Beyond FMEA: the structured what-if technique (SWIFT)

Risk Management Beyond FMEA: The structured what-if technique (SWIFT) By Alan J. Card, MPH, CPH, CPHQ James R. Ward, BEng, CEng, PhD, MIET, and P. John Clarkson, PhD, BA(Eng) Although it is probably the best-known prospective hazard analysis (PHA) tool, failure mode and effects analysis (FMEA) is far from the only option available. This article introduces one of the alternatives: The structured what-if technique (SWIFT). SWIFT is a flexible, highlevel risk identification technique that can be used on a stand-alone basis, or as part of a staged approach to make more efficient use of bottom-up methods like FMEA. In this article we describe the method, assess the evidence related to its use in healthcare with the use of a systematic literature review, and suggest ways in which it could be better adapted for use in the healthcare industry. Based on the limited evidence available, it appears that healthcare workers find it easy to learn, easy to use, and credible. Especially when used as part of a staged approach, SWIFT appears capable of playing a useful role as component of the PHA armamentarium. BA C K GR O U N D Patient safety risk management has traditionally been retrospective in nature, aimed at solving problems that have already occurred. Root cause analysis is perhaps the best-known example of this approach. There is broad consensus in the wider risk management community that waiting for something to go wrong before developing preventive measures is not enough, and since the mid-1960s a number of prospective hazard analysis (PHA) techniques have been developed in other high-risk industries. It wasn’t until the 1990s, however, that these techniques began to make their way into the healthcare sector, and it was not until 2001 that the Joint Commission began requiring their use by organizations they accredit.(1) One of the most widely used formal PHA techniques in healthcare is failure mode and effects analysis (FMEA),(2) and its siblings, healthcare failure mode and effects analysis (HFMEA)(3) and failure mode, effects, and criticality analysis (FMECA). (4) But, although FMEA may be the first method to come to mind when many healthcare risk managers consider PHA, it is far from the only method available. Through its Patient Safety Research Portfolio, the UK Department of Health funded a research project that led to the development of a prospective hazard analysis (PHA) tool kit to support the integration of PHA techniques into healthcare risk management.(5) The PHA tool kit includes a new method—the © 2012 American Society for Healthcare Risk Management of the American Hospital Association Published online in Wiley Online Library (wileyonlinelibrary.com) • DOI: 10.1002/jhrm.20101 AMERICAN SOCIETY FOR HEALTHCARE RISK MANAGEMENT • VOLUME 31, NUMBER 4 23 Exhibit 1: Pre-existing PHA and Supporting Methods Included in the PHA Tool Kit.(5) Hazard and operability studies (HAZOP) The structured what-if technique (SWIFT) Human error assessment and reduction technique (HEART) Failure mode and effects analysis (FMEA) Barrier analysis Influence diagrams Fault-tree analysis (FTA) Event-tree analysis (ETA) Absolute probability judgment (APJ) Risk matrices preliminary risk review (PRR)—which provides an initial risk assessment and also serves as a scoping and screening tool, helping users to determine which parts of the system or process of interest require a more in-depth analysis. (5,6) This research also identified a number of existing PHA techniques and supporting techniques that might be usefully applied to the healthcare sector, and prioritized 10 of these for initial inclusion in the tool kit. (See Exhibit 1 for the full list.) This article introduces one of these methods: the structured what-if technique (SWIFT).(7, 8) T HE ST RU C TU RED W H AT-IF TEC H N I Q U E ( SWI FT) Introduction The structured what-if technique (SWIFT) is a systemsbased risk identification technique that employs structured brainstorming, with the use of predeveloped guide words/headings (e.g., timing, amount, etc.) in combination with prompts elicited from participants (which often begin with the phrases “What if. . .” or “How could. . .”) to examine risks and hazards at a systems or subsystems level. This differentiates it from its precursor, the hazard operability studies (HAZOP) method, which is similar, but identifies hazards through a detailed review of lowlevel processes, subcomponents of equipment, etc.(8) SWIFT is essentially a magnifying glass to HAZOP’s (or FMEA’s) microscope. By focusing on high-level processes, it can often be conducted more quickly than more detailoriented methods. Indeed, 1 industry source reports that a SWIFT risk assessment can be conducted in as little as ⅓ the time required for a HAZOP-based approach,(9) a 24 result that was replicated in a study comparing SWIFT to HFMEA in a healthcare setting.(10) This time savings is a significant advantage. The corresponding disadvantage is that some hazards may be overlooked when using the SWIFT approach that would be identified using the more detail-oriented HAZOP or FMEA.(10) SWIFT need not be used on a stand-alone basis, however. As with the PRR described above, it can be used as the first part of a staged approach to identify quickly processes and subsystems for which it would be worth the investment of conducting an FMEA, HAZOP, FTA, or other detail-oriented risk assessment. This approach has the potential to reduce the overall amount of time and tedium(11) involved significantly, without sacrificing rigor. Similarly, although the outputs of a SWIFT are qualitative, the technique can be used to identify subsystems/processes that could benefit from a quantitative PHA approach.(12) Because SWIFT is a workshop-based technique in which potential risks are elicited from participants, it is important to assemble the right team when using this approach. Ideally this should include the representation of all stakeholder groups and those with the most intimate knowledge of the system or process being assessed (often frontline workers). SWIFT is very dependent on participants’ knowledge of the systems and processes being assessed. In addition to producing a more valid risk assessment, including these participants can have another important benefit: Participating in the SWIFT can enhance commitment to new and existing risk controls.(12) Procedure The methodological contribution of SWIFT is as a technique for hazard identification (when asking “How could. . .”), and risk identification (when asking “What if. . .”). But in practice it is typically supplemented by risk analysis [i.e., characterizing and estimating the risk(13)], risk evaluation [i.e., determining whether the risk is acceptable, or requires action(13)], and risk treatment planning [i.e., developing and assessing action plans to control risk(13)].(12) The risk analysis and risk control generation are often generic (using no particular method), whereas the risk evaluation and risk control evaluation typically use a real or implied risk matrix. Alternatively, a more rigorous assessment may be undertaken in which a supporting technique such as barrier analysis(5, 12, 14) or influence diagrams(5, 15, 16) is substituted for the generic risk analysis. A suggested procedure for conducting a risk assessment with the use of SWIFT is illustrated in Exhibit 2. U S E O F S WI FT I N H E A LTH C A R E We conducted a systematic literature review on 09 January 2012 to identify articles describing the use of SWIFT in healthcare risk management. Table 1 describes the search strategy. JOURNAL OF HEALTHCARE RISK MANAGEMENT • VOLUME 31, NUMBER 4 DOI: 10.1002/jhrm Exhibit 2: A Procedure for a Risk Assessment with the Use of SWIFT (Based on the PHA Waterfall Model)(5) 1. Prepare the guide words: The facilitator should select a set of guide words to be used in the SWIFT. 2. Assemble the team: Select participants for the SWIFT workshop based on their knowledge of the system/process being assessed and the degree to which they represent the full range of stakeholder groups. 3. Background: Describe the trigger for the SWIFT (e.g., a regulatory change, an adverse event, etc.). 4. Articulate the purpose: Clearly explain the purpose to be served by the SWIFT (e.g., to improve patient satisfaction scores). 5. Define the requirements: Articulate the criteria for success (e.g., no lost revenue over the next 5 years from reduced compensation as a result of low patient satisfaction scores). 6. Describe the system: Provide high-level textual and graphical descriptions of the system or process to be risk assessed. Do not get bogged down in detail. 7. Identify the risks/hazards: This is where the structured what-if technique is applied. Use the guide words/headings to each system, high-level subsystem, or process step in turn. Participants should use prompts starting with the phrases like “What if…” or “How could…” to elicit potential risks/hazards associated with the guide word. For instance, if the process is “Keep the patient informed about his or her condition,” and the guide word is “time, timing or speed,” prompts might include: “What if the patient is told about his or her condition while still sedated?” (wrong time) or “How could the patient be left waiting too long without an update on his or her condition?” (wrong timing). 8. Assess the risks: With the use of either a generic approach or a supporting risk analysis technique, estimate the risk associated with the identified hazards. In light of existing controls, assess the likelihood that they could lead to harm and the severity of harm they might cause. Evaluate the acceptability of these risk levels, and identify any aspects of the system that may require more detailed risk identification and analysis. 9. Propose actions: Propose risk control action plans to reduce the identified risks to an acceptable level. 10. Review the process: Determine whether the SWIFT met its objectives, or whether a more detailed risk assessment is required for some parts of the system. 11. Overview: Produce a brief overview document to communicate the results of the SWIFT. 12. Additional risk assessment: Conduct additional risk assessments using more detailed or quantitative techniques, if required. Papers were assessed to determine the guide words used, the amount of time invested, the participants’ perceptions of the technique, and whether SWIFT was paired with the use of a method to support risk control generation and analysis. Table 1. Literature Review Search Strategy This search resulted in two hits, only one of which described the use of SWIFT in healthcare.(7) (The other was an advertisement.) We were also aware of two reports from the gray literature (i.e., research reports not available from bibliographic databases like those above) that described the use of SWIFT in healthcare.(5, 10) Interestingly, all three studies took place within the United Kingdom. Databases Queried Search Terms Peer-reviewed journal article EMBASE Medline PsycINFO CINAHL Health Business Elite “Structured what-if technique” (In any field) The article by Smith et al.(7) describes the use of SWIFT to identify risks in a system with a very broad scope: “Nonoperative risks associated with adult elective surgery under general anaesthesia.” DOI: 10.1002/jhrm “Structured what if technique” (In any field) It is difficult to imagine addressing a problem of the same breadth with FMEA, given the technique’s time-intensive AMERICAN SOCIETY FOR HEALTHCARE RISK MANAGEMENT • VOLUME 31, NUMBER 4 25 and detail-oriented nature. Even with the use of SWIFT, this assessment required the UK National Patient Safety Agency (NPSA) to convene panels of experts in a series of sessions over 5 days totaling somewhere between 240 and 300 person-hours (30 hours of total session time, with 8–10 participants per session). If a SWIFT can be assumed to take approximately ⅓ the time of an FMEA, this would imply that it would require on the order of 800 person-hours to assess this system using an FMEA. the other, so the 2 groups were not able to communicate with one another until both had finished. The authors described the guide words used as: “environmental factors, human causes, operating errors, maintenance/calibration, health and safety, communications, etc.” Their study identified 102 risks, and resulted in 95 risk control recommendations. The SWIFT took 2 hours, whereas the HFMEA required 5.5 hours (plus a 30-minute lunch break, for a total of 6 hours). Participants provided positive feedback for both methods. They were satisfied with the process, and confident in the results; however, those results differed significantly between the 2 methods. Although there were many areas of agreement between the 2 methods, more than 50% of the risks identified by each method were not picked up by the other approach. And both produced results that were significantly different from what the researchers found through RCA and ethnographic research. As with other risk assessment techniques [e.g., RCA(17) or FMEA(18)], SWIFT does not provide direct support for the generation or analysis of robust risk controls. Nor was a separate risk control method used to support this step. This is reflected in one of the limitations noted by the authors, namely: These findings contribute to a growing consensus in the research literature that accurate risk identification in healthcare requires the triangulation of data from multiple sources.(19–23) Using SWIFT as part of a staged approach with FMEA might therefore be not only more efficient, but also more effective than using either alone. The high priority given in many of the recommendations to awareness of risks and training to reduce them. These are superficially cheap and easy solutions but can serve to divert attention away from underlying system factors to the people who are closest to the patient, those whose actions are most likely to lead to a visible adverse event. The guide words used were not described, and SWIFT was not used in combination with any method to support the risk control process. The authors describe SWIFT as “. . .straightforward to learn and easy to use,” and noted as a positive side effect the fact that the participants learned a lot from each other during the sessions; in fact, many of the proposed actions focused on the dissemination of risk controls that were already being used by some participants. They also suggested that risk assessments with a broad scope could lead to more systemic risk control than taking a singleissue focus. Based on the findings of this article, SWIFT, with its relative speed and its ability to serve as a scoping and screening tool, would appear to be a good tool for this job. Gray literature A comparison with HFMEA As part of a broader study of e-health, researchers in the United Kingdom compared the use of SWIFT and HFMEA to examine the risks involved in the technologies used in an anticoagulant service.(10) SWIFT was used in this case as a stand-alone method, not with HFMEA as in the staged approach, but in parallel, as a way of testing the methods against one another. In order to make the comparison as rigorous as possible, the researchers allocated each method to 1 of 2 groups of volunteers; the volunteers in each group were matched in terms of profession and seniority. The same facilitators led both sessions, and these were conducted one right after 26 P H A TO O L K I T E VA LU ATI O N As part of the evaluation research for the PHA tool kit, 2 pilot case studies were conducted in which SWIFT was one of the components used. Among the guide words used were too soon/too late/doesn’t happen/wrong order, etc. Both of these case studies included the use of more than 1 PHA method, and did not aim to assess SWIFT, as such. And in both cases, artificial time constraints imposed by the nature of the study meant that the risk assessments could not be fully completed. However, the participants generally found the technique credible and easy to use. SWIFT was not used in combination with any method to support the risk control process. A D A P TI N G S WI FT FO R TH E H E A LTH C A R E S E C TO R Guide words for a more complex environment SWIFT originated in the chemical process industry as a faster and easier alternative to HAZOP.(12) Healthcare is generally considered to be a far more complex system, and some of the original guide words for SWIFT (see Exhibit 3) may be confusing or difficult to apply to the healthcare environment. The method is flexible, however, and the facilitator can choose any guide words that seem appropriate. A proposed set of guide words is shown in Exhibit 4. JOURNAL OF HEALTHCARE RISK MANAGEMENT • VOLUME 31, NUMBER 4 DOI: 10.1002/jhrm Exhibit 3: Original SWIFT Guide Words(8) The original SWIFT guide words • Material problems • External events or influences • Operating errors and other human factors • Analytical or sampling errors • Equipment or instrumentation malfunction • Process upsets of unspecified origin • Utility failures • Integrity failure or loss of containment • Emergency operations • Environmental release Exhibit 4: Proposed Guide Words for SWIFT in Healthcare Wrong: Person or people Examples: Wrong patient surgery, Referral to the wrong specialist, Treatment delivered by staff suffering from fatigue Wrong: Place, location, site, or environment Examples: Wrong-site surgery, Retained surgical sponges, Failure to isolate a patient with SARS, Unwarranted patient discharge, Poor lighting Wrong: Thing or things Examples: Poorly designed equipment, Wrong medication, Wrong syringe Wrong: Idea, information, or understanding Examples: Poor communication at handoff, Patient misinformed, Incorrect understanding of who is responsible for a given function Wrong: Time, timing, or speed Examples: Treatment is not delivered in a timely manner, Patient is not given time to process a cancer diagnosis before the appointment ends, Medication not delivered on schedule Wrong: Process Examples: Wrong surgical procedure performed, Failure to perform a suicide risk assessment for a mental health patient, Workarounds used to avoid complying with safety procedures Wrong: Amount Examples: Understaffing, Drug overdose, Drug shortages, Capacity shortfalls during a disaster DOI: 10.1002/jhrm Another option might be to use the category headings from the NPSA fishbone diagram tool: patient factors, individual (staff ) factors, task factors, communication factors, team factors, education and training factors, equipment and resources, working condition factors, organizational and strategic factors.(24) Risk control generation and analysis Like other commonly used risk assessment methods,(17) SWIFT originated in an industrial setting, where its use would typically be led by fully trained safety/reliability engineers. In that context, it is perhaps safe to assume that a good understanding of the risk will necessarily lead users to devise robust and appropriate risk controls. In healthcare, however, this tends not to be the case. As the article by Smith et al.(7) illustrates, healthcare workers find it very difficult to generate high-quality risk controls, or to differentiate between those that will prove robust and those that are weak—or possibly even harmful.(14,25–30) None of the articles we found described the use of a method to support the risk control process after a SWIFT. In this, it is no different from RCA(17) or FMEA.(18) There appears to be no widely used method to support robust risk control in healthcare risk management, regardless of the risk assessment technique employed. There is a pressing need for the development of tools to support risk control after both prospective and retrospective risk assessment. Cost effectiveness SWIFT appears to be considerably less time consuming than FMEA(9, 10), and produces overlapping but different results.(10) In common with FMEA, it also produces different results from RCA and ethnographic research, even when the same system is examined.(10) There is an increasing weight of evidence that comprehensive healthcare risk assessment requires the triangulation of data from multiple sources and techniques.(10,19–23) Additional research is required to define the most costeffective role for each, including the correct balance between retrospective and prospective risk assessment methods in healthcare risk management. C O N C LU S I O N The structured what-if technique (SWIFT) is a flexible, high-level risk identification technique that can be used on a stand-alone basis, or as part of a staged approach to make more efficient use of detail-oriented methods like FMEA. Based on the limited evidence available, it appears that healthcare workers find it easy to learn, easy to use, and credible.(5, 7, 10) Different risk identification methods produce different results, and triangulation between multiple methods is probably the best way to achieve an accurate understanding of the risks in a given system.(19–21) Especially when used as part of a AMERICAN SOCIETY FOR HEALTHCARE RISK MANAGEMENT • VOLUME 31, NUMBER 4 27 staged approach, SWIFT appears capable of playing a useful role as a component of the PHA armamentarium. RE FE REN C ES 1. Croteau R. Risk assessing risk assessment. Jt Comm J Qual Patient Saf. 2010;36(8):35–37. Available at: http://msn.jcrinc.com/Common/Documents/ OnlineExtras/RAHS10Extras/s2.pdf. Accessed January 10, 2012. 2. Israelski EW, Muto WH. Human factors risk management as a way to improve medical device safety: a case study of the Therac-25 radiation therapy system. In: Gosbee JW, Gosbee LL, eds. Using Human Factors Engineering to Improve Patient Safety. Oakbrook Terrace, IL: Joint Commission Resources; 2005: 123–132. 3. DeRosier J, Stalhandske E, Bagian JP, Nudell T. Using healthcare failure mode and effect analysis: The VA National Center for Patient Safety’s prospective risk analysis system. Jt Comm J Qual Improv. 2002;28(5):248–267. 4. Dhillon BS. Methods for performing human reliability and error analysis in healthcare. Int J Healthcare Qual Assur. 2003;16(6):306–317. Available at: http://www.emeraldinsight .com/10.1108/09526860310495697. Accessed December 15, 2011. 5. Ward J, Clarkson J, Buckle P, et al. Prospective Hazard Analysis: Tailoring Prospective Methods to a Healthcare Context. 2010. Available at: http://www. haps2.bham.ac.uk/publichealth/psrp/PS035_Project_ Summary.shtml. Accessed March 1, 2012. 6. Card AJ, Harrison H, Ward J, Clarkson PJ. Using prospective hazard analysis to assess an active shooter emergency operations plan. J Healthc Risk Manag. 2012:31(3):34–40. 7. Smith A, Boult M, Woods I, Johnson S. Promoting patient safety through prospective risk identification: Example from peri-operative care. Qual Saf Healthcare. 2010;19(1):69–73. Available at: http://qshc .bmj.com/content/19/1/69.abstract. 8. Maguire R. Safety Cases and Safety Reports. Meaning, Motivation and Management. Burlington, VT: Ashgate Publishing; 2006:101–103. 9. Lloyd’s Register. HAZOP and Hazard Identification Services. 2008;2010(4 January). Available at: http://www.lloydsregisterasia.com/services/pdfs/ risk-management-services-factsheet2.pdf. Accessed January 10, 2012. 10. Potts HWW, Keen J, Denby T, et al. Towards a Better Understanding of Delivering E-Health Systems: A 28 Systematic Review Using The Meta-Narrative Method and Two Case Studies. Final Report. London, England: NIHR Service Delivery and Organisation Programme; 2011. 11. Burgmeier J. Failure mode and effect analysis: An application in reducing risk in blood transfusion. Jt Comm J Qual Improv. 2002;28(6):331–339. 12. ISO. ISO 31010: Risk Management—Risk Assessment Techniques. Geneva, Switzerland: ISO; 2009. 13. ISO. ISO 31000: Risk Management—Principles and Guidelines on Implementation. Geneva, Switzerland: ISO; 2009. 14. Lyons M, Woloshynowych M, Adams S, Vincent C. Error Reduction in Medicine: Final Report to the Nuffield Trust. London, England: Nuffield Trust; 2005. 15. Lee RC, Ekaette E, Kelly K-L, et al. Implications of cancer staging uncertainties in radiation therapy decisions. Med Decis Mak. 2006;26(3):226–238. Available at: http://mdm.sagepub.com/cgi/content/ abstract/26/3/226. 16. Gomez M, Bielza C, Fernandez del Pozo JA, Rios-Insua S. A graphical decision-theoretic model for neonatal jaundice. Med Decis Mak. 2007;27(3):250–265. Available at: http://mdm. sagepub.com/cgi/content/abstract/27/3/250. 17. Card AJ, Ward J, Clarkson PJ. Successful risk assessment may not always lead to successful risk control: A systematic literature review of risk control after root cause analysis J Healthc Risk Manag. 2012:31(3):6–12. 18. Grout JR. Mistake-proofing the design of healthcare processes. 2007. Available at: http://www.ahrq.gov/ qual/mistakeproof/. 19. Olsen S, Neale G, Schwab K, et al. Hospital staff should use more than one method to detect adverse events and potential adverse events: Incident reporting, pharmacist surveillance and local real-time record review may all have a place. Qual Saf Healthcare. 2007;16(1):40–44. 20. Sari AB-A, Sheldon TA, Cracknell A, Turnbull A. Sensitivity of routine system for reporting patient safety incidents in an NHS hospital: Retrospective patient case note review. BMJ (Clin Res Ed). 2007;334(7584):79. 21. Kessels-Habraken M, Van der Schaaf T, De Jonge J, Rutte C, Kerkvliet K. Integration of prospective and retrospective methods for risk analysis in hospitals. Int J Qual Healthcare. 2009;21(6):427–432. 22. Hogan H, Olsen S, Scobie S, et al. What can we learn about patient safety from information sources JOURNAL OF HEALTHCARE RISK MANAGEMENT • VOLUME 31, NUMBER 4 DOI: 10.1002/jhrm within an acute hospital: A step on the ladder of integrated risk management? Qual Saf Healthcare. 2008;17(3):209–15. Available at: http://www.ncbi. nlm.nih.gov/pubmed/18519628. 28. Wallace LM, Spurgeon P, Earll L. Evaluation of the NPSA 3 Day Root Cause Analysis Training Programme: Final Report (Rev.). 2006. Department of Health, London. 23. Wet C de, Bowie P. The preliminary development and testing of a global trigger tool to detect error and patient harm in primary-care records. Postgrad Med J. 2009;85(1002):176–180. Available at: http://www .ncbi.nlm.nih.gov/pubmed/19417164. 29. Mills PD, Neily J, Luan D, Stalhandske E, Weeks W. Using aggregate root cause analysis to reduce falls and related injuries. Jt Comm J Qual Pat Saf. 2005;31(1):21–31. 24. National Patient Safety Agency. Root Cause Analysis Investigation: Fishbone Diagram Tool. August 6, 2010. Available at: http://www.nrls.npsa.nhs.uk/ resources/?entryid45=75605. Accessed January 10, 2012. 25. Pham JC, Kim GR, Natterman JP, et al. ReCASTing the RCA: An improved model for performing root cause analyses. Am J Med Qual. 2010;25(3):186–191. 26. Youngson GG, Flin R. Patient safety in surgery: Nontechnical aspects of safe surgical performance. Patient Saf Surg. 2010;4(1):4. 27. Percarpio KB, Watts BV, Weeks WB. The effectiveness of root cause analysis: what does the literature tell us? Jt Comm J Qual Saf. 2008;34(7):391–398. DOI: 10.1002/jhrm 30. Mills D, Neily J, Kinney LM, Bagian J, Weeks WB. Effective interventions and implementation strategies to reduce adverse drug events in the Veterans Affairs (VA) system. Qual Saf Healthcare. 2008;17(1):37–46. A BO U T TH E A U TH O R S Alan J. Card, MPH, CPH, CPHQ, is a doctoral candidate at the University of Cambridge Engineering Design Centre and President/CEO of Evidence-Based Health Solutions, LLC. James R. Ward, BEng, CEng, PhD, MIET, works for the University of Cambridge as a researcher in patient safety, and for a number of hospitals in the local region. P. John Clarkson, PhD, BA(Eng), is a professor at the University of Cambridge and director of the engineering Design Centre. AMERICAN SOCIETY FOR HEALTHCARE RISK MANAGEMENT • VOLUME 31, NUMBER 4 29