayed in
the staff lounge.
Each staff
member
received $75
if the PU
incidence was
below target
goal (incidence
<3%) set by
administration.
Staff
reprimanded for
non-completion.
Staff terminated
for not
completing
training during
extension
period.
Implementation
Details
Outcomes:
Benefits
Influence of
Contexts on
Outcomes
Author/
Year
Description
of PSP
Study
Design
Abel et al.
22
2005
Process of care
Pre-post
system changes
in collaboration
with a state QIO
Target safety
problem:
PUs
Key elements:
Collaborative
with a state QIO,
intervention tool
kit, nurses aid
and licensed staff
training
Theory or Logic
Model
Description of
Organization
Contexts
Implementation
Details
Outcomes:
Benefits
NS
20 facilities in
Texas
Average residents:
100
Average Medicare
beds: 15
External: Identified
from 143 Medicarecertified skilled
nursing facilities as
having high rates of
PUs and a high
volume of residents
receiving preventive
care
Organizational
Characteristics:
Selected due to
accessibility to state
QIO (Texas Medical
Foundation [TMF])
Teamwork,
Leadership, Culture:
NS
Implementation tools:
TMF provides
tools
o Nurses Station
Reference
Cards
o Pocket
Assessment
Card
o Mobility
Program
o Fax
Communication
Form
o Care Planning
Tool
o Resident
Patient and
Family
Education
Brochure
Tool kit components
based on information
Length: 2 years
Process:
Monthly onsite
visits by TMF
Tools modified
Periodic
progress
assessment
Successes:
Performance
significantly
improved on 8 of
12 QIs
Management
maintains
autonomy which
promoted
continued
commitment and
a sense of
ownership
Barriers:
Staff resistance
Staff turnover
and variation in
new staff
orientation often
contributed to
clinical or
operational
practices that
were
inconsistent with
their protocol
requirements.
Incomplete risk
assessments
Monitoring
systems not
appropriately
used
Incidence rate:
Pre: 13.6%
Post: 10.0%
Significant
improvements in
8 QIs (baseline vs.
re-measurement):
Proportion of
residents with
appropriate risk
assessment
completed within
2 days of
admission
(2.2% vs.
15.3%;
p<0.0001)
Proportion of
high-risk
residents with
appropriate care
plan for ALL
selected triggers
for high-risk
residents
(10.1% vs.
21.8%;
p<0.0001)
Proportion of
high-risk
residents whose
care reflects the
triggered care
plan
interventions
(2.0% vs. 9.8%;
p<0.0001)
Proportion of
residents with
PUs that receive
weekly skin
D-146
Influence of
Contexts on
Outcomes
Although there are
areas for
improvement, the
implementation of
process of care
system changes by
NHs in a
collaborative
relationship with a
QIO may yield
improvements in
measures of patient
outcomes (e.g., PU
incidence).
Abel et al. also
indicated that the
10 facilities with the
highest [QI] scores
at re-measurement
showed a trend
toward a lower [PU]
incidence rates than
the 10 facilities with
the lowest [QI]
indicator scores at
re-measurement
(S = 125.5,
p = 0.07).
Facilities with the
highest QI scores
versus facilities with
the lowest QI
scores (baseline vs.
re-measurement;
PU incidence rate,
%):
High scoring group:
12.3% vs. 7.7%
Low scoring group:
14.8% vs. 12.2%
Facilities with the
Author/
Year
Description
of PSP
Study
Design
Theory or Logic
Model
Description of
Organization
Contexts
Implementation
Details
Outcomes:
Benefits
Documented risk
factors not acted
upon
Addressing
Barriers: Monthly
D-147
Influence of
Contexts on
Outcomes
assessments
greatest
(12.6% vs.
improvement versus
32.8%;
facilities with the
p<0.0001)
least improvement
in QI scores
Proportion of
facility-acquired (baseline vs. reand community- measurement; PU
incidence rate, %):
acquired PUs
with appropriate High scoring group:
ulcer description 13.1% vs. 7.1%
within 24 hours Low scoring group:
14.0% vs. 12.8%
of ulcer
recognition
(53.5% vs.
68.9%;
p = 0.035)
Proportion of
residents with
PUs and mobility
issues using a
pressure relief
mattress/overlay
(50.7 vs. 76.7;
p<0.0001)
Proportion of
residents
identified as
high risk (as per
MDS) using a
pressure relief
mattress/overlay
(33.0% vs.
53.4%;
p = 0.003)
Proportion of
residents whose
treatment orders
and care plan
interventions for
PUs reflect
facility wound
Author/
Year
Rantz et al.
23
2001
Description
of PSP
Statewide
implementation
of Show-Me QI
report
Target safety
problem:
Comprehensive
Key elements:
Workshops,
Minimum Data
Set (MDS)
Quality Indicator
(QI) feedback
reports, clinical
consultation
Study
Design
Theory or Logic
Model
RCT
NS
Group 1:
Workshop
plus feedback
reports
Group 2:
Workshop,
feedback
reports and
clinical
consult
Group 3:
Control
Description of
Organization
87 nursing homes
in Missouri
Bed size:
160:
10
61120: 52
120+:
25
Contexts
External: NS
Organizational
Characteristics:
Adequate experience
with transmitting
MDS data
electronically
Teamwork,
Leadership, Culture:
NS
Implementation tools:
Educational
workshop
RAI manual
RAPs
CPG (AHRQs)
Comparative
feedback ShowMe QI report
(quarterly)
GCNS consult
D-148
Implementation
Details
Length: 1 year
Process:
Core group
receives ShowMe QI report in
workshop;
subsequent
quarterly reports
sent to
administrator
and DON
GCNS help
interpret report,
assess resident
problems, and
document care
15 facilities
(Group 2) had
1 on-site visits
and GCNS calls
18 facilities
(Group 2) had
only 1 call and
limited GCNS
calls
13 quality
indicator
outcome
measures were
evaluated
Successes:
Reduction in
pressure ulcers
(overall and lowrisk) for residents in
facilities using
GCNS
Barriers:
Outcomes:
Benefits
care protocol
(1.3% vs. 4.9%;
p = 0.0505)
Secondary
regression
analysis:
MDS QI 29
Pressure Ulcers
(overall):
Case mix: 0.156
Time Pre-Post:
0.240
Intervention: 0.026
Group X Time:
0.085 (p0.10)
MDS QI 29lr
Pressure ulcers
low risk:
Case mix: 0.417
Time Pre-Post:
0.037
Intervention: 0.064
Group X Time:
0.057 (p0.10)
Influence of
Contexts on
Outcomes
A subset of Group
2 nursing homes
that were intensely
involved with the
intervention
showed
improvement in
MDS QI scores for
five outcome
measures including
MDS QI 29
(pressure ulcers).
Nursing homes
that did have
continuous quality
improvement
systems in place
were often part of
larger health care
systems that have
ongoing support
from a quality
improvement
expert.
Author/
Year
Description
of PSP
Study
Design
Theory or Logic
Model
Description of
Organization
Contexts
Implementation
Details
Short staff
Staff turnover
especially nurse
RAI coordinator
Taking care
themselves
Cancelled site
visit at last
minute
Additional time
needed to
correct
inaccurate MDS
assessments
Teams mired in
the MDS
assessment
process and
coding issues
Difficulty
convincing staff
to use
continuous QI
principles
Addressing
Barriers:
Stronger
incentives to use
GCNS
GCNS more
local
More flexible site
visit times
Extend time to
implement
change
Use teams to
address
problems
Post accomplishments
D-149
Outcomes:
Benefits
Influence of
Contexts on
Outcomes
Author/
Year
Ryden et al.
24
2000
Description
of PSP
Protocol
implementation
by APNs
Target safety
problem:
Comprehensive
Key elements:
APNs assist staff
to implement
care plan;
APNs provide
direct care to
residents
Study
Design
Theory or Logic
Model
Controlled
Havelocks (1974)
before-andmodel of effective
after
research utilization
APN
treatment
(2 facilities)
vs. usual care
(1 facility)
Description of
Organization
3 privately-owned
facilities located in
suburban
MinneapolisSt. Paul area;
certified for
Medicare
Contexts
Implementation
Details
Multiple nurses
responsible for
RAI process
Use of quality
manager on staff
to support care
delivery
improvements
Leadership buy
in to QI
Sustainability: NS
External: NS
Length: 6 months
Organizational
Process:
Characteristics:
RAs assess
APNs work with head
risk/collect data
nurse who works
2 APNs reassess
with physician or
risk, analyze data
GNP
(10 hrs/week per
Teamwork,
facility)
Leadership, Culture: APNs meet with
NS
residents
Implementation tools:
15-30 min/wk
AHRQ CPG
Successes:
Staff education
6 months of APN
Work with nursing treatment
assistants
significantly
APNs participate improved 3 of 4
clinical problems
in conferences
compare to usual
and wound care
care
rounds
Barriers:
High turnover of
unlicensed staff
Addressing
Barriers: NS
Sustainability:
A wound care
committee was
established at
1 facility.
D-150
Outcomes:
Benefits
Influence of
Contexts on
Outcomes
APN Treatment
(n = 86)
Pre: 19.8
Post: 3.5
2
x = 3.01(1),
p = 0.04, one-tailed
Usual Care
(n = 111)
Pre: 17.3
Post: 10.0
The relatively
short time (10 hr
per week in each
nursing home) and
the high turnover
rates of unlicensed
staff (range of
11%-45%) reduced
opportunities for
each APN to
establish
relationships with
staff.
D-151
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
D-152
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
D-153
Evidence Tables for Chapter 22. Inpatient Intensive Glucose Control Strategies To
Reduce Death and Infection (NEW)
Table 1, Chapter 22. Large trials (n > 500) evaluating the health outcome effects of IIT
Patient population
Single or multi-center
Country
Implementation/
Context
Diabetes
mellitus (%)
Glucose
target, T v
C (mg/dL)
SICU
Single center
10
Belgium
Insulin protocol
was developed
and use overseen
by study
investigators.
13
80-110
v 180-200
Neurosurgical ICU
Single center
58
Italy
Efforts made to
limit nursing
turnover. New
nursing staff
worked with
experienced staff.
Study conducted
in a hospital that
had already
conducted similar
IIT study in SICU
patients. Authors
note the nurse:bed
ratio of 2.5 was
not changed for
study.
NR
80-110 v
180-200
16
80-110 v
180-200
MICU
Single center
59
Belgium
111 v 153
(p<0.001)
D-154
ICU mortality:
24.2 v 26.8%
(p=0.31)
Hospital
mortality: 37.3
v 40.0%
(p=0.33)
RR 0.93; 95%
CI 0.81-1.08
90d mortality:
35.9 v 37.7%
(p=0.53)
Other reported
outcomes*
TvC
Quality
Renal
replacement
4.8 v 8.2%
(p=0.007)
Sepsis
4.2 v 7.8%
(p=0.0003)
Fair
Sepsis 2.9 v
3.3% (p=NS)
Long-term
disability:
40.2 v 41.1%
(p=0.98)
Infection 0.7 vs
0.8% (p=NS)
Renal
replacement
20.8 v 22.7%
(p=0.50)
Fair
MICU
Single center
59
Belgium
Patient population
Single or multi-center
Country
Implementation/
Context
Diabetes
mellitus (%)
Glucose
target, T v
C (mg/dL)
MICU
Multicenter
20
Germany
No details
provided
30
80-110
v 180-200
MICU/SICU
Multicenter
60
Europe
Characteristics
from each study
site were reported.
Median nurse:bed
ratio was 2. ICUs
ranged widely in
size, patient
volume, and
number of
glucometers per
ICU.
24/7 ICU coverage
by intensivists.
Protocol designed
by
multidisciplinary
team at study site.
Physicians and
nurses attended
training sessions
before and during
study.
Three month staff
training period
before study.
17 T, 22 C
(p=0.031)
80-110
v 140-180
32 T,
48 C
(p<0.001)
80-110
v 180-200
115 v 171
(p<0.001)
13 T,
12 C
(p=NS)
80-110 v
180-200
120 v 149
(p,0.001)
MICU/SICU
Single center
18
Saudi Arabia
MICU/SICU
Single center
53
Colombia
D-155
Other reported
outcomes*
TvC
Quality
Renal
replacement
27.5 v 22.5%
(p=0.001)
MICU
Multicenter
20
Germany
Renal
replacement
(patient days)
519 v 523
(p=0.75)
Fair
Renal
replacement
11.7 v 12.1%
(p=0.89)
Sepsis 36.9 v
40.9% (p=0.35)
Fair
Infection 27.2 v
33.2% (p=NS)
Renal
replacement
10.8 v 13%
(p=0.45)
Fair
Patient population
Single or multi-center
Country
Implementation/
Context
MICU/SICU
Pre-trial pilot
54
Multicenter International
studies carried out
to test/improve
insulin protocol.
Final
computerized
insulin protocol
algorithm
accessible to
study sites
through a central
Web site. No clear
explicit training
prior to study.
Acute MI
No details
Multicenter CCU
provided
61
Sweden
Acute MI
62
Multicenter Europe
No details
provided
Diabetes
mellitus (%)
Glucose
target, T v
C (mg/dL)
20
80-108 v
<180
39
126-198 v
NR
77
established
DM; 23 new
DM
of < 1y
24 hours:
T: 172.8
(59.4)
C: 210.6
(73.8)
p < .001
group 1 and
24 hours:
2: 126-180
group 1:
group 3:
163.8 (54.0),
NR
group 2:
163.8 (50.4),
group 3:
180.0 (64.8)
p = .0001
D-156
Other reported
outcomes*
TvC
Renal
replacement
15.4 v 14.5%
(p=0.34)
Sepsis
12.8 v 12.4%
(p=0.57)
Quality
Fair
Fair
Poor
Patient population
Single or multi-center
Country
Implementation/
Context
Diabetes
mellitus (%)
Glucose
target, T v
C (mg/dL)
D-157
References
1.
2.
3.
4.
5.
6.
7.
8.
D-158
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
D-159
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
47.
48.
49.
50.
51.
52.
53.
54.
55.
42.
43.
44.
45.
46.
D-160
56.
57.
58.
59.
D-161
60.
61.
62.
63.
Number of trials
Sample size
AMSTAR criteria
Bicarbonate
N-acetylcysteine and
bicarbonate
14
2290
10
Evidence of
benefit for
intervention
N
10
1594
12/2009
Iso-osmolar contrast
36
7166
9/2004
N-acetylcysteine
22
2746
11
11
N
N
8/2007
Iso-osmolar contrast
25
3270
11
4/2008
10/2007
2/2009
Not stated
11/2008
11/2006
9/2008
12/2008
1/2008
11/2008
Bicarbonate
Bicarbonate
Bicarbonate
Bicarbonate
Bicarbonate
N-acetycysteine
Bicarbonate
Bicarbonate
Bicarbonate
Iso-osmolar contrast
Renal replacement
therapy
Bicarbonate
N-acetylcysteine
Statins
Bicarbonate
4
7
18
9
17
26
7
17
12
16
573
1307
3055
2043
2448
3352
1734
2633
1854
2763
8
9
11
7
9
9
6
11
8
10
N
Y
N
N
N
Y
Y
Y
Y
Y
N
751
10
16
12
23
1090
1677
1194
3563
11
9
8
11
Y
Y
N
N
Literature
search end date
11/2008
Brown, 2009
2,3
2/2009
From, 2010
Gonzales, 2007
Heinrich, 2009
6
Ho, 2008
7
Hogan, 2008
8
Hoste, 2009
9
Joannidis, 2008
10
Kanbay, 2009
11
Kelly, 2008
12
Kunadian, 2010
13
Meier, 2009
14
Navaneethan, 2009
15
Reed, 2009
Song, 2010
16
17
Trivedi, 2009
18
Trivedi, 2010
19
Zhang, 2011
20
Zoungas, 2009
6/2010
10/2008
1/2008
7/2010
12/2008
Intervention evaluated
D-162
References
1.
2.
3.
4.
5.
6.
7.
8.
D-163
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
D-164
Anwar ul,
1
2010
Description of
PSP
Contexts
Implementation
Details
Education
sessions with
quarterly
reinforcement
Outcomes:
Benefits
Mortality : ICU
mortality of
patients
admitted to ICU
from floor (total
sample 77)
Results:50% to
15%
Statistics:
p=0.001 OR
0.18 (0.09-0.35)
Cardiac arrest:
Results: 5.2 to
2.7/
1000 admits
Statistics:
p=0.004
OR=.52 (0.122.26)
D-165
Outcomes: Influence
Harms of Contexts
on
Outcomes
Comments
Author,
year
Bader,
2
2009
Description of
PSP
D-166
Author,
year
Benson,
3
2008
Description of
PSP
Contexts
Implementation
Details
Outcomes:
Benefits
Credentialing,
information and
education
interventions
(email, newsletter
articles, rounding,
informational
sessions at
meetings), clinical
practice protocols
developed
Mortality :
average
mortality per
month
Results:9%
decrease (no
actual rates or
stats reported)
Statistics: NR
Cardiac arrest:
58.7% reduction
in codes per
1000
admissions
Results: 9.41
vs. 3.89
Statistics p =
.0065
National
Database of
Nursing Quality
Indicators
(NDNQI)
Failure to
Rescue rate
Results:19.5%
reduction (no
actual rates or
stats reported)
Statistics: NR
D-167
Outcomes: Influence
Harms of Contexts
on
Outcomes
Comments
Author,
year
Description of
PSP
Contexts
Implementation
Details
Outcomes:
Benefits
D-168
Outcomes: Influence
Harms of Contexts
on
Outcomes
Comments
Author,
year
Chan,
5
2008
Description of
PSP
Contexts
Implementation
Details
Outcomes:
Benefits
D-169
Outcomes: Influence
Harms of Contexts
on
Outcomes
Comments
Chose as a primary
outcome total hospital
code rate (including ICU
codes) and found no
benefit. ICU patients are
not part of RRS exposure
group. Their non-ICU
(general ward) codes did
drop significantly.
Author,
year
Gerdik,
6
2010
Description of
PSP
Contexts
Implementation
Details
Outcomes:
Benefits
Pilot program
Mortality : total
followed by
Results: 32.5
campus wide
vs.
implementation 8 31.0/1000
months later.
admits
Worked with UHC Statistics: ns
collaborative in
developing
Cardiac arrest:
implementation. Results: 25.2
Secured
vs. 17.4/
stakeholders, then month
added patient and Statistics: none
family activation given
ICU
readmisssion
Results: no data
given
Statistics: ns
change
D-170
Outcomes: Influence
Harms of Contexts
on
Outcomes
Comments
Author,
year
Hanson,
7
2009
Description of
PSP
Contexts
Implementation
Details
Outcomes:
Benefits
Criteria
Mortality : ward
development,
(not total) but
Collaborative
included those
participation (IHI), with DNR (i.e.
planning,
expected and
education, hospital unexpected)
wide
Results: 1.5 vs.
implementation
0.45/
1000 admits
Statistics: RR=
0.30 (0-1.04) p=
0.07
Cardiac arrest:
ward
Results: 1.27
vs. 0.45/
1000 admits
Statistics RR=
0.35 (0-1.24)
p=0.126
time between
codes
Results: 2512 to
9418 patient
days
Statistics: not
given
Total hospital
mortality
Results: 9.64
vs. 7.31/
1000 admits
Statistics:
RR=0.076 (01.03) p= 0.078
D-171
Outcomes: Influence
Harms of Contexts
on
Outcomes
Comments
Author,
year
Hatler,
8
2009
Konrad,
9
2010
Description of
PSP
MET consists of
Pre-post
ICU nurse and ICU
277717
physician
admissions
afferent and
efferent limbs,
education
NA
900 bed
teaching
hospital in
Sweden
Contexts
Implementation
Details
Outcomes:
Benefits
Cardiac arrest:
Results: 0.93
vs. 0.63/
1000 discharge
Statistics not
given, may be
ns
direct and online Mortality :
intranet education, adjusted total
pocket cards for Results: RR 0.9
alert criteria with Statistics:
an education
p=0.003
period during the
initial
Cardiac arrest:
implementation
Results: 1.12
vs. 0.83/
1000
admissions
Statistics
p=0.035
Outcomes: Influence
Harms of Contexts
on
Outcomes
Comments
Team structure,
alert criteria,
documentation
development,
education
180 day
mortality
Results: 37%
vs. 15.8%
Statistics: NR
LOS
Results: no
change
D-172
Adjusted
Only study to report longmortality
term mortality
was
significantly
decreased
in both
medical and
surgical
patients
Author,
year
Kotsakis,
10
2011
Description of
PSP
Implementation
Details
Outcomes:
Benefits
3 phases,
Mortality: total
development
hospital
1.education phase mortality: 10 vs.
2. pilot phase
9.6/1000 admits
when team only Statistics: NS
avail M-F during
Cardiac arrests:
day
Results: 1.9
3.Full 24/7
vs.1.8/ 1000
7d/week
implementation. admits
Statistics: NS
MET and Code
Blue Team were
the same group ICU mortality
Results: 0.3 vs.
(unified team)
0.1/ 1000
hospital admits
Statistics:
p=0.05
ICU
readmission
Results: NR
Statistics: NR
D-173
Outcomes: Influence
Harms of Contexts
on
Outcomes
Comments
Prospectively collected
after implementation
Author,
year
Laurens,
11
2011
Description of
PSP
Multi-component
MET consisted of
anesthesiologist,
medical house
officer and ICU/ED
nurse. responds to
any patient outside
ICU
describes the alert
criteria, education
process and RRS
process
Contexts
Implementation
Details
Outcomes:
Benefits
One month
Mortality:
education program unadjusted
prior to
hospital
introduction of the Results: 9.9 vs.
MET with ongoing 7.5/ 1000
admissions
education.
Formal training for Statistics:
MET team
RRR=24.2%
members and
p=0.003
index cards for
staff with alert
Cardiac arrests:
criteria
Results: 77 vs.
42/1000 admits
Statistics:
RRR=45.5%
p=0.0025
ICU admissions
Results: 22.4 to
17.6/ 1000
admissions
Statistics:
RRR=21.4%
p=0.003
D-174
Outcomes: Influence
Harms of Contexts
on
Outcomes
Comments
Author,
year
Lighthall,
12
2010
Description of
PSP
Contexts
Implementation
Details
Outcomes:
Benefits
D-175
Outcomes: Influence
Harms of Contexts
on
Outcomes
Comments
Author,
year
Rothberg,
13
2011
Description of
PSP
Implementation
Details
Outcomes:
Benefits
Outcomes: Influence
Comments
Harms of Contexts
on
Outcomes
Initial
Mortality :
Stratified
Codes called for medical
implementation on Overall hospital
analyses by crises declined for units
2 med floors then mortality
codes within outside critical care only;
critical care Rate of MET activation
spread to entire Results: 22
deaths/
vs. codes (18 calls/1000
hospital over 3
months; Education 1000
outside
admissions)
included meetings, admissions
critical care:
e-mails, and
across study
posters; anyone period
could
Statistics: NS
activate;75% calls
from med, 20%
Cardiac arrest:
from surgical
Cardiac arrests
did not change
significantly
Results: 7.3 to
4.2/1000
admissions
Statistics
p<0.0001
Rate of fatal
codes/1000
admissions
Results: Delta =
0.06 (no specific
pre
post rate
reported in text,
graphed in
figure 4 only)
Statistics: p =
.65
D-176
Author,
year
Sarani*,
15
2011
Description of
PSP
NA
Academic
hospital in US.
Size not given
Implementation
Details
Outcomes:
Benefits
Created MET as
part of MERIT
study, they were a
MET hospital in
that study
Mortality :
unexpected
Results: 0.58
vs. 0.30/1000
admits in last
time period
Statistics:
p<0.05
Cardiac arrest:
Results: 0.78
vs. 0.25/1000
admits in last
time period
Statistics
p<0.001
Limited, states
cardiac surgical
service did not
participate but
nothing beyond
that
Unanticipated
ICU admission
Results: 0.65
vs. 0.89/1000
admits in last
time period
Statistics: ns
Mortality:
hospital
mortality
Results:
Medical: 4.29
vs. 3.23%,
p<0.001;
Surgical: 1.21
vs. 1.11%
Statistics: ns
Cardiac arrest
Results: 4.07
vs. 2.32/1000
discharges
Statistics:
p<0.001
describes criteria
for RRS and the
structure
D-177
Outcomes: Influence
Harms of Contexts
on
Outcomes
Comments
Author,
year
Scott,
16
2009
Shah*,
17
2011
Description of
PSP
Pre-post
231,305
patient days,
61,389
admissions
NA
3 affiliated
academic
hospitals in
the US.
Contexts
Implementation
Details
Outcomes:
Benefits
1 month pilot
Cardiac arrest:
followed by house- Results: 7 vs. 2/
wide
1000 patient
implementation
days
Statistics:
unknown
Pre-intervention Mortality: Inperiod followed by hospital:
a 9 month roll-out Results: 2.4%
followed by full
vs. 2.06%,
intervention period 1.94%, 2.46% in
subsequent
postimplementation
period,
respectively
Statistics:
p=0.03,
0.01,and 0.83
respectively for
each postimplementation
period.
Cardiac arrests
Results: 0.83
vs. 0.98/
1000 final
period
Statistics: p=0.3
D-178
Outcomes: Influence
Harms of Contexts
on
Outcomes
Comments
Author,
year
Tibballs,
18
2009
Description of
PSP
Contexts
Implementation
Details
Outcomes:
Benefits
Included intensive
education, hiring
additional ICU
nurses
Mortality: total
in-hospital
Results; 4.38
vs. 2.87/
1000 admits
Statistics: RR=
0.65 (0.57-0.75)
p<0.0001
Mortality:
unexpected
general ward
Results: 0.12
vs. 0.04/
1000
Statistics:
RR=0.35 (0.130.92) p=0.03
Cardiac arrest:
unexpected
non-ICU
Results: 0.19
vs. 0.17/
1000 admits
Statistics
RR=0.91 (0.501.64) p=0.75
Cardiac arrest:
preventable
non-ICU
Results: 0.16
vs. 0.07/1000
admits
Statistics:
RR=0.45 (0.20.97) p=0.04
* one reviewer had indicated that the article did not apply but a subsequent reviewer included and data was available
D-179
Outcomes: Influence
Harms of Contexts
on
Outcomes
Comments
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
D-180
10.
11.
12.
13.
14.
15.
16.
17.
18.
Buist,* 2007
Calzavacca,
3
2010
Chen, 2010
Cretikos,
5
2007
Description of
RRS
offered in
Appendix A but
appendix not with
pdf. uses two
tiered mechanism
for calling for
assistance
Study Design
prospective
evaluation of
breaches of
PACE system
before and after
changes
before after
design
cohort
comparison
(early MET time
period and
another time
period several
years later)
Main Study
Description of Organization
objectives
to assess if new
750 bed tertiary university
strategies could
affiliated hospital
improve the time to
delivery of MET
components as
compared to previous
MET system
too assess impact of
change programs
(education program
for new interns, nurse
liaisons, and
development
programs for
hosuestaff) on
incidence of cardiac
arrest
Does maturation of a
RRS improve the
failure to rescue rate
(recognition of
deterioration) and the
associated outcomes
Implementation
Comments
Themes
centralized activation
quantitative
system, review of all
events, automatic
escalation to code team if
MET did not respond
within 30 min, institution
of nurse educator for
training and compliance
400 bed suburban teaching
nurse liaison, career
hospital (first one in the world development and
to have a true MET)
education/oreintation
To assess the
process components
of MET
implementation that
correlated with
utilization
D-181
change in delayed
activations (late
recognition),
unanticipated ICU
admission
institution of NFR (DNR)
orders
effect of teaching
hospital, metropolitan
hospital, patient location
and time of activation
knowledge of activation
Quantitative but only to
criteria, understanding of utilization rates not
MET purpose,
outcomes
perceptions of readiness
for change, overall
attitude to MET program
Author, year
Donaldson,
6
2009
Foraida,
7
2003 ,
DeVita,
8
2004
Genardi,
9
2008
Description of
Study Design
RRS
not known as it
multi-modal
involved multiple
(qualitative using
hospitals, probably interviews
varied
ICU registrar,
prospective
Anesthesia, ICU
nurse, resp
therapy; 8 defined
roles-Team leader
airway manager,
airway assistant,
procedure
physician, chest
compressions,
runs medication/
equipment chart,
recorder, bedside
nursing
not given
prospective
Main Study
objectives
Identify factors
associated with
successful
implementation,
develop plans to help
others replicate such
success, standardize
process measures,
evaluate impact
through nurse
perceptions.
to determine if
specific educational
and feedback
interventions would
increase MET
utilization
to revitalize their
existing RRT and
improve on code
reductions
Description of Organization
multiple (>500 hospitals,
nested within 9 multihospital
grantee organizations)
Implementation
Themes
Extra resources, rapid
transfer, communication
enhancement, one stop
shopping(single team
assessment), robust
early adopters vs. late or
poor functioning RRS
D-182
Comments
Very qualitative, did not
define successful RRSs by
any objective criteria
Quantitative data on
utilization and incidence of
cardiac arrest but not
mortality
Author, year
Jones,
10
2006
Jones,
11
2006
Jones,
12
2010
Shapiro,
13
2010
Description of
RRS
Pre-intervention
had a unified
code/MET team
with
anesthesiology,
ICU and
cardiology
registrars, ICU
nurse and primary
service physician,
post intervention
separate the
functions dropping
the cardiology and
anesthesiology
members from the
separate MET
ICU registrar, ICU
nurse and
receiving unit
medical registrar.
Separate from the
code team
Rapid response
nurse (2 dedicated
positions),
patients on-call
physician
various, different
hospitals
Study Design
prospective
before after trial
Main Study
objectives
to assess whether
systems changes in
existing MET would
increase utilization
rate
Description of Organization
350 bed tertiary university
affiliated hospital
Implementation
Comments
Themes
Team composition
Quantitative data for
(separation of unified
utilization rates and
code/MET into separate incidence of true code calls
teams with separate
activations), Method of
activation (changing the
activation methods to
separate the teams),
Triggers (changing alert
criteria for calling a MET)
re-education on purpose
of MET, criteria, and the
changes
prospective
interventional but
continuous as
opposed to
before after with
defined
intervention
change
prospective
mixed, mostly
semi-structured
focus groups
assess education
program to increase
utilization of existing
MET
education, improved
communication, on-thejob aids (e.g., posters,
observational charts),
differences in MET usage
for medical vs. surgical
admissions
to determine if
mandatory activation
of MET improves
outcomes compared
to elective activation
to determine nurses
perceptions of RRS
impact on practice
and what constitutes
a successful RRS
multiple
D-183
Quantitative data on
utilization rate but it is
continuous so may wish to
exclude
Author, year
Williams,*
14
2011
Description of
RRS
RRT model with
ICU nurse, ED
nurse, reps
therapist
Study Design
focus group
methodology
Main Study
objectives
clarify nurses
perceptions of RRS
Description of Organization
156 bed community hospital
Implementation
Themes
experience with
activation, composition of
teams, concerns about
activating a RRT
advantage of RRT to
nurses and patients
D-184
Comments
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
D-185
11.
12.
13.
14.
Was the
allocation
sequence
adequately
generated?
Was the
allocation
adequately
concealed?
Were baseline
characteristics similar?
Were
incomplete
outcome data
adequately
addressed?
No
Were
baseline
outcome
measurements
similar?*
Unclear
Unclear
Was knowledge
of the allocated
interventions
adequately
prevented during
the study?
No
Anwar ul,
1
2010
2
Bader, 2009
Benson,
3
2008
Campello,
4
2009
5
Chan, 2008
6
Gerdik, 2010
Hanson,
7
2009
8
Hatler, 2009
9
Konrad, 2010
Kotsakis,
10
2011
Laurens,
11
2011
Lighthall,
12
2010
Rothberg,
13
2011
Santamaria,
14
2010
Sarani*,
15
2011
16
Scott, 2009
17
Shah*, 2011
Tibballs,
18
2009
No
No
No
No
No
No
Unclear
Unclear
No
No
No
No
No
Yes
Yes
No
No
Unclear
Unclear
No
No
Yes
Unclear
Unclear
No
Yes
Unclear
Unclear
No
Yes
Yes
No
No
No
Yes
Unclear
Unclear
No
No
No
Unclear
Unclear
Unclear
No
No
No
Yes
Yes
Yes
No
Yes
Yes
No
No
No
No
No
No
Unclear
Yes
Yes
No
No
Yes
Unclear
Unclear
No
No
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
No
No
Yes
No
Unclear
No
Yes
Yes
No
No
Yes
Unclear
Unclear
No
Yes
Yes
No
Unclear
Unclear
No
Yes
No
Unclear
Yes
No
No
Yes
Unclear
Unclear
No
Yes
Yes
No
No
No
Unclear
Unclear
No
Yes
Yes
No
No
No
No
No
No
Unclear
Yes
Unclear
No
Yes
No
Unclear
Unclear
Yes
No
No
No
Yes
Unclear
Yes
Unclear
Yes
Yes
D-186
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
D-187
10.
11.
12.
13.
14.
15.
16.
17.
18.
D-188
Evidence Tables for Chapter 32. Promoting Engagement by Patients and Families To
Reduce Adverse Events (NEW)
Table 1, Chapter 32. Evidence table: patients engagement
Author,
year
Weingart,
1
2004
McGuckin,
2
2004
Description of
PSP
Study
design
Multiple
Sample
interventions
size
or multifaceted
interventions
Proving patients RCT
with
personalized
209
medication list
to help prevent
medication
errors
Theory or Description of
logic
organization
model
No
No
Boston
teaching
hospital
A 24-bed
inpatient
rehabilitation
unit located in
an acute care
university
hospital
Contexts
Implementatio Measurement
n details
tool
Organizational
characteristics:
a 40-bed unit;
The unit used
paper
medication
order forms that
were faxed to
the pharmacy
and entered into
the hospitals
electronic
pharmacy
information
system; CPOE
not available at
time of study
Teamwork,
leadership,
culture: Nurse
manager was
member of
research team
Patient surveys;
identification of
med incidents
through
interviews of
pharmacists,
housestaff,
electronic
review
D-189
Outcomes:
Benefits
Outcomes: Influence of
Harms
contexts on
outcomes
Comments
adverse drug
rate between
intervention and
control
8.4% versus
2.9%, p=0.12
close-call rate
between
intervention
patients and
controls
(7.5% versus
9.8%) p=0.57
patients aware of
drug-related
mistakes during
the
hospitalization11%
Hand Hygiene
per resident day
5 to 9.7 during
intervention, 6.7
at 6 weeks, 7.0
at 3 months.
p<0.001 for all
timepoints
Patients
asked
physicians
40% of time,
nurses 95%
of time
% of patients
comfortable
asking - 75%
% of HCWs
washing hands
when asked by
patient-60%
Author,
year
Description of
PSP
Study
design
Multiple
Sample
interventions
size
or multifaceted
interventions
3
Stone 2007 Patient
Pre-post
empowerment
(materials
187 acute
telling patients hospitals
to ask HCWs to
clean their
hands).
Included other
interventions as
well as patient
engagement:
bedside alcohol
hand rub, ward
posters
changed
monthly, pts
encouraged to
ask HCWs to
clean their
hands). An
optional
component was
six-monthly
audit and
feedback of
hand hygiene
Theory or Description of
logic
organization
model
No
187 acute
hospitals
Contexts
Implementatio Measurement
n details
tool
Implementation
Tools: National
Patient Safety
Agencys Clean
Your Hands
Campaign
(CYHC) seeks
to improve 293
healthcare
workers
(HCWs) handhygiene
behaviour in
England and
Wales
Outcomes:
Benefits
Monthly median
alcohol hand rub
(AHR) use: 44
pre to 56 post;
p<0.001
Combined
median use of
AHR and soap:
13.2 to 31
ml/patient bedday;
Health careassociated
infection rates:
No changes
apart from
seasonal
changes in
norovirus and
CDAD
D-190
Outcomes: Influence of
Harms
contexts on
outcomes
increase may
have been
confounded
by a change
in soap/AHR
provider
Comments
limitations of
self-reported
data; high
response rate;
targeting use of
AHR, changed
many aspects of
hand-hygiene
behaviour,
increasing AHR
use in particular,
across the acute
sector of the
NHS without
reducing soap
usage. Audit and
feedback, a
component
emphasized
much less than
AHR and
posters, was
less widely
implemented.
Weingart,
1
2004
McGuckin,
2
2004
Stone
3
2007
Was the
allocation
sequence
adequately
generated?
Was the
allocation
adequately
concealed
?
Were baseline
outcome
measurement
s similar?*
Were baseline
characteristics
similar?
Were incomplete
outcome data
adequately
addressed?*
Yes
Unclear
Yes
No
Unclear
Yes
No
Yes
No
No
Yes
Yes
No
No
No
No
No
No
Unclear
not applicable
Unclear
No
Yes
Yes
D-191
Was knowledge
of the allocated
interventions
adequately
prevented during
the study? *
References
1.
2.
3.
D-192
Abstoss,
1
2011
Description of
PSP
Study
design
Multiple
interventions or
multifaceted
interventions
7 interventions:
-Culture:
Feedback on
errors (posters
and emails), QI
education and
training (TV
channel and
curriculum)
-System: CPOE,
medication
management
(pharmacist), pt
safety report
form revisions
Sample
size
Pre-post
Post 2009
(n = 85,
resp. rat =
90%)
Theory or Description of
logic
organization
model
Contexts
University of
Organizational:
Michigans
characteristics2007C.S. Mott
2009
Childrens
Hospital PICU
Implementation
details
Measurement
tool
Outcomes:
Benefits
Safety Attitudes
Questionnaire
(SAQ) *only 13
items related to
medication
error/reporting
are reported in
this study
Culture
survey:
Teamwork
climate:
52.8% to
71.8%
agreement;
Safety
climate:
54.6% to
63.4%
agreement
(not sig))
(p < .01) and
(p = .13)
4 cultural & 3
system-level
interventions
Reported
errors
resulting in
harm
0.56 to 0.15
events/10,000
doses
p<0.01
Overall error
reporting rate
3.16 to
3.95/10,000
doses
ns
D-193
Outcomes:
Harms
Influence
of
contexts
on
outcomes
comment
Comments
**Still
abstracting
AdamsPizarro,
2
2008
Description of
PSP
Study
design
Multiple
interventions or
multifaceted
interventions
Regional
improvement
collaboratives;
ICU initiatives
included
rounding and
daily goals,
ventilator bundle
Sample
size
Theory or Description
logic model
of
organization
IHI
20 ICUs
breakthrough
429 in 14 series
hospitals model,
(individual Culture
response Improvement
Guide
rate =
65%)
Pre-post
Contexts
Implementation
details
External:
Collaborative,
state safety
organization
Measurement Outcomes:
tool
Benefits
D-194
Culture
survey:
only 3 of 12
domains
showed
improvement;
decrease by
13.6 in one
overall
measure
(Safety
Grade) and
Overall
Perception of
Safety
decreased
1%
Outcomes: Influence
Harms
of
contexts
on
outcomes
comment
Comments
Also included
EDs and ORs
- only
including ICU
data here.
Only 56% of
ICUs
completed
both surveys
and are
included
Blegen,
3
2010
Description of
PSP
Study
design
Multiple
interventions or
multifaceted
interventions
Triad for Optimal
Patient Safety
(TOPS): (1) team
training, (2) unit
based safety
team, (3) patient
engagement in
daily goals
Sample
size
Pre-post
-3 inpatient
medical
units nested
in 3 different
hospitals;
post n = 368
(response
rate = 81%
Theory or Description of
logic
organization
model
1 academic, 1
community, 1
integrated
healthcare
system; 26-34
beds; 1nurse:45 patients
Contexts
Implementation
details
Organizational:
characteristics:
All hospitals
located in San
Francisco Bay
area, CA;
Differed in
physician care
model,
pharmacy
presence on
the unit, and
use of
information
technology,
2006-2007
Leadership:
leadership
provided
protected time
for a unit-level
project
champion on
each unit
D-195
AHRQ Hospital
Survey on
Patient Safety
(HSOPS)
Culture
survey:
Increases in
mean
dimension
score
significant
for 5 of 10
dimensions
p < .05 for
all 5
Site x time
interaction
on 6
dimensions;
Post diffs
among
professional
groups, 3
culture
dimensions
Comments
-Significant site x
time interactionsculture scores
for one hospital
did not change
or changed in
negative
direction without that unit,
the analysis in
the other 2
hospitals
showed changes
in all 10
dimensions (no
indication of
which unit was
dropped)
-Differences also
found between
professional
groups on 3
dimensions,
overall
perceptions of
safety, and
frequency of
event reporting;
nurses tended to
score culture
most positively
post intervention
compared to
physicians and
pharmacists
Author,
year
Cooper,
4
2008
Description of
PSP
Study
design
Multiple
interventions or
multifaceted
interventions
Simulation-based
anesthesia crisis
resource
management
training.
Sample
size
pre-post
with control
hospitals
293,
response
rate 38%
Theory or Description of
logic
organization
model
Contexts
Implementation
details
4 academic
Not reported
medical
centers
associated with
Harvard. 2
control
academic
medical
centers. in
Massachusetts.
One-day, 6 to 7
hour simulationbased anesthesia
crisis resource
management
training session in
4 hospitals. 2
control hospitals,
staff did not
receive training.
D-196
Patient Safety
Climate in
Healthcare
Organizations
(PSCHO)
Culture
survey:
No
significant
differences
Comments
Description of
PSP
Study design
Sample size
Multiple
interventions or
multifaceted
interventions
Donahue, prepare
5
2011
paraprofessionals
to communicate
changes in
patient status
using structured
communication,
including
reducing cultural
barriers to
interdisciplinary
communication
Pre-post
Theory or Description
logic model
of
organization
IHI Spread
for Change
111 (41%) post
Framework,
survey
Crew
(paraprofessionals Resource
only)
Management
techniques
Danbury
Hospital,
Danbury,
CT; Not
described
Contexts
Leadership:
Chief Nursing
Executive,
Chief Medical
Officer involvement
and
messaging
Implementation
details
Measurement
tool
Outcomes:
Benefits
AHRQ
Hospital
Survey on
Patient Safety
(HSOPS)
Culture survey:
Reported
change
(improvement
of >10%) on
4/42 survey
items
Use of
structured
communication
(SBAR)
74% to 90%
increase
Not reported
Rapid
response
events
reported
Increase from
351 to 487;
Decrease in
number of
RRS events
that led to
code events
(29% pre, 22%
post)
Not reported
D-197
Outcomes: Influence
Harms
of
contexts
on
outcomes
comment
Comments
Also includes
some
qualitative data
from focus
groups
describing
change in
communication
after
implementation
Author,
year
Frankel,
7
2008
Description of
PSP
Study design
Sample size
Multiple
interventions or
multifaceted
interventions
rigorous
Pre-post
WalkRounds
702 of all
professions, 21
patient care
areas, resp rate
60% (n =1, 256;
45% RN)
Theory or Description
logic model
of
organization
Contexts
Implementation
details
D-198
Measurement
tool
Outcomes:
Benefits
Safety
Attitudes
Questionnaire
(SAQ)
Culture survey:
62% to 77% in
1 hospital,
46% to 56% in
2nd hospital
p=0.03 and
0.06 in the 2
hospitals
Outcomes: Influence
Harms
of
contexts
on
outcomes
comment
Comments
Only 2 of 7
hospitals
complied fully
with approach
& only those
results are
reported; also
only SAQ
results from
units with
>50%
response rate
are reported
(about half of
the units). 2
hospitals did
not implement
Walk Rounds
rigorously -no
significant
change in
those units
Structured
interdisciplinary
rounds each
weekday using
structured
communication
tool;
Interdisciplinary
working group
met for 12 weeks
before
implementation,
developed format,
frequency, timing
Safety
Attitudes
Questionnaire
(SAQ)
Culture survey:
Teamwork climate:
Mean, control = 77.3,
intervention = 82.4;
Safety climate, mean:
Control = 75.4,
Intervention = 76.5
Teamwork (p = .01);
Safety (n.s., p = .90))
Length of Stay
adjusted LOS was
0.19 days longer for
the intervention unit
vs. control
n.s. (p = 0.17)
Cost
adjusted cost was
$24.05 less for the
intervention unit vs.
control
n.s. (p = 0.94)
D-199
Comments
Teamwork
climate was
significantly
higher for
intervention
unit nurses
(83.5 vs. 74.2,
p = .005), but
there was no
significant
difference on
safety climate.
Teamwork and
safety climate
were rated
higher by
intervention
Physicians but
not statistically
significant
Author,
year
OLeary,
8
2011
Description of
PSP
Study
design
Multiple
Sample
interventions or
size
multifaceted
interventions
Structured Inter- concurrent
Disciplinary
control
Rounds
49 nurses in
both units
(84%
response
rate)
Theory or Description
logic
of
model organization
Contexts
every weekday,
30 minutes, led
by nurse manager
and unit medical
director, used
structured
communication
tool
Safety
Attitudes
Questionnaire
(SAQ)
(teamwork
and safety
domains)
Culture survey:
median 75
intervention, 61.1
control
p=0.03
Rating of quality of
communication and
collaboration with
hospitalizes
80% intervention vs.
54% control
p<0.01
Teamwork climate
median 85.7
intervention, 61.6
control
p=0.008
D-200
Comments
Author,
year
Paine,
10
2010
Description of
PSP
Study
design
Multiple
interventions or
multifaceted
interventions
Comprehensive
Unit-Based
Safety Program,
hospital-wide
training, culture
score goal setting
Sample
size
Hospital-wide,
multiple
interventions
Pettker,
11
2009
Pre-post
5461
surveys
post (144
units), 79%
response
rate
protocol
Pre-post
standardization,
creation of
not
patient safety RN reported
position and
patient safety
committee, team
skills training
risk-reduction
clinical practices
and creation of a
comprehensive
culture of safety.
Theory Description
or logic
of
model organization
Contexts
Tertiary-level
academic
medical
center. OB
service
averages
5500
admissions
per year.
Organizational:
characteristics:
academic
medical center,
OB service,
urban/suburban
Implementation Measurement
details
tool
Outcomes:
Benefits
Culture survey:
Improvements in
all SAQ domains
except stress
recognition
p<0.001
Incremental
Safety Attitudes
2004-2006:
Questionnaire
(SAQ)
expert review,
protocol
standardization,
patient safety
RN and
committee, team
skills training,
fetal heart
monitoring
certification,
crew resource
management
team training
Culture survey:
%reporting good
teamwork climate
& good safety
climate improved
from 38.5% to
55.4% and 33.3%
to 55.4%,
respectively.
D-201
Adverse
outcomes index
indicators
3.3% pre to 1.6%
post
P= .011
Outcomes: Influence
Harms
of
contexts
on
outcomes
comment
Comments
Culture scores
decreased in
17 of 144 units,
details not
reported,
informal
interviews
suggested that
manager
turnover, unit
construction,
and
implementation
of new IT may
have
contributed to
lower scores
Author,
year
Pettker,
12
2011
Riley,
14
2011
Description of
PSP
Study
design
Multiple
interventions or
multifaceted
interventions
NOT
ABSTRACTEDthis article
already
abstracted - just
culture results are
here
Sample
size
TeamSTEPPS
didactic training
program,
TeamSTEPPS
with in-situ
simulation
training exercises
Didactic training
with in-situ
simulation v.
didactic only
Theory
or logic
model
Description
of
organization
Contexts
Implementation Measurement
details
tool
Not
recorded
post 183 response
rate 72%
Reasons 3
rural/suburban
model
US
community
not reported
labor and
how many
delivery units,
completed
50-66 beds
survey
Cluster
RCT
Organizational:
characteristics:
midwest, 20052008
Outcomes:
Benefits
Culture survey:
Statistically
significant
improvements on
4 domains,
worsening 1
(Perceptions of
favorable working
conditions, no
change 1
<p 0.05 for 5 (4
improved, 1
worse)
Local tailoring of Safety
Culture survey:
TeamSTEPPS;
Attitudes
perinatal
simulation
Questionnaire outcomes Weighted Adverse
included detailed (SAQ)
Outcomes Score
debriefing// 1
%change -37%
hospital full, -1% didactic,
TeamSTEPPS
+43% control
didactic training
p<0.05 for full
(condensed); 1 intervention
TeamSTEPPS
with series of insitu training
exercises,
repeated until
staff targets were
met
D-202
Outcomes: Influence
Harms
of
contexts
on
outcomes
comment
Comments
NOT
ABSTRACTEDthis study
already
abstracted for
the other article
reporting on
this - just
culture results
are here. NO
ROB done
either.
cluster RCT
Author,
year
Sexton,
15
2011
Description of
PSP
Study
design
Multiple
interventions
or multifaceted
interventions
CUSP
(Comprehensive
Unit-Based
Safety
Program), as
part of Keystone
ICU project
Sample
size
Pre-post
Pre n =
4,260
(overall
res. rate =
71%; 99
ICUS);
Post n =
3,533
(overall
res. rate =
73%; 71
ICUs)
Theory
or logic
model
Description
of
organization
71 Michigan
hospitals
71 ICUS
-68%
Teaching
-31% Faithbased
-27% bed sz.
>=500
-25% bed sz.
200-299
Contexts
Implementation
details
Measurement
tool
Organizational: CUSP
Safety Attitudes
characteristics: Intervention_4Steps: Questionnaire
see description (1) Educate staff on (SAQ)
science of safety;
(2) Identify
errors/defects, (3)
partner with senior
leadership, (4) Use
tool to learn from
one defect per
month//ICU project
teams created. In
addition to CUSP,
also implemented
intervention to
reduce CLABSI
and/or VAP, and
Daily Goals
checklist.
Implementation
period for each
intervention approx.
3 months.
D-203
Outcomes:
Benefits
Culture survey:
The overall mean
percent positive
scores increased
significantly from
baseline to
follow-up. Nine
ICUs met 60%
positive criteria
for success in
2004
Mean safety
culture %positive
(pre = 42.5%,
post = 52.2%, p <
.001); (Results
for specific
questions
reported in Table
2)
Outcomes: Influence
Harms
of
contexts
on
outcomes
comment
Hosp. size,
faithbased:
Gain
higher for
faith-based
and small
h., though
diffs. not
tested
directly
Comments
Only reported
data from ICUs
that reported
culture surveys
at both time
points (71 of
127 total ICUs)
Author,
year
Thomas,
16
2005
Tiessen,
17
B., 2008
Description of
PSP
Study
design
Multiple
Sample
interventions
size
or
multifaceted
interventions
Executive
Cluster
walkrounds
RCT
evaluate and
reevaluate
patient safety
culture,
encourage
patient safety
learning, share
stories, weekly
executive
walkrounds,
prioritize
improvement
efforts, identify
staff safety
concerns,
implement
improvements
8 point patient
safety plan
Pre-post
not stated;
35%
response
rate
Theory
or logic
model
Description
of
organization
Contexts
Implementation Measurement
details
tool
711 bed
Not reported
academic
tertiary care
hospital
inpatient units.
In units
randomized to
receive
executive walk
rounds
intervention,
executives
rounded once
every 4 weeks
for 3 visits per
unit.
88-bed, acute External:
The patient
care, rural
Canadian Council safety practices
on Health
community
were rolled out,
Services
hospital in
hospital-wide
Accreditation new over a 2 year
Ontario,
Patient Safety
Canada.
period (2005Required
2007).
Practices,
upcoming
accreditation
survey
Organizational:
characteristics:
hospital financial
situation issues,
staffing changes
and reductions
Leadership:
created Patient
Safety Action
Plan
D-204
Outcomes:
Benefits
Outcomes: Influence of
Harms
contexts on
outcomes
comment
Comments
Patient Safety
Climate in
Healthcare
Organizations
(PSCHO)
Culture survey:
significant
improvement on
only 2 of 30
statements,
significant
decrease on one
statement
one
statement
did show
decrease 46
to 29%
p=0.01
Hospital
financial
issues may
have
impacted
effectiveness
Significant
improvement
on 2
statements:
asking for help
not a sign of
incompetence,
and If I make a
mistake that
has serious
consequences,
I tell someone
about it.
Author,
year
Timmel,
18
2010
Description of
PSP
Study
design
Multiple
interventions
or
multifaceted
interventions
CUSP; video,
identify safety
hazards, learn
from defects,
teamwork and
communication
tools
Sample
size
Pre-post
n = 28
(100%
response
rate)
Theory
or logic
model
Description
of
organization
Contexts
1 surgical unit
in a large,
urban
academic
medical center
Leadership:
Senior hospital
executive
participated as
part of CUSP
team
Implementation
details
Measurement
tool
Outcomes:
Benefits
Culture survey:
Safety climate:
80% to 90%;
Teamwork
climate: 56% to
80%; hospital
management:
39% to 47%; Unit
Management:
62% to 68%;
Working
conditions: 48%
to 55%; Stress
recognition: 45%
to 46%
all composite
scores except
stress recognition
significantly
improved from 06
to 08 (p < .001)
Nurse turnover
3/12 FTEs left in
2006; 0/16 left in
2008 and 2009
D-205
Outcomes: Influence
Harms
of
contexts
on
outcomes
comment
Comments
Author,
year
Description of
PSP
Multiple
interventions or
multifaceted
interventions
Edwards, Ad-hoc safety
6
rounds,
2008
enhancements to
error reporting
system and
related
education,
standardized
communication
protocols
(SBAR), transfer
of care checklist,
implementation of
EMR system
Study
design
Sample
size
Pre-post
n = 428
(Response
rate =
32%)
Theory Description
or
of
logic organization
model
2 metropolitan
pediatric
tertiary care
hospitals in
same health
system; 1
academic, 1
community
Contexts
Implementation
details
Organizational: 2 initiatives to
see description address error &
perceptions of
safety: Safety
Leadership:
rounds, reporting
supported
system updates; 3
survey
initiatives designed
to improve hand
offs and transitions
across units:
SBAR, transfer
checklist, EMR
5 components,
staggered
implementation
over time
D-206
Culture
survey
results:
6/11
domains
significantly
increased, 1
significantly
decreased,
2 no
significant
change.
P< 0.01
community
hospital
scored
higher on 3
domains
compared
to
academic
hospital
Comments
Author,
year
Description of
PSP
Study
design
Multiple
interventions
or
multifaceted
interventions
Pronovost, Comprehensive
13
2005
unit based
safety program:
(1) assess
culture, (2)
science of
safety
education, (3)
staff
identification of
safety
concerns, (4)
senior
executive
involvement, (5)
improvements
implemented
from #3, (6)
efforts
documented,
(7) results
shared, (8)
reassessment
of culture
Sample
size
Quasistepped
wedge
design
WICU n =
64
(response
rate =
86%);
SICU n =
23
(Response
rate =
84%)
Theory
or logic
model
Description
of
organization
Contexts
2 ICU units in
a large
metropolitan
tertiary care
hospital
Leadership:
Each unit was
adopted by a
senior level
executive;
dedicated
improvement
team to support
intervention
implementation
Implementation Measurement
details
tool
Forms for
Safety climate
collecting/sharing scale (SCS)
improvement
success stories,
daily goals
sheets, tool for
medication errors.
Science of safety
education, staff
identify how next
patient will be
harmed and how
to prevent
Outcomes:
Benefits
Culture survey
results (nurse
turnover and
length of stay)
WICU : 35%
positive climate to
52% positive
climate post; SICU
35% positive
climate to 68%
positive climate
post
Stats reported for
individual
questions, but not
for overall domain
changes; 8/10 p
<0.05 in WICU,
3/10 in SICU had p
< .05
WICU: decreased
from 9% to 2%;
SICU: Decreased
from 8% to 2%
P = NS
CUSP (8 step
version)
WICU: decreased
from 2 days to 1
day; SICU:
Decreased from 3
days to 2.3 days
P< 0.05
D-207
Outcomes: Influence
Harms
of
contexts
on
outcomes
comment
Comments
pre-post study
in both units
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
D-208
10.
11.
12.
13.
14.
15.
16.
17.
18.
Evidence Tables for Chapter 34. Effect of Nurse-to-Patient Staffing Ratios on Patient Morbidity
and Mortality
Table 1, Chapter 34. Evidence table
Author, Year
Unruh and
1
Zhang, 2012
A conceptual
model is
presented that
relates case mix,
location,
ownership, size,
and payer mix
with changes in
nurse staffing
over time
None
Outcomes:
Harms
D-209
Influence of
Contexts on
Outcomes*
Overall
Risk of
Bias
Case mix,
urban status,
Medicaid, and
HMO days of
care were all
positively
related to
changes to
failure-torescue
High
High patient
Low
turnover also
associated with
worse
outcomes
Author, Year
Twigg et al.,
3
2011
Contexts
Outcomes:
Harms
Influence of
Contexts on
Outcomes*
Overall
Risk of
Bias
For all
patients and for
medical and surgical
patients the death
rate decreased
significantly
None
mentioned
None
mentioned
High
Lower patient-to-nurse
staffing ratios were
associated with lower
30-day mortality and
failure to rescue
25% of RNs
None
reported they mentioned
perceived
decreased
support from
LVNs, 34% of
RNs reported
decreased
support from
unlicensed
personnel
High
Australia
Nurse hours of care and skill
mix
Academic states assessed
No assessment of existing
quality / safety infrastructure
Organizational infrastructure
described in terms of
comprehensive clinical
services being provided
Aiken et al.,
4
2010
Not a study of
an
intervention
per se, rather
California
legislation
mandated
certain nurseto-patient
ratios
Cross-sectional None
22,336 hospital
nurses in
California
D-210
Author, Year
Harless and
5
Mark, 2010
Contexts
Outcomes:
Harms
Influence of
Contexts on
Outcomes*
Overall
Risk of
Bias
Not relevant
None
mentioned
None
mentioned
High
None
mentioned
High
USA
Numerous financial and
economic payer variables
No assessment of quality /
safety infrastructure
No assessment of
organizational infrastructure
SCTL: Not assessed
39 Michigan hospitals
USA
Nurse staffing estimated by
taking the ratio of each
hospitals FTE for RN and
dividing by patient-days
No assessment of existing
quality / safety infrastructure
No assessment of
organizational infrastructure
SCTL: Not assessed
D-211
Author, Year
Aiken et al.,
7
2008
Contexts
Outcomes:
Harms
Influence of
Contexts on
Outcomes*
Overall
Risk of
Bias
None
mentioned
Better care
environments
were found to
be associated
with lower 30
day mortality
and lower 30day failure-torescue.
High
None
mentioned
High
USA
RN staffing mean patients
per nurse, Nurse education ,
A composite score of the care
environment, encompassing
subscales from the Nursing
Work Index
Academic status assessed
No assessment of existing
quality safety infrastructure
High vs. low technology
assessed
Cho et al.,
8
2008
None
SCTL:
Practice Environment Scale
(PES) of the Nursing Work
Index (NWI) gets at a related
concept.
236 hospitals
Not relevant, not Each additional
None
a study of an
assigned patient per mentioned
Korea
intervention
RN in secondary
hospitals was
Mean years of ICU nurse
associated with a 9%
experience, RN staffing
increase in odds of
dying; there was no
SCTL:
statistically significant
Nonassociation of nursing
US/UK/Canada/Australia/New
experience with
Zealand study
mortality.
D-212
Author, Year
None
Convenience
sampling of 396
patients
Outcomes:
Harms
No statistically
None
significant
mentioned
associations were
found in risk-adjusted
ICU mortality.
Influence of
Contexts on
Outcomes*
Overall
Risk of
Bias
None
mentioned
High
None
mentioned
High
Greece
Daily nursing workload/
workload exposure
None
SCTL:
NonUS/UK/Canada/Australia/New
Zealand study
54 neonatal ICUs.
Not relevant, not Higher specialist
None
a study of an
nursing provision was mentioned
UK
intervention
statistically
significantly
Total number of RNs per
associated with a
shift, Nursing provision ratio
lower risk-adjusted
per shift, Specialist nursing
observed mortality
provision ratio per shift
rate.
Academic status not reported
No assessment of existing
quality / safety infrastructure
No assessment of
organizational complexity
SCTL:
Not assessed
D-213
Author, Year
Mark et al.,
11
2007
Contexts
Influence of
Contexts on
Outcomes*
Overall
Risk of
Bias
No relationship was
None
found between inmentioned
hospital pediatric
death and nurse
staffing for
hospitalized California
pediatric patients.
None
mentioned
High
None
mentioned
High
Outcomes:
Harms
RN staffing, Licensed
vocational nurse (LVN)
staffing, Unlicensed hours of
care provided per patient day
Academic status assessed
No assessment of existing
quality/safety infrastructure
Presence of pediatric ICU or
NICU
SCTL:
Not assessed
30 acute trusts in 1998.
UK
Mean hospital patient-nurse
ratio derived from survey of
nurses
Academic status assessed
No assessment of existing
quality/safety infrastructure
No assessment of
organizational infrastructure
SCTL:
Not assessed
D-214
Author, Year
Stone et al.,
13
2007
Outcomes:
Harms
Influence of
Contexts on
Outcomes*
A conceptual
framework was
presented that
related the
potential
contributions of
patient
characteristic,
structures of
care, and
administrative
processes
including
organizational
climate, staffing,
overtime and
wages on patient
outcomes.
An increase in
catheterassociated
bloodstream
infections in
organization
with a more
positive
organizational
climate
No significant High
relationship
was observed
between
organizational
climate and 30day mortality.
D-215
No significant
relationship was
observed between
overtime use and 30day mortality.
Overall
Risk of
Bias
Author, Year
Tourangeau,
Doran, et al.,
14
2006
A conceptual
framework was
presented that
included
numerous
variables in six
categories:
condition of the
hospital practice
environment,
nurse staffing,
physician
expertise, nurse
and nurse
employment
characteristic,
care
management
processes (use
of care
maps/protocols),
and hospital
type/location on
30-day mortality.
Outcomes:
Harms
D-216
Influence of
Contexts on
Outcomes*
Overall
Risk of
Bias
Author, Year
Estabrooks et
15
al., 2005
Contexts
Outcomes:
Harms
D-217
Influence of
Contexts on
Outcomes*
Overall
Risk of
Bias
NurseHigh
physician
collaboration
was associated
with lower 30
day mortality
rates.
Author, Year
Halm et al.,
16
2005
Contexts
USA
RN staffing
Academic status not reported
No assessment of existing
quality / safety infrastructure
No assessment of
organizational complexity
SCTL:
Maslach Burnout Inventory
assessed
D-218
Outcomes:
Harms
Influence of
Contexts on
Outcomes*
Overall
Risk of
Bias
None
mentioned
High
Author, Year
Person et al.,
17
2004
None
D-219
Outcomes:
Harms
Lower in-hospital
None
mortality rates were
mentioned
associated with higher
RN staffing in
hospitals.
Higher in-hospital
mortality rates were
associated with higher
licensed
vocational/practical
staffing in hospitals.
Influence of
Contexts on
Outcomes*
Overall
Risk of
Bias
None
mentioned
High
Author, Year
Aiken et al.,
18
2003
Contexts
Outcomes:
Harms
Influence of
Contexts on
Outcomes*
Overall
Risk of
Bias
None
mentioned
High
None
mentioned
High
SCTL:
Not assessed
17 inpatient units in one
university hospital
Thailand
Ratio of total nursing staff to
patients, Proportion of RN to
total nursing staff, Mean
years RN experience,
Percentage of baccalaureateeducated nurses.
SCTL:
Non- US/UK/Canada/
Australia/New Zealand study
D-220
Author, Year
Aiken et al.,
20
2002
Contexts
None
Not relevant, not Higher patient-toa study of an
nurse ratio found to be mentioned
associated with higher
intervention
30-day mortality (p <
.001). Odds of patient
death increased by
7% for every
additional patient in
nurse workload.
USA
RN staffing
Academic status assessed
No assessment of existing
quality or safety
High vs. low technology
assessed
SCTL:
Not assessed
D-221
Outcomes:
Harms
Influence of
Contexts on
Outcomes*
Overall
Risk of
Bias
None
mentioned
High
Author, Year
None
Outcomes:
Harms
None
Not relevant, not No statistically
a study of an
significant
mentioned
intervention
relationships were
found between inhospital mortality rates
USA
and nurse staffing
Eight nurse staffing indicators
indicators.
were assessed: Number of
RN hours of nursing care per
Two statistically
patient day, Number of
significant
licensed practical nurse hours
relationships were
per patient day, Number of
found between lower
aid hours of care per patient
hospital failure-today, Total hours of nursing
rescue rates and
care per patient day,
nurse staffing:
Proportion of RN hours of all
For medical patients, a
hours of nursing care,
higher proportion of
Proportion of licensed
hours of care provided
practical nurse hours of all
by RNs.
hours of nursing care,
Number of hours of care
For surgical patients, a
provided by licensed nurses
greater number of
(RN + practical nurse) per
hours of care provided
patient day, RN hours as a
by RNs.
proportion of licensed nurse
hours.
799 hospitals in 11 states
from 1997 and 1998 fiscal
years
D-222
Influence of
Contexts on
Outcomes*
Overall
Risk of
Bias
None
mentioned
High
Author, Year
A conceptual
framework was
presented that
included eight
domains: nurse
staffing, nurse
skill mix,
professional role
support, nurse
characteristic,
nurse practice
environment
condition,
continuity of
registered nurse
care provider,
and other
determinants on
30-day mortality.
D-223
Outcomes:
Harms
Influence of
Contexts on
Outcomes*
Overall
Risk of
Bias
Condition of
High
nursing practice
environment
was not
associated with
lower 30-day
mortality.
Author, Year
Tarnow-Mordi
23
et al., 2000
None
Outcomes:
Harms
Influence of
Contexts on
Outcomes*
Overall
Risk of
Bias
None
mentioned
High
D-224
References
1.
2.
3.
4.
5.
6.
7.
D-225
8.
10.
11.
12.
13.
14.
15.
16.
17.
18.
21.
19.
22.
20.
23.
D-226
Evidence Tables for Chapter 35. Patient Safety Practices Targeted at Diagnostic
Errors (NEW)
Table 1, Chapter 35. Evidence table
Author, Year
Diagnostic Error
CarewMcColl,
(49)
1983
Diagnostic
interpretation of
radiographs
Robson,
(58)
1985
Diagnostic
interpretation of
radiographs
Ciatto,
(40)
1995
Diagnostic errors
during readings of
mammograms
Lind, 1995
(53)
Diagnostic errors in
surgical pathology
reports
Experimental Intervention
Patient or Related
Outcome
Identification of a missed
injury
Review of x-rays in an
accident and emergency
department
D-227
Notes
Missed diagnosis;
misdiagnosis
Missed diagnosis;
misdiagnosis
Missed diagnosis;
misdiagnosis;
proof of concept
Missed diagnosis
Misdiagnosis;
delayed
diagnosis
Author, Year
Bruner,
(47)
1997
Dudley,
(29)
1997
Thiesse,
(59)
1997
Lufkin,
(54)
1998
Espinosa,
(50)
2000
Diagnostic Error
Experimental Intervention
Patient or Related
Outcome
Diagnostic
Review of brain or spinal
Substantial and serious
discrepancies in brain cord biopsy results by a
neuropathology diagnostic
and spinal biopsy
neuropathology consultation errors
reports
service
D-228
Notes
Misdiagnosis
Missed diagnosis
Diagnostic
discrepancy:
proof of concept
Misdiagnosis
Missed diagnosis;
misdiagnosis;
delayed
diagnosis
Author, Year
Nam, 2001
(30)
Westra,
(60)
2002
Canon,
(48)
2003
Kwek, 2003
Nordrum,
(44)
2004
*Howard,
(42)
2006
(52)
Diagnostic Error
Experimental Intervention
Patient or Related
Outcome
Misdiagnosis due to
Repeat colonic transit study Success rate post
inaccurate
in patients with chronic
colectomy for chronic
interpretation of
constipation and suspected constipation
colonic transit study in colonic inertia to confirm the
patients with chronic diagnosis prior to colectomy
constipation
Missed diagnosis of
A secondary review of
Treatment modification
head and neck
histopathologic diagnoses
based on changes in
cancers
diagnoses
Detection of polyps
Secondary reading of
Diagnostic accuracy of
and/or colorectal
barium enemas
polyps and carcinomas
carcinomas
D-229
Notes
Misdiagnosis
Missed diagnosis;
misdiagnosis
Missed diagnosis;
misdiagnosis
Missed diagnosis;
misdiagnosis
Misdiagnosis;
delayed
diagnosis
Missed diagnosis
Author, Year
(45)
Raab, 2006
Singh, 2006
(46)
Duijm, 2007
(41)
Manion,
(43)
2008
Parameswara
(56)
n, 2008
Diagnostic Error
Experimental Intervention
Patient or Related
Study Design: Result
Outcome
Diagnostic errors in
Pre/post: Double viewing did not lower the
Pre sign-out double viewing Incorrect diagnoses (that
interpretation of
frequency of cytologic-histologic correlation
of all pulmonary cytology
could impact patient
pulmonary cytology
slides
management and outcome) false-negative errors. Double viewing
slides (based on
detects errors in up to 1 of every 37 cases.
correlation of cytology
While the double cytology slide viewing
and surgical
was helpful at some project sites in
specimens histology
detecting pre-sign out error, the
results)
intervention did not significantly reduce
error frequencies at any of three study
sites. Agreement with subsequent surgical
diagnosis was moderate when definitive
diagnoses were made.
Diagnostic error for
Urgent referral; timingPresence of cancer
Other: Retrospective review and audit;
head and neck cancer based referral to reduce
diagnosis, and the time
86% of urgent patients were seen within 2
delay in diagnosis and
delay to reach diagnosis
weeks. 24% had oral squamous cell
influence on diagnostic
carcinoma.
pathway
Missed breast cancer Independent double reading Cancer detection rates and Other: Additional reading by technologists
diagnosis
of mammograms by two
referral rates of women with increased the cancer detection rate by 0.36
cancers per 1,000 women and the referral
mammography
positive screening results
rate by 0.13%.
technologists beyond the
from any reader
standard double reading by
two radiologists
Clinically significant
Second opinion in pathology Diagnostic accuracy no
Other: Retrospective chart review of major
diagnostic errors
to expose clinically
major disagreement in
diagnostic disagreements (2.3% of
(varied clinical
significant errors
pathology that would impact reviewed cases). Second opinion for
conditions)
treatment or prognosis
clinically significant error validated in 34 of
132 cases reviewed.
Missed abnormal
Sampling the remaining
Identification of pathologic
Other: Review sampling showed pathologic
findings in histology
tissue of colorectal biopsies abnormalities in remaining
abnormality in 3.9% cases. New diagnostic
specimens
originally diagnosed as
biopsy tissues (originally
information identified in 1.7% of cases, but
normal- with additional step reported as normal)
lost in 1.3% of cases (present in initial
sections- to reveal
sections but not in remaining tissue).
pathologic abnormalities
D-230
Notes
Missed diagnosis;
misdiagnosis
Delayed
diagnosis
Missed diagnosis
Missed diagnosis;
misdiagnosis
Missed diagnosis;
delayed
diagnosis
Author, Year
(57)
Raab, 2008
Murphy,
(55)
2010
Hamady,
(71)
2005
Diagnostic Error
Experimental Intervention
Patient or Related
Outcome
Incorrect surgical
Two diagnostic error
Although this was a
pathology diagnoses detection processes:
retrospective review of
targeted review of a random surgical pathology
5% of surgical pathology
specimens, the study
specimens, and focused
evaluated: a) impact
review by 3 subspecialty
identified diagnostic errors
pathologists of cases with a could have had on patient
perceived higher level of
outcomes, including
diagnostic uncertainty or
management, and b)
lack of standardization in
whether the patient
terminology
experienced subsequent
harm
Missed colonic and
Double reporting by two
Identification of clinically
extra-colonic lesions
radiologists of minimal
relevant colonic and extrain minimal preparation preparation CT of colon
colonic lesions that could
Computer
(MPCTC) in elderly patients impact future patient
Tomography (CT) of
management
colon
Incorrect
Pathology reports that
Malignancy status of tumor,
interpretation of
received discrepant
indicated course of
pathology reports for interpretations from a
treatment and expected
thyroid cancer
referring and receiving
prognosis
clinician were reviewed by a
third clinician blinded to the
thyroid cancer diagnosis
Educational Interventions
McCarthy,
Incorrect diagnosis by Teach parents Acute Illness Number of infants with
(86)
1990
parents of symptoms Observation Scale (AIOS)
serious illnesses
of serious illness
to detect childs illness vs.
3-point global scoring
system for evaluating
chance of serious illness
D-231
Notes
Missed diagnosis
Misdiagnosis;
proof of concept
Misdiagnosis
Author, Year
Diagnostic Error
Fridriksson,
(87)
2001
Misdiagnoses of
sudden onset
headache (an early
sign of ruptured
aneurysm)
Thaler,
(88)
2010
Errors in ECG
readings due to
switched electrode
cables
Experimental Intervention
A community teaching
program on educating local
physicians about sudden
onset of headache in
subarachnoid hemorrhage
(SAH); continuous
interaction between
neurosurgeons and local
physicians including
seminars on SAH, individual
follow-up of all referred
patients
A 45 min teaching session
for ICU nurses and
physicians about correct
ECG recording and errors
resulting from improper
electrode placements
Personnel Changes
Incorrect diagnoses of Comparison of diagnostic
De Lacey,
(84)
radiographs in
1980
accuracy between casualty
accident and
officers and radiologists
emergency
departments
(37)
Sakr, 1999
Clinically important
errors, including
errors in the diagnosis
pathway (i.e., history,
physical examination,
and radiographic
interpretation errors)
Patient or Related
Outcome
Early misdiagnoses of
ruptured aneurysms;
aneurysm surgery rates,
surgery outcomes and
morbidity and mortality
outcomes at 6 months post
SAH
D-232
Missed diagnosis
Notes
Missed diagnosis;
misdiagnosis
Missed diagnosis;
misdiagnosis
Author, Year
Diagnostic Error
Klassen,
(92)
1993
Missed positive
radiographic findings
(fracture, dislocation
or effusion) after
trauma
Schriger,
(39)
2001
(93)
Experimental Intervention
Patient or Related
Outcome
Notes
Diagnostic accuracy;
improvement of patient
outcomes (mortality and
morbidity) by identifying
missed injuries
Number of positive
radiographic findings;
number of missed positive
radiographic findings and
long-term clinical
importance thereof, in
pediatric trauma patients
Missed diagnosis
Missed diagnosis
Biffl, 2003
Missed injuries in
Level I trauma center
Soundappan,
(94)
2004
Missed injuries
associated with
trauma (pediatric)
Incidence of missed
diagnosis
D-233
Missed diagnosis;
delayed
diagnosis
Missed diagnosis
Author, Year
Diagnostic Error
*Perno,
(25)
2005
Delayed diagnosis of
injury in a Pediatric
Trauma Center
Ursprung,
(95)
2005
Diagnostic errors
related to laboratory
tests or radiologic
studies; delays in
patient care or
information transfer/
communication
(additional errors
probed)
Incorrect
interpretation of
frozen sections of
pathology specimens
(98)
Raab, 2006
(96)
Raab, 2006
Improved diagnostic
accuracy from Pap
test
Experimental Intervention
Patient or Related
Study Design: Result
Outcome
Implementation of a
Delayed diagnoses of injury Other: DDI occurred in 15 (0.46%) of
trauma patients. Previous study by same
Pediatric Trauma Response (DDI) in admitted pediatric
group revealed 4.3% DDI, an almost 10team and trauma service for trauma patients
fold decrease between the two studies.
severely injured children in
Among the 15 DDI cases in the latter
Pediatric Trauma Centers
study, 13 diagnoses were identified by
tertiary examination, and 2 patients were
discharged without diagnosis and returned
to the hospital after worsening symptoms.
Real time patient safety
Impact of errors (i.e., delays Other: 338 errors detected; 27 of 36 items
auditing during routine
in patient services or errors on checklist detected >1 error. Significant
clinical work in the ICU (36- in information transfer) on
safety errors were detected promptly and
item patient safety checklist patient clinical management rapid changes in policy and practice
focused on several errors
and on adverse outcomes
ensued.
including diagnostic errors)
Number of frozen
permanent section
discordant results and
deferred diagnoses (that
could impact patient
management and outcome)
D-234
Notes
Missed diagnosis;
delayed
diagnosis
Missed diagnosis;
misdiagnosis;
delayed
diagnosis
Author, Year
(97)
Raab, 2006
(99)
Diagnostic Error
Patient or Related
Outcome
Thyroid gland fine
Standardized terminology
Diagnostic accuracy of the
needle aspiration
scheme (Toyota Production FNA interpretation; surgery
(FNA) diagnostic error System Process Redesign) rates and repeated FNA
for reporting of cytologic
rates
results from thyroid fine
needle aspirations (FNA)
Raab, 2008
Mueller,
(100)
2010
Geriatric health
problems previously
unknown to a general
practitioner (GP) and
overlooked treatment
needs
De Vries,
(101)
2011
Surgical diagnosis
accuracy
Experimental Intervention
D-235
Notes
Missed diagnosis
Missed diagnosis;
misdiagnosis
Missed diagnosis;
delayed
diagnosis
Missed diagnosis;
misdiagnosis;
delayed
diagnosis; proof
of concept
Author, Year
Technique
Attard,
(72)
1992
Diagnostic Error
Experimental Intervention
Patient or Related
Outcome
Incorrect diagnosis in
patients presenting
with abdominal pain
Resnick,
(73)
1996
Incorrect diagnosis of
urinary incontinence
in nursing home
women
Misdiagnoses of urinary
stress incontinence
Borgstein,
(74)
1997
Incorrect appendicitis
diagnosis
Vermeulen,
(75)
1999
Incorrect appendicitis
diagnosis
Prieto,
(76)
2003
Incorrect indication of
surgical margins of
melanocytic lesions in
en face frozen
compared to
permanent paraffinembedded sections
Diagnostic accuracy;
whether surgery was
deemed necessary or not
D-236
Notes
Misdiagnosis
Missed diagnosis;
misdiagnosis
Misdiagnosis;
delayed
diagnosis
Author, Year
Diagnostic errors
based on altered
physical examination
findings
Thomas,
(26)
2003
Kokki, 2005
Hewett,
(78)
2010
Diagnostic Error
(77)
Delay in diagnosis or
decrease in
diagnostic accuracy of
physical examination
findings for
appendicitis
Experimental Intervention
Patient or Related
Outcome
Morphine sulfate (MS)
Patient disposition and
administered for pain during ultimate diagnosis (including
diagnostic process
presence and severity of
physical findings)
Notes
Missed diagnosis;
misdiagnosis
Missed diagnosis;
misdiagnosis;
delayed
diagnosis
Missed colorectal
Cap-fitted colonoscopy,
Missed colorectal adenoma
adenoma diagnosis in which allows for flattening of diagnosis in colonoscopy
colonoscopy
haustral folds and/or
improves mucosal exposure
Wellwood,
(104)
1992
D-237
Missed diagnosis;
misdiagnosis;
delayed
diagnosis
Missed diagnosis
Missed diagnosis;
misdiagnosis
Author, Year
Diagnostic Error
Experimental Intervention
Patient or Related
Outcome
Implementation of a
Diagnostic accuracy (proxy
computerized acute cardiac CCU or telemetry unit
ischemia time-insensitive
admission)
predictive instrument (ACITIPI)
Notes
Missed diagnosis
or delayed
diagnosis
(presumed from
appropriate
admissions)
Implementation of an acute
cardiac ischemia predictive
instrument, similar to acute
cardiac ischemia timeinsensitive predictive
instrument (ACI-TIPI),
calculated and delivered in
hardcopy to clinicians
Selker,
(105)
1998
Missed diagnosis of
acute cardiac
ischemia
Pozen,
(117)
1984
Missed diagnosis of
acute cardiac
ischemia
Kuperman,
(28)
1999
Time interval between Computer system to detect Interval from when a critical
laboratory results and critical conditions and notify result was available for
clinical action
the physician
review until appropriate
treatment administered
D-238
Missed diagnosis,
misdiagnosis
Delayed
diagnosis
Author, Year
Bogusevicius,
(27)
2002
Major, 2002
(32)
*Poon,
(106)
2002
(107)
Gur, 2004
Diagnostic Error
Acute mechanical
small bowel
obstruction
Experimental Intervention
Patient or Related
Outcome
Computer-aided diagnosis
Time to diagnosis;
(CAD) and contrast
morbidity, mortality,
radiography for diagnosis of sensitivity, specificity and
acute mechanical small
positive/negative predictive
bowel obstruction
values
Kakeda,
(108)
2004
Cupples,
(109)
2005
Breast cancer
detection rates (from
mammography
screening program)
Unsafe workups
Ramnarayan,
(110)
2006
Fenton,
(111)
2007
Breast cancer
detection rates
Implementation of
computer-aided detection
(CAD) program
Implementation of
computer-aided detection
(CAD) program
Implementation of
computer-aided detection
(CAD) technology to assist
in the interpretation of
mammography
Mortality
Although no patient
outcomes studied (only
usage patterns and users
satisfaction studied) related
to unnecessary delays in
patient care
Diagnostic accuracy of
breast cancer
Notes
Missed diagnosis;
misdiagnosis;
delayed
diagnosis
D-239
Missed diagnosis;
misdiagnosis
Missed diagnosis;
misdiagnosis
Missed diagnosis
Missed diagnosis;
misdiagnosis
Missed diagnosis;
misdiagnosis
Author, Year
(112)
Diagnostic Error
Experimental Intervention
Patient or Related
Outcome
Starting in 2005, SMS text
Time delay between receipt
message notifications with
of clinically significant
patient critical values sent to information and appropriate
clinician. From 2001-, a
course of patient treatment
callback system had been in
place to send patient critical
values to clinicians
Park, 2008
Interval between
results and clinical
action
*Piva,
(113)
2009
*Singh,
(114)
2009
Inadequate
communication of
abnormal cancerscreening test results
in electronic health
records
Misdiagnosis of noninfectious conditions
as cellulitis
David,
(115)
2011
Etchells,
(33)
2011
Although no patient
outcomes studied, timely
physician notification could
have had impact on patient
events
Electronic medical record
Timely follow-up of
alert for positive fecal occult abnormal cancer screening
blood (FOBT) cancer
test results (FOBT) to
screening test results
reduce missed or delayed
diagnoses of colorectal
cancer
Visually-based
Number of patients admitted Other: In 18/28 of misdiagnoses, VCDDSS
to the hospital with an
computerized diagnostic
included the correct diagnosis, while in
incorrect cellulitis diagnosis only 4/28 cases did the physician identify
decision support system
(VCDDSS, also named
the correct diagnosis.
VisualDx) to generate an
improved differential
diagnosis
Real-time clinical alerting
Clinical actions completed
Experimental Design: Based on laboratory
systems for critically
in response to the alerts
values, 50% of potential clinical actions
abnormal laboratory values (that could affect patient
occurred when the alert system was on as
via text messages sent to
outcome) and patients
well as 50% while off. Adverse events
physicians using
adverse events
within 48h were actually higher in cases
alphanumeric pagers or
while alert system was on (42%) than while
smart phones (decision
off (33%) but this difference only
support also provided via
approached significance (p = .06).
smart phones or hospital
intranet)
D-240
Notes
Delayed
diagnosis
Delayed
diagnosis
Missed diagnosis;
delayed
diagnosis
Misdiagnosis
Delayed
diagnosis
Author, Year
Diagnostic Error
Experimental Intervention
Patient or Related
Study Design: Result
Outcome
Fitzgerald,
Errors during
Real time computerRCT: Error free resuscitation in 16% of
Patient morbidity and
(34)
2011
reception and
prompted evidence-based
mortality; including length of baseline controls and 21.8% intervention.
resuscitation of
decision support system
Predicted mortality rate 11%, but actual of
hospital stay
severely injured adult (with decision and action
5.2%, meant insufficient power for
trauma patients
algorithms) during reception
analyzing a true mortality difference
(including errors in the and resuscitation of
statistically. No significant reduction in
diagnosis pathway)
severely injured adults in
sepsis or adult respiratory distress
Level I adult trauma center
syndrome, but aspiration pneumonia was
reduced from 5.3% (control group) to 2.5%
(intervention).
Olsson,
Missed or delayed
Neural network-based
Diagnostic accuracy (proxy Other prospective design: Compared to
(116)
2006
diagnosis of acute
decision making tool added for indicated treatment for
cardiology attending, interns regularly
cardiac ischemia
to ECG results to
ST-segment elevation
treating chest pain patients in the
recommend statistical
myocardial infarction)
emergency department classified 68%
likelihood that results
ECGs indicating ischemia and 92% of
normal ECGs correctly without the decision
indicated thrombolytic
aid. After switching to the decision aid two
agents and
weeks following baseline, the interns rates
revascularization
changed to 93% and 87%, respectively,
with significant increases in sensitivity and
decreases in specificity.
Multiple Intervention Types
Additional Review Methods and Educational Interventions
Interpretation of
Film review process and
Diagnostic accuracy
Seltzer,
Other: Retrospective review; seniority
(61)
radiographs
education sessions with
1981
positively correlated to diagnostic
medical students
accuracy. 80% of abnormalities were
thought to be of clinical importance. First
year residents had an omission rate of
6.1% while second and third year residents
had 4.8%.
Interpretation of
Thomas,
Red star report reminds or
Diagnostic accuracy
Other: Internal audit system; Red star
(64)
radiographs
1992
indicates something
reports issued in 2.8% of cases. 0.7% of
possibly missed or
patients needed to return for follow-up due
incorrectly interpreted.
to incorrect interpretations. Less than 50%
Educational conference
required an alteration to treatment.
held to discuss results of
reports
D-241
Notes
Misdiagnosis;
delayed
diagnosis
Missed diagnosis,
misdiagnosis
Missed diagnosis;
misdiagnosis;
proof of concept
Missed
diagnosis;
misdiagnosis
Author, Year
Diagnostic Error
Experimental Intervention
Patient or Related
Outcome
Additional Review Methods, Educational Interventions, and Structured Process Changes
Kundel,
Diagnostic accuracy
Visual (gaze-duration
Proper interpretation of
(62,63)
1990
of pulmonary nodule threshold algorithm)
chest radiographs
interpretation
feedback to radiologists
based upon eye-position
recordings. Re-review of
radiographs
Additional Review Methods, Educational and Technology-based Systems Interventions
McPhee,
Missed cancer
Cancer screening
Cancer screening test
(62)
1989
diagnosis
reminders, audit with
performance
feedback or control (no
intervention). Half the cohort
was also provided with
educational course; 6
intervention cohorts with
medical residents randomly
assigned.
Additional Review Methods and Personnel Changes
Trotter,
Diagnostic errors from Interpretation of skin
Clinical importance of
(67)
2003
interpretation of skin
biopsies by general
discrepant skin biopsy
biopsies by general
pathologists (vs.
results between general
pathologists (vs.
dermatopathologists)
pathologists and
dermatopathologists)
dermatopathologists
D-242
Notes
Missed diagnosis;
misdiagnosis;
proof of concept
Missed diagnosis;
misdiagnosis;
delayed
diagnosis
Missed diagnosis;
misdiagnosis
Author, Year
Ross, 1996
Goodyear,
(70)
2008
(65)
Diagnostic Error
Incorrect vertebral
fracture diagnoses
Laboratory error
Experimental Intervention
Patient or Related
Outcome
Blinding of X-ray readings to Incorrect vertebral fracture
film sequence and patient
diagnoses (that can impact
identity for the detection of patient management and
vertebral fractures
outcome)
Peldschus,
(68)
2005
Lung lesions/cancer
detection
Missed diagnosis;
misdiagnosis
Missed diagnosis;
misdiagnosis
D-243
Notes
Missed diagnosis;
misdiagnosis;
proof of concept
Missed diagnosis;
misdiagnosis
Author, Year
Diagnostic Error
(9)
Delayed sepsis
detection in surgical
intensive care
Moore, 2009
Experimental Intervention
Patient or Related
Study Design: Result
Notes
Outcome
Mortality as a proxy of
Other Prospective Design, Pre/Post: Of
Delayed
delayed diagnosis of sepsis 4,991 sepsis screens with 920 patients
diagnosis
across 927 admissions to the surgical ICU,
the sepsis early identification tool and
protocol yielded a sensitivity of 96.5%,
specificity of 96.7%, positive predictive
value of 80.2%, and negative predictive
value of 99.5%. Compared to the year
before implementing the sepsis tool,
mortality from severe sepsis and septic
shock decreased from 35.1% to 24.2%.
The authors reported that mortality in the
medical and cardiovascular ICUs did not
decrease notably at the same location
during the study period.
Reduction in clinically
significant errors; late error
detection
D-244
Author, Year
Diagnostic Error
Experimental Intervention
Patient or Related
Outcome
Diagnostic accuracy
D-245
Notes
Missed diagnosis;
misdiagnosis
Missed diagnosis;
misdiagnosis;
delayed
diagnosis
Missed diagnosis;
misdiagnosis
Author, Year
Diagnostic Error
Experimental Intervention
Rollman,
(38)
2002
Ravery,
(82)
2000
(81)
Patient or Related
Outcome
Diagnosis; treatment plan
Diagnostic accuracy of
prostate cancer
Diagnostic accuracy of
prostate cancer
Prostate cancer
detection
Diagnostic accuracy of
prostate cancer
Notes
Overall rate of
misdiagnoses and rate of
serious misdiagnoses
leading to patient recall and
treatment change
Missed diagnosis;
misdiagnosis
Missed diagnosis
Missed diagnosis
D-246
Diagnostic Error
Type of
Intervention
Experimental Intervention
Compared
intervention
Description of
Outcome
Placebo
Wrong Diagnosis
0.22 (0.05-0.98)
0.96 (0.73-1.27)
Diagnostic accuracy
(based on information
from follow-up
visits/hospital
discharges)
Missed adenoma
diagnoses (per
adenomas)
False positives
False negatives
33.33 (2.01-554.09)
Hewett,
(78)
2010
Missed colorectal
adenoma diagnosis in
colonoscopy
McCarthy,
(86)
1990
Incorrect diagnosis by
parents of symptoms of
serious illness
EI
Klassen,
(92)
1993
Missed positive
radiographic findings
(fracture, dislocation or
effusion) after trauma
Misdiagnosis of
appendicitis
SPC
Wellwood,
(104)
1992
TBS
Regular high
resolution
colonoscopy
3-point global
scoring system for
evaluating the
chance of serious
illness
Physicians carrying
out standard
procedures
D-247
0.63 (0.41-0.99)
Author
Bogusevicius,
(27)
2002
Diagnostic Error
Type of
Intervention
Missed acute mechanical TBS
small bowel obstruction
Experimental Intervention
Computer-aided diagnosis for
diagnosis of acute mechanical
small bowel obstruction (SBO)
Compared
intervention
Contrast
radiography
Description of
Outcome
False positives for
complete SBO
False negatives for
complete SBO
False positives for
partial SBO
False negatives for
SBO
D-248
Patients with
radiographs ordered
0.94 (0.75-1.18)
Author
McPhee,
(62)
1989
Diagnostic Error
Type of
Intervention
Missed cancer diagnosis ARM, EI, and
TBS
Experimental Intervention
Computer generated list of
overdue tests at patients visits
(cancer screening reminders)
Compared
intervention
No intervention
No intervention
Patient education
No intervention
Placebo
D-249
Description of
Outcome
Further cancer
screening (Results
given as postintervention compliance
scores relative to
standards according to
the American Cancer
Society
recommendations)
Further cancer
screening (results given
as post-intervention
compliance scores
relative to standards
according to the
American Cancer
Society
recommendations)
Further cancer
screening (results given
as postintervention
compliance scores
relative to standards
according to the
American Cancer
Society
recommendations)
significant
Inappropriate
management (surgery
or patient observation)
Admissions for
observation or
discharge home
Surgeries
Repeat physician visit
for abdominal pain
within 7 days
Possible incorrect
surgical management
0.22 (0.05-0.98)
Statistically
significant
Statistically
significant**
P = 0.50
P = 0.51
2.84 (0.31-26.08)
2.84 (0.31-26.08)
Author
Diagnostic Error
Kuperman,
(28)
1999
Delays between
laboratory results and
clinical action
(37)
Sakr, 1999
Wellwood,
(104)
1992
Rollman,
(38)
2002
Type of
Intervention
TBS
Clinically important
PC
errors, including errors in
the diagnosis pathway
(i.e., history, physical
examination, and
radiographic
interpretation errors)
Misdiagnosis of
TBS
appendicitis
Missed depression
diagnosis
Experimental Intervention
Compared
intervention
Computer system to detect
No automatic
critical laboratory conditions and notification for alerts
notify the physician via Hospitals
paging system
Use of nurse practitioner in
Use of Junior
providing care in the emergency Doctors in the
department
emergency
department
Description of
Outcome
Time to appropriate
treatment
Unplanned follow-up
visits
0.65 (0.45-0.96)
Admissions
Surgeries
0.91 (0.84-0.99)
0.98 (0.82-1.16)
1.01 (0.91-1.12)
1.09 (0.90-1.32)
0.95 (0.49-1.87)
No diagnostic aid
D-250
1.01 (0.64-1.59)
0.95 (0.49-1.87)
1.19 (0.63-2.25)
0.75 (0.44-1.25)
1.19 (0.63-2.25)
Author
Diagnostic Error
Type of
Intervention
Patient Outcomes
(37)
Sakr, 1999
Clinically important
PC
errors, including errors in
the diagnosis pathway
(i.e., history, physical
examination, and
radiographic
interpretation errors)
Bogusevicius, Missed acute mechanical TBS
(27)
2002
small bowel obstruction
Fitzgerald,
(34)
2011
Kuperman,
(28)
1999
Delays between
laboratory results and
clinical action
TBS
TBS
Experimental Intervention
Compared
intervention
Description of
Outcome
0.94 (0.68-1.30)
Contrast
radiography
Mortality
Morbidity outcome
5 (0.25-100.97)
1.33 (0.32-5.58)
Control (without
computer-aided
decision support
system)
Error rate
Morbidity from shock
management
Aspiration pneumonia
Sepsis
0.89 (0.79-1.00)
P = 0.03
P = 0.046
Not statistically
significant
Not statistically
significant
ARDS (acute
respiratory distress
syndrome)
Functional
Not statistically
independence measure significant
score
Hospital length of stay Not statistically
significant
P < 0.001
Transfusion of blood
productions
Mortality
1.15 (0.65-2.03)
P = 0.11
Computer system to detect
No automatic
Time to resolution of
critical laboratory conditions and notification for alerts alerting conditions
notify the physician via hospitals
P = 0.41
Adverse events
paging system
D-251
Author
Diagnostic Error
Rollman,
(38)
2002
Missed depression
diagnosis
Type of
Intervention
SPC
Experimental Intervention
Active care: PCPs were exposed
to advisory messages on the
paper encounter-based upon
AHCPRs guidelines AND advise
to click on the computer desk top
icon to obtain further treatment
advise from the EMR intranet site
Compared
intervention
Passive care: PCPs
provided with a
reminder of their
patients depression
diagnosis on the
paper encounter
form to treat
depressive
episodes, but
offered no details on
how to do so
Usual care
Description of
Outcome
Nonimprovement of
depressive symptoms
Nonimprovement of
depressive symptoms
0.88 (0.65-1.19)
Nonimprovement of
depressive symptoms
0.93 (0.70-1.25)
D-252
PRIME-MD report
not given to the
Physician
Author
Diagnostic Error
Type of
Intervention
TBS
Experimental Intervention
Compared
intervention
Computer system to detect
No automatic
critical laboratory conditions and notification for alerts
notify the physician via hospitals
paging system
Description of
Effect Size (95%
Outcome
CI)*
Delays between
Kuperman,
TUO+PO: Adverse
1.20 (0.78-1.84)
(28)
laboratory results and
1999
events
clinical action
(cardiopulmonary
arrest, ICU admissions,
strokes, acute renal
failure, death, need for
surgery)
Abbreviations: AHCPR = Agency for Health Care Policy and Research; ARM = additional review methods; DAO = diagnostic accuracy outcome; EI = educational intervention;
EMR = electronic medical record; ICU: intensive care unit; nss = not statistically significant; PC = personnel change; PCP = primary care physician; PO = patient outcomes;
PRIME-MD: Primary Care Evaluation of Mental Disorders; SBO = small bowel obstruction; SPC = structured process change; ss = statistically significant; T= technique; TBS =
technology-based systems intervention; TUO = therapeutic use outcome.
*Effect size is relative risk except for Fitzgerald et al. where error rate was used; McPhee et al,where difference in scores post intervention was used and Kuperman et al.al where
time to appropriate treatment was used.
Results were significant for: stool occult blood testing, rectal examination, sigmoidoscopy, pelvic exam, breast exam, mammography AND non-significant for Pap smear.
Results were significant for: breast exam, mammography AND non-significant for: occult blood test, rectal exam, sigmoidoscopy, Pap smear, pelvic exam
**Results were significant for breast exam AND non-significant for mammography.
D-253
References
1.
2.
3.
4.
5.
6.
D-254
7.
8.
9.
10.
11.
12.
13.
21.
14.
22.
15.
23.
24.
25.
26.
27.
16.
17.
18.
19.
20.
D-255
28.
29.
30.
31.
32.
33.
34.
D-256
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
48.
49.
Carew-McColl M. Radiological
interpretation in an accident and emergency
department. Br J Clin Pract. 1983;37(1112):375-7. Epub 1983/11/01. PubMed
PMID: 6671078.
50.
51.
52.
Kwek BH, Lau TN, Ng FC, Gao F. Nonconsensual double reading in the Singapore
Breast Screening Project: benefits and
limitations. Annals of the Academy of
Medicine, Singapore. 2003;32(4):438-41.
Epub 2003/09/13. PubMed PMID:
12968545.
53.
54.
55.
46.
47.
D-257
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.
D-258
70.
71.
72.
73.
74.
75.
D-259
76.
77.
78.
79.
80.
81.
82.
83.
84.
85.
86.
87.
88.
D-260
89.
90.
91.
92.
93.
94.
95.
96.
97.
98.
99.
100.
101.
D-261
102.
103.
104.
105.
106.
107.
108.
109.
110.
111.
112.
113.
114.
D-262
115.
116.
117.
118.
119.
120.
121.
122.
D-263
Evidence Tables for Chapter 36. Monitoring Patient Safety Problems (NEW)
The layout of this evidence table is customized based on the data reported by the included studies. Some columns in the evidence
tables for other PSP topics are not included in this table or merged with other columns. For example, the Description of
Organization column is merged into the Context column. There is no Theory or Logic Model column in this table because none
of the included studies reported such data. The customized layout allows the data collected to fit into the table appropriately.
Table 1, Chapter 36. Evidence from studies comparing methods for detecting patient safety problems
Author/
Year
Olsen
1
2007
Description of PSP
Study Design
Contexts
This is a prospective
observational study. Data on AEs
were collected on 288 patients
discharged from adult medical
and surgical units in an acute care
hospital
D-264
Author/
Year
Wetzels
2
2008
Description of PSP
adverse event was determined for each case.
Each event was classified according to the stage of
care and a mutually exclusive problem category
(diagnosis, overall assessment of patients
condition including comorbidities, technical
problems occurring during a procedure, infectionrelated, general problems with ongoing monitoring
and management of patients and medicationrelated problems). Record review was also carried
out by members of the clinical team caring for the
patients. But in this report only the data collected by
the external assessors were used.
Five methods for identifying adverse events in
general practice:
1) Physician reported adverse events
The physicians recorded all events using a
simplified computerized registration form based on
an existing international taxonomy for errors in
general practice. The physicians registered event
date, birth date of patient, gender, event category
(practice administration (archive; medical record;
appointment; other), diagnostic (wrong diagnosis;
delayed diagnosis; missed diagnosis; other),
therapeutic (wrong, incomplete; delayed; none,
though it should be; other), communication (with
patients; with caregivers; other), and additional
remarks and/or context.
2) Pharmacist reported adverse events
The pharmacist recorded events from her point of
view using an adjusted form developed for this
purpose. Event date, birth date of patient, gender,
practice, event category (prescribing error (wrong
prescription, wrong administration; wrong dose;
other), adverse reaction (adverse reaction; allergic
reaction; overdose; interaction; contra-indication;
other), dispensing error (too late; wrong medicine;
wrong dose; other), and additional remarks or
context were recorded.
3) Patients experiences of adverse events
In the waiting room of the two practices samples of
50 patients, consecutively visiting the practice,
were invited to complete a questionnaire on
Study Design
A prospective observational
study, comparing the five
methods in two general practices
in a period of five months (May to
October 2006)
A total of approximately
8,250 patients were registered
with the two practices
D-265
Contexts
Author/
Year
Ferranti
3
2008
Description of PSP
experienced problems with safety of their health
care in the previous six months. A drop box was
used to collect the completed questionnaires.
Questions were derived from items of the Medical
Harvard Study, and from questions of two survey
studies. The questionnaire guaranteed anonymity
of participating patients.
4) Assessment of a random sample of medical
records
Thirty medical records per physician were randomly
selected from patients who had visited their general
practice in the observation period. Anonymous
medical records, containing the information from
this period were printed out. Two clinical
researchers examined the information
independently. They scrutinized the records for
indications of events and, when found, categorized
the event (errors in office administration, diagnosis,
treatment or communication with their
subcategories); and added demographic data of the
patient. Subsequently the physicians discussed
their findings and reached consensus.
5) Assessment of all deceased patients
One physician examined the medical records of the
patients who had died in the period of the study for
events. The same registration form and analysis
procedure as for the audits of medical records was
used.
Two ADE detection systems were studied:
1) Voluntary reporting
The safety reporting system was developed as a
home-grown web application to provide a single
point of entry for voluntary reporting and allow
standardized evaluation of safety events across
Duke University Health System (DUHS). All DUHS
employees may access the reporting system and
are encouraged to report any safety events
witnessed, including near misses. Although
anonymous reporting is possible, DUHS policy
supports a non-punitive culture of safety. Safety
reporting system captures a myriad of event types
including medication/intravenous-related, blood
Study Design
Contexts
Computerized surveillance
detected 710 ADEs (6.93/1,000
patient days), whereas voluntary
reporting identified 205 ADEs
(1.96/1,000 patient days). For
each major drug category
(anticoagulants, hypoglycemia,
narcotics and benzodiazepines,
and miscellaneous), surveillance
and voluntary reporting detected
significantly different event rates.
Most surveillance events were
hypoglycemia-related, whereas
most voluntarily-reported events
D-266
Author/
Year
Levinson
4
2010
Description of PSP
transfusions, surgical, falls, treatment/testing,
dissatisfied patient, and others. Each
medication/intravenous-related report was
investigated by a team of 4 medication safety
pharmacists and scored for severity before
submission to a multidisciplinary leadership team
for review and confirmation. All events with a
severity score were deemed adverse drug events
(ADEs).
2) Computerized surveillance
The DUHs computerized ADE-S system was
deployed by an internal team of technical and
safety experts. Each evening, ADE-S evaluates
medication, laboratory, and patient demographic
information against a set of clinical rules or triggers
to detect potential ADEs or evolving unsafe
conditions. Nearly 130 rules have been deployed
since the systems inception, but only 14 high-risk
rules with high true-positive rates were considered
in surveillance. These 14 rules span 3 main
categories: abnormal laboratory results, use of
antidotes, and drug-lab combinations. Adverse drug
event surveillance delivers an electronic daily report
to a web-based surveillance application that details
all triggers fired by the system. This list was
evaluated by 3 clinical pharmacists who perform a
chart review to determine whether an ADE
occurred. Pharmacists identified all possible
medications involved in the event and assigned a
causality score using the Naranjo algorithm and a
severity score using the DUH 7-point scale. All
events scored with causality Q5 and a severity Q3
were considered ADEs. Pair wise inter-rater
reliability scores (J statistic) exceeded 0.88 for each
rater pair.
The following safety problem monitoring methods
were assessed:
1) Nurse Reviews
Contracted registered nurses reviewed medical
records for each sampled 278 Medicare
beneficiarys hospitalization. Nurses used a
standardized review process developed by the
Study Design
D-267
Contexts
Author/
Year
Description of PSP
Study Design
Contexts
D-268
Author/
Year
LevtzionKorach
5
2010
Description of PSP
Healthcare Research and Qualitys (AHRQs)
Patient Safety Indicator (PSI) software program to
hospital administrative billing data for the 278
sample Medicare beneficiary hospitalizations.
AHRQ developed the PSI software to monitor
health care quality using administrative data, such
as patient demographics (e.g., age, gender), and
diagnoses and procedure codes. The PSI software
is based upon a series of algorithms that detect 20
provider-level complications that indicate possible
events (e.g., death of a low-risk patient).
The following safety problem monitoring methods
were assessed:
1) An incident reporting system
The hospital used a commercially available Webbased incident reporting system. Hospital
personnel could report confidentially through any
hospital computer using a secure login and could
report anything that they perceive might be an
issue. Each adverse event report contains the
reporters initial comments and a section for the
departmental manager to clarify issues further and
add comments and actions. The manager is
responsible for reviewing each report and assigning
one or more contributing factors from a drop-down
list of 50 potential contributing factors. For the most
important reports, management will have direct
conversation with the reporters after the evaluation
is complete
2) Reports to hospital risk management
A nurse-lawyer leads the risk management team.
Physicians and nurses, in about equal numbers,
call the team to report adverse events and poor
patient outcomes. Risk management staff members
investigate each case and determine on the basis
of the estimated risk whether to report the case to
the malpractice carrier. This information is collected
manually with no systematic categorization and is
entered in an electronic index. Risk management
also provides information back to managers or
frontline individuals so that risks can be mitigated.
3) A patient complaints database
Study Design
Contexts
This is a prospective
observational study, comparing
the five safety problem detection
methods. Data were collected for
a 22-month period from
May 10, 2004, to
February 28, 2006.
8,616 incident reports (involving
13,255 contributing factors),
1,003 risk management reports,
4,722 patient complaints
(involving 6,617 specific
problems), 61 walk rounds
(involving 572 comments), and
322 malpractice claims (involving
949 issues) were evaluated.
D-269
Author/
Year
Description of PSP
Study Design
D-270
Contexts
system.
Equipment (15.7%),
electronic records (12.2%), and
environment/infrastructure
(12.1%) were the leading
categories in executive walk
rounds but were ranked low in the
other systems.
In the incident reports,
identification issues (24.4%) and
falls (16.8%) were the leading
categories but were barely
represented in the other systems.
Overall, there is a low level of
consistency across the five
methods. The highest correlations
between the different categories
across the methods were
between malpractice claims,
reports to risk management, and
patient complaints. The adverse
event reporting system and
executive walk rounds had low
and negative correlation with the
other four systems.
Author/
Year
Tinoco
6
2011
Description of PSP
records, adjustor notes). Nurse coders assign one
or more (from 170) risk management issues,
factors that may have contributed to the allegation,
injury, or initiation of the claim/suit. There are clear
definitions, standardized coding algorithms, and
collaboration between coders leading to high interrater reliability. The data are stored in an electronic
database that is available for querying, analysis,
and generation of reports. There are about
30 claims per year.
Two methods for detecting inpatient adverse drug
events (ADEs) and hospital-associated infections
(HAIs) were studied: A computerized surveillance
system (CSS) or manual chart review (MCR)
For CCS, the HELP (Health Evaluation through
Logical Processing) system was used. This
electronic system manages billing and
administrative codes for each hospital admission,
as well as information from several clinical
domains: admission, discharge, and transfer
(ADT)/registration, pharmacy, laboratory,
microbiology, nurse charting, and physician
narratives, etc. The physician narratives stored in
the HELP system as freetext documents include
emergency department report, admission history
and physical report, consultant note, radiology
report, surgical procedure note, and discharge
summary.
The HAI detection criteria used by CSS were
originally based on the guidelines from the Study of
the Efficacy of Nosocomial Infection Control and
the Centers for Disease Control and Prevention
(CDC). In addition to routine HAI surveillance, daily
urine samples from all catheterized patients were
obtained as part of an existing, hospital-wide
urinary catheter surveillance program. The ADE
detection criteria used by CSS include various
clinical triggers such as medication discontinuation
orders, dose decrease orders, antidote orders,
laboratory test orders, abnormal laboratory test
results and vital signs. Suspected cases are
flagged by CSS and reported to surveillance
Study Design
Contexts
D-271
Author/
Year
Description of PSP
Study Design
Contexts
D-272
References
1.
2.
3.
D-273
4.
5.
6.
Study, Year
Adler et al, 2009
Intervention
Readmission
144
156
CCT
General medical
Geriatric/General
medical
44
45
RCT
General medical
49
47
RCT
124
119
CCT
General medical
Geriatric/General
medical
Geriatric/Mixed
Diagnoses
Geriatric/Mixed
Diagnoses
Geriatric/General
medical
83
83
1235
158
371
379
64
64
626
581
43
30
CCT
RCT
RCT
CCT
General medical
229
243
RCT
General medical
186
175
RCT
Geriatric
Geriatric/General
medical
200
200
201
199
General medical
Mixed patient
population
59
18
315
316
117
115
66
66
376
373
13
14
RCT
15
CCT
16
17
CCT
RCT
11
12
10
General medical
Geriatric/General
medical
Costs
Control
CCT
Adverse Event
Acute care/ED
utilization
Patient Population
18
RCT
CCT
CCT
RCT
Geriatric
Geriatric/General
medical
General medical and
surgical
D-274
Sample Size
Study
Design
Study, Year
20
21
Intervention
Readmission
21
20
314
340
Geriatric
Mixed patient
population
Geriatric/General
medical
356
350
231
221
25
29
746
792
186
177
20
20
181
181
RCT
General medical
Geriatric/Mixed
Diagnoses
Geriatric/General
medical
Geriatric/General
medical
Geriatric/General
medical
185
181
CCT
General medical
1014
1248
RCT
Geriatric/Mix
Diagnoses
49
49
RCT
General medical
250
250
RCT
22
RCT
23
CCT
RCT
25
CCT
26
RCT
27
RCT
Control
RCT
Naylor, 1990
Patient Population
Geriatric/General
medical
Geriatric/General
medical
28
29
30
31
32
RCT
Adverse Event
Costs
General medical
84
92
RCT
General medical
391
371
35
CCT
84
749
176
178
CCT
General medical
Geriatric/General
medical
Geriatric/General
medical
469
355
RCT
Mixed Diagnoses
453
166+
190 (2
groups)
453
43 (0-4 day
stay)+ 137 (5+
day stay)
64
68
36
RCT
37
38
39
CCT
40
RCT
Geriatric/General
medical
Geriatric/General
medical
D-275
RCT
34
33
Acute care/ED
utilization
Sample Size
Study, Year
Voss et al, 2011
41
42
43
Study
Design
Patient Population
Control
Intervention
Readmission
CCT
General medical
736
257
CCT
General medical
366
396
695
D-276
Adverse Event
Costs
Acute care/ED
utilization
Study, Year
1
Adler et al, 2009
2
Al-Rashed et al, 2002
3
Balaban et al, 2008
4
Bolas et al, 2004
5
Brand et al, 2004
6
Coleman et al, 2004
7
Coleman et al, 2006
8
Courtney et al, 2009
9
Cowan et al, 2006
10
Dellasega et al, 2000
11
Einstadler et al, 1996
12
Forster et al, 2005
13
Gallagher et al, 2011
14
Gillespie et al, 2009
15
Gow et al, 1999
16
Graumlich et al, 2009
17
Hellstrom et al, 2011
Hogan and Fox,
18
1990
19
Jack et al, 2009
20
Koehler et al, 2009
21
Lim et al, 2003
22
Lipton and Bird, 1994
Makowsky et al,
23
2009
24
Martin et al, 1994
25
Mudge et al, 2006
26
Naylor et al, 1999
27
Naylor, 1990
28
Nazareth et al, 2001
29
Nikolaus et al, 1999
30
Palmer et al, 2002
31
Parry et al, 2009
32
Schillig et al, 2011
33
Schnipper et al, 2006
34
Scullin et al, 2007
Study
Design
CCT
CCT
RCT
RCT
CCT
CCT
RCT
RCT
CCT
RCT
CCT
RCT
RCT
RCT
CCT
RCT
CCT
Allocation
Sequence
Random?
No
Unclear
Unclear
Yes
Unclear
No
Yes
Yes
No
Yes
No
Yes
Yes
Yes
No
Unclear
No
Allocation
Concealed?
No
Unclear
Unclear
Unclear
Unclear
No
Unclear
Yes
Yes
Yes
No
Yes
Yes
Yes
Unclear
No
Unclear
Baseline
Outcomes
Similar?
Unclear
Unclear
Unclear
Unclear
Unclear
Unclear
Unclear
Unclear
Unclear
Yes
Unclear
Yes
Unclear
Unclear
Unclear
Unclear
Unclear
Baseline
Characteristics
Similar?
No
Unclear
Yes
Yes
Unclear
No
No
Yes
No
Yes
Yes
Yes
Yes
Unclear
Unclear
Unclear
Yes
Incomplete
Data
Addressed?
Unclear
Unclear
Yes
Unclear
Unclear
Unclear
Unclear
Unclear
Unclear
Yes
Unclear
Yes
No
Unclear
Unclear
Unclear
Yes
Outcomes
Assessed Blind
to Intervention?
Unclear
Unclear
Unclear
Yes
Unclear
Unclear
Yes
Yes
Yes
Yes
Unclear
Yes
Yes
Yes
Unclear
Yes
Unclear
No
Contamination?
Unclear
Unclear
Unclear
Unclear
Yes
Unclear
Yes
Unclear
Yes
Yes
Unclear
Yes
Unclear
Unclear
Unclear
Yes
Unclear
Free of
Selective
Outcome
Reporting
Risk?
Unclear
Unclear
Unclear
Unclear
Unclear
Yes
Yes
Unclear
Yes
Yes
Yes
Yes
Unclear
Yes
No
Yes
Yes
CCT
RCT
RCT
RCT
RCT
No
Yes
Yes
Yes
Yes
No
Yes
Yes
Yes
Yes
Unclear
Unclear
Unclear
Unclear
Yes
Unclear
Unclear
Yes
Yes
Yes
Yes
Unclear
Unclear
Yes
Yes
Unclear
Yes
Unclear
Yes
Yes
Unclear
Unclear
Yes
Unclear
Yes
Yes
Unclear
Unclear
Yes
Yes
No
Unclear
Unclear
Yes
Yes
CCT
RCT
CCT
RCT
RCT
RCT
RCT
CCT
RCT
RCT
RCT
RCT
No
Yes
No
Yes
Yes
Yes
Yes
No
Yes
No
Yes
Yes
No
Unclear
No
Yes
Unclear
Unclear
Yes
No
No
No
Yes
Yes
Unclear
Unclear
Unclear
Unclear
Unclear
Yes
Unclear
Unclear
Unclear
Unclear
Unclear
Unclear
Unclear
No
Unclear
Yes
Unclear
Unclear
Unclear
Unclear
Yes
No
Yes
Yes
Unclear
Yes
Unclear
Yes
Unclear
Unclear
Unclear
Unclear
Unclear
Unclear
Unclear
Unclear
Unclear
Unclear
Yes
Yes
Unclear
Yes
Yes
Unclear
Yes
No
Yes
Yes
Yes
Yes
Unclear
Unclear
Unclear
Unclear
Unclear
Unclear
Yes
Unclear
Yes
Unclear
Unclear
Yes
Unclear
Yes
Yes
Yes
Yes
Unclear
Yes
Yes
Yes
Yes
Unclear
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
No
No
Yes
D-277
Free from
other
Bias?
No
Yes
Yes
Yes
Unclear
Unclear
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
No
Unclear
Unclear
Study, Year
35
Scullin et al, 2011
36
Siu et al, 1996
37
Steeman et al, 2006
38
Stewart et al, 1998
39
Styrborn et al, 1995
40
Thomas et al, 1993
41
Voss et al, 2011
42
Walker et al, 2009
Weinberger et al,
43
1996
Study
Design
CCT
RCT
CCT
RCT
CCT
RCT
CCT
CCT
Allocation
Sequence
Random?
No
Yes
Unclear
Yes
No
Yes
No
No
Allocation
Concealed?
Unclear
Unclear
Unclear
Unclear
No
Unclear
No
No
Baseline
Outcomes
Similar?
Unclear
Unclear
Unclear
Yes
No
Unclear
Yes
Unclear
Baseline
Characteristics
Similar?
Unclear
Unclear
Yes
Unclear
No
Yes
No
Unclear
Incomplete
Data
Addressed?
Unclear
Unclear
Unclear
Unclear
No
Yes
Unclear
Unclear
Outcomes
Assessed Blind
to Intervention?
Yes
Yes
Yes
Unclear
Unclear
Yes
Yes
Unclear
No
Contamination?
Unclear
No
Unclear
Yes
Yes
Unclear
Yes
Yes
Free of
Selective
Outcome
Reporting
Risk?
Yes
Yes
Unclear
Unclear
Unclear
Yes
Unclear
Yes
RCT
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Unclear
D-278
Free from
other
Bias?
Unclear
No
Unclear
Yes
No
No
Yes
Yes
Yes
Source
Adler et al,
1
2009
Al-Rashed et
2
al, 2002
Balaban et al,
3
2008
Bolas et al,
4
2004
Brand et al,
5
2004
Coleman et
6
al, 2004
Coleman et
7
al, 2006
Courtney et
8
al, 2009
Cowan et al,
9
2006
Dellasega et
10
al, 2000
Einstadler et
11
al, 1996
Forster et al,
12
2005
Gallagher et
13
al, 2011
Gillespie et al,
14
2009
Gow et al,
15
1999
Graumlich et
16
al, 2009
Hellstrom et
17
al, 2011
Hogan and
18
Fox, 1990
Risk
Stratification
Individualized Patient
Record
MultiPatient
Facilitation of disciplinary
Engagement Communication
Team
Dedicated
Discharge
Advocate
Medication
Reconciliation
Post-discharge Interventions
Facilitated
Clinical
Outreach
Medication
Follow-up to Patients Reconciliation
D-279
Pre-discharge Interventions
Source
Jack et al,
19
2009
Koehler et al,
20
2009
Lim et al,
21
2003
Lipton and
22
Bird, 1994
Makowsky et
23
al, 2009
Martin et al,
24
1994
Mudge et al,
25
2006
Naylor et al,
26
1999
Naylor,
27
1990
Nazareth et
28
al, 2001
Nikolaus et al,
29
1999
Palmer et al,
30
2002
Parry et al,
31
2009
Schillig et al,
32
2011
Schnipper et
33
al, 2006
Scullin et al,
34
2007
Scullin et al,
35
2011
Siu et al,
36
1996
Steeman et
37
al, 2006
Stewart et al,
38
1998
Risk
Stratification
Individualized Patient
Record
MultiPatient
Facilitation of disciplinary
Engagement Communication
Team
Dedicated
Discharge
Advocate
Medication
Reconciliation
Post-discharge Interventions
Facilitated
Clinical
Outreach
Medication
Follow-up to Patients Reconciliation
D-280
Pre-discharge Interventions
Source
Styrborn et al,
39
1995
Thomas et al,
40
1993
Voss et al,
41
2011
Walker et al,
42
2009
Weinberger et
43
al, 1996
Risk
Stratification
Individualized Patient
Record
MultiPatient
Facilitation of disciplinary
Engagement Communication
Team
Dedicated
Discharge
Advocate
Medication
Reconciliation
Post-discharge Interventions
Facilitated
Clinical
Outreach
Medication
Follow-up to Patients Reconciliation
D-281
Study
Design
Bridging
Intervent
ion (Y/N)
Total
number of
intervention
s
30-day
readmiss
ion rate control
group
30-day
readmissi
on rate interventi
on group
ARR - 30-day
readmissions
Al-Rashed et
2
al, 2002
CCT
29.5%*
11.1%*
18.4%
Balaban et al,
3
2008
Coleman et al,
6
2004
RCT
8.2%
8.5%
-0.3%
2.0%
2.1%
-0.1%
CCT
13.8%
8.9%
4.9%
14.2%
11.0%
3.2%
Coleman et al,
7
2006
RCT
11.9%
8.3%
3.6%
Courtney et al,
8
2009
RCT
~15%,
only in
graphics
~5%, only
in
graphics
Dellasega and
10
Zerbe, 2000
Einstadler et
11
al, 1996
Forster et al,
12
2005
Gow et al,
15
1999
Jack et al,
19
2009
RCT
NR
NR
CCT
14.4%
19.8%
-5.4%
19.2%
21.8%
-2.6%
RCT
13.0%
19.4%
-6.4%
7.8%
9.6%
-1.8%
CCT
15.2%
16.7%
-1.5%
RCT
20.7%
14.9%
5.8%
D-282
30-day
ED visits
- control
group
30-day ED
visits intervention
group
ARR - 30day ED
visits
Statistically
significant
ARR (for
either
readmits or
ED visits)
Y
Comment
*Outcomes
measured at 15-22
days post-discharge
Significant
difference for
readmissions, but
not ED visits
te
kkappears only in
graphical format;
p<0.05 for
comparison with
control for
Data presented only
in graphical format;
significant
improvement for ED
visits
Exact rates not
supplied
N
24.5%
16.5%
8.0%
Significant
difference for ED
visits only
Author, Year
Study
Design
Bridging
Intervent
ion (Y/N)
Total
number of
intervention
s
30-day
readmiss
ion rate control
group
30-day
readmissi
on rate interventi
on group
Koehler et al,
20
2009
RCT
38.1%
10.0%
-28.1%
11.2%
15.4%
-4.2%
5.8%
5.5%
0.3%
Lipton and
21
Bird, 1994
Palmer et al,
30
2002
Parry et al,
31
2009
Schnipper et
33
al, 2006
Steeman et al,
37
2006
41
Voss, 2011
Walker et al,
42
2009
ARR - 30-day
readmissions
30-day
ED visits
- control
group
30-day ED
visits intervention
group
ARR - 30day ED
visits
Statistically
significant
ARR (for
either
readmits or
ED visits)
Y
Comment
Composite
outcome of ED
visits and
readmissions
i.e. 284 CCT.284
Control 284
Intervention 284
General Medical
284 Mixed 284
RCT
CCT
Y
16.7%
6.8%
9.9%
RCT
RCT
30%
30%
0%
CCT
5.1%
2.8%
2.3%
CCT
CCT
Y
Y
6
7
18.6%
18.0%
12.8%
22.1%
5.8%
-4.1%
D-283
12.3%
9.5%
2.8%
Y
N
Composite
outcome of total ED
visits and
readmissions
Outcomes
measured at 14
days post discharge
Table 5, Chapter 37. Studies reporting adverse events (including adverse drug events)
Study and
Country
Study
Design
Population
Sample
Size
Intervention
Control
Balaban et al.,
3
2008
US
RCT
General
medical
(Safety net)
96
Discharge-transfer
intervention
Forster et al.,
12
2005
Canada
Gallagher et al.,
13
2011
Ireland
RCT
General
medical
361
RCT
Geriatric
400
Gillespie et al.,
14
2009
Sweden
Graumlich et al.,
16
2009
US
Hellstrom et al.,
17
2011 *
Sweden
RCT
Geriatric/
General
Medical
Mixed Patient
Population
400
CCT
Geriatric
210
RCT
Geriatric/
General
medical
General
Medical/
Mixed (on
warfarin
therapy)
40
Naylor, 1990
27
Schillig et al.,
32
2011
US
US
RCT
RCT
631
500
ARR (95%
Cl)
Usual care
Discharge coordination
led by nurse specialist
Usual care
Inpatient medication
screening using
validated criteria
(STOPP/START)
Pharmacist discharge
counseling and postdischarge follow-up
Software-assisted
discharge
Usual care
Usual care
12 months
1 month
Usual care
All post-discharge
adverse events (7.3
vs. 7.3%, N/A)
-
Pharmacist-led
systematic medication
reconciliation and
review
Comprehensive
discharge planning led
by nurse specialist
Pharmacist-directed
anticoagulation service
Medication-related
readmissions (24.2 vs.
4.8%, NR)
Probable adverse drug
event (5.4 vs. 5.4%,
N/A)
Composite of
medication-related
admissions & ED visits
(12.0 vs. 5.6%, 0.138)
-
3 months
Post-discharge
infection rates (50 vs.
33.3%, NR)
-
12 weeks
Inpatient &
30 days
(combined)
Usual care
Usual care
Usual care
Composite endpoint
(14.8 vs. 10.0%, 0.104)
INR >5 (14.8 vs. 9.6%,
0.076)
Major bleeding (0.8 vs.
0.4%, 0.563)
Thrombosis (0 vs. 0%,
N/A)
D-284
6.4% (-1.214.8)
Other Adverse
ARR (95%
Events (Control vs.
Cl)
intervention rates, pvalue)
Failure to complete
recommended
outpatient work-up
(31.3 vs. 11.5%, 0.11)
Post-discharge
adverse events (22.8
vs. 23.6%, 0.87)
Falls (8.4 vs. 5.8%,
0.332)
Timing
NR
30 days
6 months
Study and
Country
Schnipper et al.,
33
2006
US
Study
Design
Population
Sample
Size
Intervention
Control
RCT
General
medical
176
Pharmacist discharge
counseling and postdischarge follow-up
Usual care
*This was a nonrandomized controlled trial. All other studies are randomized controlled trials.
D-285
ARR (95%
Cl)
Other Adverse
ARR (95%
Events (Control vs.
Cl)
intervention rates, pvalue)
-
Timing
30 days
2.
3.
4.
5.
6.
7.
8.
9.
D-286
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
28.
20.
29.
30.
31.
32.
33.
34.
35.
36.
37.
21.
22.
23.
24.
25.
26.
27.
D-287
38.
39.
40.
41.
42.
43.
D-288
D-289