Etiology of Shock in The Emergency Department: A 12-Year Population-Based Cohort Study

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

SHOCK, Vol. 51, No. 1, pp.

60–67, 2019

ETIOLOGY OF SHOCK IN THE EMERGENCY DEPARTMENT: A 12-YEAR


POPULATION-BASED COHORT STUDY

Jon Gitz Holler, * Helene Kildegaard Jensen, * Daniel Pilsgaard Henriksen, †


Lars Melholt Rasmussen, ‡ Søren Mikkelsen, § Court Pedersen, jj
and Annmarie Touborg Lassen *
*Departmentof Emergency Medicine, Odense University Hospital, Odense C, Denmark; † Department of
Respiratory Medicine, Odense University Hospital, Odense C, Denmark; ‡ Centre for Individualized
Medicine in Arterial Diseases, Department of Clinical Biochemistry and Pharmacology, Odense University
Hospital, Odense C, Denmark; §Department of Anesthesiology and Intensive Care Medicine, Odense
University Hospital, Odense C, Denmark; and jjDepartment of Infectious Diseases, Odense University
Hospital, Odense C, Denmark

Received 24 Apr 2016; first review completed 12 May 2016; accepted in final form 30 Nov 2016

ABSTRACT—Introduction: The knowledge of the etiology and associated mortality of undifferentiated shock in the
emergency department (ED) is limited. We aimed to describe the etiology-based proportions and incidence rates (IR) of
shock, as well as the associated mortality in the ED. Methods: Population-based cohort study at a University Hospital ED in
Denmark from January 1, 2000, to December 31, 2011. Patients aged 18 years living in the ED-catchment area
(N ¼ 225,000) with a first-time ED presentation with shock (n ¼ 1,553) defined as hypotension (systolic blood pressure
100 mm Hg) and 1 organ failures were included. Discharge diagnoses defined the etiology and were grouped as follows:
distributive septic shock (SS), distributive non-septic shock (NS), cardiogenic shock (CS), hypovolemic shock (HS),
obstructive shock (OS), and other conditions (OC). Outcomes were etiology-based characteristics, annual IR per
100,000 person-years at risk (95% confidence intervals [CIs]), mortality at 0 to 7-, and 0 to 90 days (95% CIs) and hazard
rates (HR) at 0 to 7, 8 to 90 days (95% CIs). Poisson and Cox regression models were used for analyses. Results: Among
1,553 shock patients: 423 (27.2%) had SS, 363 (23.4%) NS, 217 (14.0%) CS, 479 (30.8%) HS, 14 (0.9%) OS, and 57 (3.7%)
OC. The corresponding IRs were 16.2/100,000 (95% CI: 14.8–17.9), 13.9/100,000 (95% CI: 12.6–15.4), 8.3/100,000 (95%
CI: 7.3–9.5), 18.4/100,000 (95% CI: 16.8–20.1), 0.5/100,000 (95% CI: 0.3–0.9), and 2.2/100,000 (95% CI: 1.7–2.8). SS IR
increased from 8.4 to 28.5/100,000 during the period 2000 to 2011. Accordingly, the 7-, and 90-day mortalities of SS, NS, CS,
and HS were 30.3% (95% CI: 25.9–34.7) and 56.2% (95% CI: 50.7–61.5), 12.7% (95% CI: 9.2–16.1) and 22.6% (95% CI:
18.1–27.7), 34.6% (95% CI: 28.2–40.9) and 52.3% (95% CI: 44.6–59.8), 19.2% (95% CI: 15.7–22.7), and 36.8% (95% CI:
33.3–43.3). SS (HR ¼ 1.46 [95% CI: 1.03–2.07]), and CS (HR ¼ 2.15 [95% CI: 1.47–3.13]) were independent predictors of
death within 0 to 7 days, whereas SS was a predictor within 8 to 90 days (HR ¼ 1.66 [95% CI: 1.14–2.42]). Conclusion: HS
and SS are frequent etiological characteristics followed by NS and CS, whereas OS is a rare condition. We confirm the
increasing trend of SS, as previously reported. Seven-day mortality ranged from 12.7% to 34.6%, while 90-day mortality
ranged from 22.6% to 56.2%. The underlying etiology was an independent predictor of mortality.
KEYWORDS—Emergency department, epidemiology, etiology, incidence rate, mortality, shock

INTRODUCTION
Address reprint requests to Jon Gitz Holler, MD, PhD, Department of Emergency
Medicine, Odense University Hospital, Sdr Boulevard 29, Entrance 130, 1. Floor
5000, Odense C, Denmark. E-mail: [email protected] Circulatory shock is a life-threatening condition associated
JGH conceived the study, analyzed and interpreted the data, and drafted the with substantial morbidity and mortality (1). Shock result from
manuscript. ATL conceived and coordinated the study and performed critical one, or a combination, of four different pathophysiological
appraisal of the manuscript. ATL, HKJ, CP, DPH, LMR, and SM contributed to
writing, reviewing, and revising the paper. All authors interpreted the data and mechanisms (1). Internal or external loss of fluids (i.e., trauma
critically reviewed drafts of the manuscript. All authors edited and approved the final or gastrointestinal bleeding) often cause hypovolemia, while
manuscript. intracardiac etiologies (i.e., myocardial infarction, myopathy,
The work was supported by University of Southern Denmark and an unrestricted
grant from the Philanthropic private fund TrygFonden given to the University of or a major arrhythmia) cause altered contractility and cardio-
Southern Denmark. genic failure. Extracardiac etiologies of cardiac pump failure
JGH and Professor in Emergency Medicine, ATL, are financially supported by an (i.e., pulmonary embolism, tension pneumothorax) cause
unrestricted grant provided by the philanthropic foundation TrygFonden to Univer-
sity of Southern Denmark. ATL and JGH are employed by the Faculty of Health and obstruction. Effects of inflammatory agents (distributive, i.e.,
Medical sciences, University of Southern Denmark. None of the authors have sepsis, anaphylaxis, poisoning, or other vasodilation effects)
financial interests in the project. often mediate vascular permeability and loss of vascular tone
The authors have no conflicts of interest.
Supplemental digital content is available for this article. Direct URL citation and lead to distributive shock (1, 2).
appears in the printed text and is provided in the HTML and PDF versions of this Knowledge of the frequency and associated prognosis across
article on the journal’s Web site (www.shockjournal.com). etiologies of undifferentiated shock are limited, especially in
DOI: 10.1097/SHK.0000000000000816
Copyright ß 2018 The Author(s). Published by Wolters Kluwer Health, Inc. on the emergency department (ED) (3). The research available has
behalf of the Shock Society. This is an open access article distributed under the terms mainly been limited to the post-ED period. The estimates
of the Creative Commons Attribution-Non Commercial-No Derivatives License 4.0 reported are often based on highly selected patient populations
(CCBY-NC-ND), where it is permissible to download and share the work provided it
is properly cited. The work cannot be changed in any way or used commercially (e.g., septic or cardiogenic shock) in the intensive care unit
without permission from the journal. (ICU) and are of limited value for understanding the early
60
SHOCK JANUARY 2019 ETIOLOGY OF SHOCK IN THE EMERGENCY DEPARTMENT 61

etiological characteristics of shock at the presentation in the Population-based registers


ED. Clarification of the shock-related etiologies in the pre-ICU In Denmark, the Danish National Health Service provides free individual
period could aid knowledge to clinical decision makers han- ED services to every Danish citizen. At birth the Danish Civil Registration
system assigns a unique 10-digit civil personal registry number (PRN-
dling critically ill ED patients. We therefore carried out a number) to each Danish citizen and to residents upon immigration since
population-based epidemiologic study in an area with a 1968. This unique PRN-number enables accurate linkage of the Danish
well-defined adult population. national registers (10).
The Danish National Patient Registry—Since 1995 the Danish National
The aims of the present study were to examine the etiology Patient Registry has registered all in and outpatient clinic contacts at hospitals in
of shock in an ED setting in a well-defined Danish area as Denmark assembling data regarding dates of admission, discharge, admitting
well as trends in the annual incidence and mortality across departments, and all primary and secondary discharge diagnoses (International
Classification of Diseases, version 10 [ICD-10] code system, since 1994) from
etiologies. hospitals (11, 12). At discharge every unique patient contact is assigned one primary
diagnosis and secondary diagnoses classified according to the ICD-10 system. From
The Danish Civil Registration System, we retrieved data on municipality of
PATIENTS AND METHODS residence, migration, marital and vital status, and date of birth (10).
Study design and setting
Outcome measures, exposure, and variables of interest
This is a secondary analysis of the clinical characteristics and outcomes of
adult patients with shock enrolled in a population-based cohort study at the ED Shock was defined as the presence SBP  100 mm Hg (9) and 1 organ
of Odense University Hospital, Denmark, between January 1, 2000, and failures. We used the Shock Index (SI) as a measure of cardiovascular
December 31, 2011 (4). Odense University Hospital is a 1,000-bed university dysfunction (13, 14). SI is calculated as the ratio of heart rate to SBP and
teaching hospital with all specialties represented. The ED, at Odense University included as a categorical variable (<0.7, 0.7–1, 1) (13). The following organ
Hospital, serves a mixed rural-urban population of 225,000 person (age 18). It failures were included: cardiovascular (SI  1), renal, coagulation, and hepatic.
is the only serving ED in this part of Denmark and provides primary 24-h acute Biochemical variables (creatinine, bilirubin, platelets, and international nor-
medical care, with 37,000 annual adult visits (5). The basic prehospital malized ratio [INR]) registered 180 days before and 1 day after the index date
assistance response is an ambulance staffed by two emergency medical was used to identify renal (>100 mmol/L recent S-creatinine increase, hepatic
technicians (EMTs) with competences restricted to such as initial treatment (recent S-bilirubin >42 mmol/L) and coagulative failure (recent platelet count
of patients with myocardial infarction (nitroglycerine, thrombolytic agents, <101  109/L or recent INR >1.59) (see Appendix 1, http://links.lww.com/
defibrillation), fluid, and inhalation therapy (6). From 2006 and onward, a SHK/A522 for details). SBP was measured with either an automated oscillo-
physician-staffed mobile emergency care unit manned with a physician spe- metric device or a manual cuff and sphygmomanometer. Heart rate was
cialist in anesthesiology and an EMT were added to the prehospital emergency measured with ECG, palpation or pulse oximetry. The primary outcome was
medical system (6). This unit provides prehospital advanced medical treatment the etiology of shock. Secondary outcomes were the etiological trends in annual
exceeding the competences of the EMTs (severe chest pain, sudden loss of IRs and mortality proportions and the mortality within 7- and 90 days of the
consciousness, high-velocity car crash, dyspnea, etc.) (6). index date. The primary exposure variables were the first recorded
At ED arrival patients are usually assessed in the ED and hereafter allocated SBP  100 mm Hg and HR at presentation, registered within 3 h upon arrival
and admitted to a specific specialty (orthopedic surgery, infectious diseases, and the presence of 1 organ failures.
etc.) or referred to primary care or ICU after primary ED evaluation. Patients are We used the primary discharge diagnose, assigned to each patient at
treated according to international standard algorithms (e.g., Advanced Trauma discharge and grouped these into the classification of shock states proposed
Life Support) program and the Injury Severity Score for trauma, Surviving by Weil and Shubin (15) and latter updated by Vincent et al. (2). In brief, a
Sepsis Campaign guidelines for management of severe sepsis and septic shock discharge diagnosis suggesting an infectious pathophysiology (e.g., pneumo-
[from 2006 and onward], percutaneous coronary intervention for obstructive nia) was grouped as ‘‘septic’’ (distributive), whereas non-infectious etiologies
coronary artery disease). In 2009, the Adaptive process triage (ADAPT) was causing inflammation and vasodilation were grouped as ‘‘non-septic’’ (distrib-
implemented in the ED at Odense University Hospital and is the most utive) (e.g., poisoning, diabetes). Cardiovascular diagnoses (e.g., myocardial
commonly used triage system in Denmark (7). infarction, arrhythmia) were grouped as ‘‘cardiogenic,’’ and diagnoses suggest-
ing hemorrhage, trauma, or dehydration, were grouped as ‘‘hypovolemia.’’
Conditions causing increased afterload (e.g., pulmonary embolus) where group
Participants as ‘‘obstructive.’’ Finally, diagnoses that did not meet these criteria were
Patients aged 18 years presenting to the ED with a systolic blood pressure grouped as ‘‘others.’’ See Appendix 2, http://links.lww.com/SHK/A523, for
(SBP) 100 mm Hg registered within 3 h upon arrival during the study period full list of ICD-10 codes and classification (n ¼ 1,170). For patients with
were considered eligible. Recording of vital values and blood tests were per- primary unclassifiable diagnoses (n ¼ 383) (see Appendix 3, http://link-
formed based on a clinical judgment. Patients suffering minor complaints (e.g., s.lww.com/SHK/A524), two authors (JGH and HKJ) independently and in a
sprained ankle, etc.) had not vital values measured or blood test analysis blinded manner read and evaluated all discharge summaries for information
performed. The primary date of contact defined the index date. Patients were indicating etiological characteristics and if such information was not clearly
included the first time they presented with hypotension (SBP  100 mm Hg) stated, the full medical record was reviewed.
during the study period. We excluded patients who had visited the ED with We also included information on the additional covariates: sex, age, and
hypotension between January 1, 1998, and January 1, 2000, to minimize left-sided Charlson Comorbidity Index (CCI). The latter, was used as a proxy for
censoring. Patients without a Danish personal identification number and patients comorbid illness (16). We used discharge diagnoses from the previous 10 years
residing outside the hospitals catchment area at the time of contact were excluded. to generate the CCI (0, 1–2, >2) for each enrolled patient upon the index
Patients were retrieved from the background population (N ¼ 225,000) and were contact date (16).
followed from index date until the date of death, emigration, December 31, 2011,
or completion of 90 days follow-up, whichever came first. Statistical analysis
We presented continuous and categorical data as medians (interquartile
Data sources and processing range [IQR]) and numbers (%), respectively. We used the Pearson chi-square
Database—All patients’ records from the ED were registered electronically. test for categorical variables and the Kruskal–Wallis equality-of-populations
In the records, vital parameters are consistently stated and available as rank test for continuous variables.
structured text format, including time of admission and time of measured Incidence rates—The crude annual IRs were calculated as the number of IRs
SBP and heart rate (HR). By electronic screening it was possible to identify per 100,000 person-years at risk (age 18 years) with the corresponding 95%
information on all patients with the unique recorded value of SBP and HR. The confidence intervals (95% CI) assuming a Poisson distribution. The annual IRs
method of free-text search has been validated in the context of extracting were adjusted using direct standardization to the sex- and age distribution of the
numerical data, including blood pressure recordings (8). The data extraction municipalities of the EDs catchment area midyear population in the year 2000.
process used has previously been validated in 500 random ED notes to have a The population was defined as contributing to one person-year at risk per
sensitivity of 95.8% (95% CI: 91.2–98.5) and a specificity of 100% (95% CI: resident per year in the analyses (17). The incidence rates were estimated and
99.0–100) for retrieving correct SBP (9). Data were supplemented with analyzed using a Poisson regression model. Sex, age group, calendar time in
information from the hospitals’ local and laboratory information systems to years were used in the adjusted model. Calendar time was entered in the model
define organ dysfunction. as a continuous variable. Age was divided into two predefined age intervals: 18
62 SHOCK VOL. 51, No. 1 HOLLER ET AL.

to 64 years and 65 years. The Poisson model was assessed using the Hosmer–
Lemeshow goodness-of-fit test.
Mortality analysis—Mortality for the different shock etiologies were pre-
sented in a Kaplan–Meier plot, and comparison between survival curves were
tested using log-rank test. Mortality proportions were reported at 7-, and 90-day
after the index date. Independent prognostics of mortality across etiologies were
evaluated by Cox regression and presented as unadjusted and adjusted hazard
ratios (HRs) with 95% confidence intervals (CIs) for time periods 0 to 7 days and 8
to 90 days. The model was adjusted for the following predefined confounders: sex,
age, CCI, and number of organ failures (1, 2, and 3). Model assumptions were
checked using Schoenfeld residuals and model fit evaluated using Cox–Snell
residuals. Cuzick test was used for trends in annual mortality and mortality trends
in age intervals. Statistical analyses were performed using Stata version 13.1
(Stata Corporation LP, College Station, Tex).

Ethics committee approval


The study was approved by the Danish Data Protection Agency (j.nr. 2008-
58-0035) and the Danish Health and Medicines Authority (j.nr. 3-3013-205/1).
In accordance with Danish law, observational studies performed in Denmark
do not need approval from the Medical Ethics Committee. The study was
reported according to the Strengthening the Reporting of Observational Studies
in Epidemiology statement (18).

RESULTS
FIG. 1. Flow chart of patients recruited to the study.
Participants
Of 438,191 adult ED contacts, a total 1,553 (0.4%) patients met
the criteria for shock and were included in the analysis. Reasons (n ¼ 363 [23.4%]), cardiogenic (n ¼ 217 [14.0%]), obstructive
for exclusions are presented in Figure 1. The median age was (n ¼ 14 [0.9%]) and other (n ¼ 57 [3.7%]). Patients suffering
71 years (IQR 56–81) and 53.4% were male (Table 1). Further septic shock had more comorbidity and a higher number of organ
characteristics of the population have been presented in a previous failure, compared with other etiologies.
paper (4). The most common etiology was hypovolemic (n ¼ 479 The validation of the primary unclassifiable diagnoses
[30.8%]), followed by septic (n ¼ 423 [27.2%]), non-septic (n ¼ 383) found a general interrater agreement of 70.3% (k,

TABLE 1. Baseline characteristics at time of arrival to the ED*


Distributive Distributive
Variable Total (%) (septic) (non-septic) Cardiogenic Hypovolemic Obstructive Other

N (%) 1,553 (100%) 423 (27.2%) 363 (23.4%) 217 (14.0%) 479 (30.8%) 14 (0.9%) 57 (3.7%)
Age in years, median (IQR) 70 (56–81) 74 (63–83) 56 (43–72) 75 (65–82) 72 (57–81) 72 (66–77) 68 (54–80)
Sex (%)
Male 830 (53.4%) 231 (54.6%) 183 (50.4%) 114 (52.5%) 267 (55.7%) 6 (42.9%) 29 (50.9%)
Female 723 (46.6%) 192 (45.4%) 180 (49.6%) 103 (47.5%) 212 (44.3%) 8 (57.1%) 28 (49.1%)
Age in age groups, yr (%)
18–39 147 (9.5%) 18 (4.3%) 73 (20.1%) 6 (2.8%) 40 (8.4%) 1 (7.1%) 9 (15.8%)
40–64 468 (30.1%) 92 (21.7%) 167 (46.0%) 47 (21.7%) 141 (29.4%) 2 (14.3%) 19 (33.3%)
65–84 691 (44.5%) 220 (52.0%) 98 (27.0%) 127 (58.5%) 212 (44.3%) 11 (78.6%) 23 (40.4%)
85þ 247 (15.9%) 93 (22.0%) 25 (6.9%) 37 (17.1%) 86 (18.0%) 0 (0.0%) 6 (10.5%)
CCI (%)
0 477 (30.7%) 87 (20.6%) 148 (40.8%) 55 (25.3%) 155 (32.4%) 4 (28.6%) 28 (49.1%)
1–2 589 (37.9%) 180 (42.6%) 117 (32.2%) 85 (39.2%) 186 (38.8%) 4 (28.6%) 17 (29.8%)
>2 487 (31.4%) 156 (36.9%) 98 (27.0%) 77 (35.5%) 138 (28.8%) 6 (42.9%) 12 (21.1%)
Vital variables
Systolic blood pressure, median (IQR) 88 (80–94) 88 (80–94) 90 (82–95) 88 (77–93) 86 (78–93) 89 (78–92) 87 (75–93)
Diastolic blood pressure, median (IQR) 52 (44–62) 51 (42–59) 53 (45–65) 52 (42–62) 52 (45–61) 56 (49–65) 56 (48–66)
Heart rate, median (IQR)† 101 (88–115) 104 (90–118) 100 (88–114) 104 (88–129) 100 (86–110) 111 (98–120) 100 (91–108)
Number of organ dysfunctions, n (%)
1 1,160 (74.7%) 275 (65.0%) 290 (79.9%) 178 (82.0%) 355 (74.1%) 9 (64.3%) 53 (93.0%)
2 311 (20.0%) 106 (25.1%) 59 (16.3%) 34 (15.7%) 104 (21.7%) 4 (28.6%) 4 (7.0%)
3þ 82 (5.3%) 42 (9.9%) 14 (3.9%) 5 (2.3%) 20 (4.2%) 1 (7.1%) 0 (0.0%)
Site of organ failure (%)
Cardiovascular 1,245 (80.2%) 329 (77.8%) 299 (82.4% 174 (80.2%) 378 (78.9%) 13 (92.9%) 52 (91.2%)
Renal 333 (21.4%) 133 (31.4%) 63 (17.4%) 50 (23.0%) 83 (17.3%) 2 (14.3%) 2 (3.5%)
Coagulation 387 (24.9%) 125 (29.6%) 70 (19.3%) 32 (14.7%) 149 (31.1%) 4 (28.6%) 7 (12.8%)
Hepatic 72 (4.6%) 30 (7.1%) 21 (5.8%) 5 (2.3%) 15 (3.1%) 1 (7.1%) 0 (0.0%)
*Values expressed as total number (fraction) and medians [25 percentile-75 percentile] as appropriate. Chi-squared test for categorical variables and
Kruskal-Wallis test for continuous variables.

35 Missing.
CCI indicates Charlson Comorbidity Index; ED, emergency department; IQR, interquartile range.
SHOCK JANUARY 2019 ETIOLOGY OF SHOCK IN THE EMERGENCY DEPARTMENT 63

FIG. 2. Crude annual incidence rates of shock from 2000 to 2011


stratified on etiology (septic, non-septic, hypovolemic, cardiogenic,
obstructive, and other).

0.62; 95% CI: 0.59–0.63), across all etiologies, corresponding FIG. 3. Crude incidence rates stratified by age and etiology.
to a substantial strength of agreement (19). When restricting to
specific etiologies, the interrater agreement was between Obstructive shock—Fourteen (0.9%) suffered obstructive
33.3% (obstructive) and 82.4% (septic). shock. The annual IR was 0.5/100,000 pyar (95% CI: 0.3–0.9).
Other causes—The IR of other causes was 2.2/100,000 pyar
Incidence rates (95% CI: 1.7–2.8), based on 57 patients (3.7%).
The yearly mean crude IRs across etiologies are shown in
Figure 2 and the IR stratified into age intervals (Fig. 3). Except Mortality
for septic shock, there were no significant trends in the annual The 7-day mortality of the major groups, non-septic, septic,
IR of the different etiologies during the period 2000 to 2011. cardiogenic, and hypovolemic, were 12.7% (95% CI: 9.2–
IRs of septic, non-septic, cardiogenic, and hypovolemic shock 16.1), 30.3% (95% CI: 25.9–34.7), 34.6% (95% CI: 28.2–
all increased by age. 40.9), and 19.2% (95% CI: 15.7–22.7), respectively (Table 2).
Septic shock—Of the 423 (27.2%) patients with septic shock, Accordingly, 90-day mortality was 22.6% (95% CI: 18.1–
the corresponding IR was 16.2/100,000 person-years at risk 27.7), 56.2% (95% CI: 50.7–61.5), 52.3% (95% CI: 44.6–
(pyar) (95% CI: 14.8–17.9). The IR increased from 8.4 to 28.5/ 59.8), and 36.8% (95% CI: 33.3–43.3), respectively. Kaplan–
100,000 pyar, during the period 2000 to 2011, with an average Meier curves are shown in Figure 4 with the overall estimated
adjusted annual increase of 8.5% (95% CI: 5.5–11.7). probability of 90-day survival stratified into etiologies. Trend
Non-septic shock—Three hundred sixty three (23.4%) suf- analysis of the annual 7-, and 90-day mortality proportions did
fered non-septic shock. The IR of non-septic shock was 13.9/ not show a significant change during the entire observation
100,000 pyar (95% CI: 12.6–15.4). period across etiologies of shock.
Cardiogenic shock—Of the 217 (14.0%) with cardiogenic
shock the IR was 8.3/100,000 pyar (95% CI: 7.3–9.5). Mortality related to etiology
Hypovolemic shock—Four hundred seventy four (30.8%) In the multivariate analysis (controlled for sex, age, CCI, and
suffered hypovolemic shock. The IR was 18.4/100,000 pyar number of organ failure) patients with septic (HR ¼ 1.46 [95%
(95% CI: 16.8–20.1). CI: 1.03–2.07]) and cardiogenic shock (HR ¼ 2.15 [95%

TABLE 2. Importance of etiology for death in patients presenting with shock at presentation to the ED—Cox regression
0–7 days 8–90 days
N, total N, died Crude HR Adjusted HR N, died Crude HR Adjusted HR
(%) (%) (95% CI) P** (95% CI)* P** (%) (95% CI) P** (95% CI)* P**
Etiology
Non-septic 363 (23.4) 46 (12.7) 1 1 41 (11.3) 1 1
Septic 423 (27.2) 128 (30.3) 2.55 (1.82–3.57) <0.001 1.46 (1.03–2.07) 0.036 104 (24.6) 3.13 (2.18–4.49) <0.001 1.66 (1.14–2.42) 0.009
Cardiogenic 217 (14.0) 75 (34.6) 3.10 (2.15–4.47) <0.001 2.15 (1.47–3.13) <0.001 39 (17.8) 2.31 (1.49–3.58) <0.001 1.33 (0.85–2.08) 0.216
Hypovolaemic 479 (30.8) 92 (19.2) 1.57 (1.10–2.23) 0.013 1.12 (0.78–1.60) 0.547 85 (17.8) 1.78 (1.23–2.59) 0.002 1.14 (0.78–1.67) 0.496
Obstructive 14 (0.9) 7 (50.0) 4.94 (2.23–10.93) <0.001 3.04 (1.37–6.79) 0.006 2 (14.3) 2.41 (0.58–9.94) 0.226 1.61 (0.39–6.67) 0.512
Other 57 (3.7) 14 (24.6) 2.11 (1.16–3.83) 0.015 2.05 (1.12–3.75) 0.020 5 (8.8) 0.88 (0.35–2.22) 0.782 0.76 (0.30–1.92) 0.558
*Cox proportional hazard model adjusted for sex, age as a continuous variable, Charlson comorbidity level (0, 1–2, >2) and number of organ
dysfunctions. Patients who died during the first 7 days after admission were excluded from the analyses of 8- to 90-day mortality.
**Log-rank test for equality.

CI indicated confidence interval; ED, emergency department; HR, hazard ratio.


64 SHOCK VOL. 51, No. 1 HOLLER ET AL.

septic shock, which has been reported in all areas of the world
where epidemiological studies of septic shock patients have
been conducted (25). The IR of septic shock has previously
been reported to be 31/100,000 pyar among in-hospital patients
(wards and ICUs) (26). Although the incidence of acute
myocardial infarction has decreased (27, 28), the incidence
of cardiogenic shock complicating ST-elevation myocardial
infarction (STEMI) paradoxically seem to be increasing
(29). Due to our local prehospital structure, patients suffering
STEMI are commonly triaged prehospitally and referred
directly for primary percutaneous coronary intervention and
thereby bypass the ED, which could explain the absence of an
increase in the IR of cardiogenic shock in our study.
Of notion was the decline in IR during 2008 followed by an
increase across several etiologies (Fig. 2). This trend could
FIG. 4. Kaplan–Meier curves illustrating overall 90-day survival partly be explained by a decrease of 9.6% (3,637) during 2000
across etiologies of shock. Below the curves are listed the number at risk
at corresponding intervals in survival time.
to 2008 of all adult ED contacts and a subsequent increase of
7.8% (2.897 contacts) from 2008 to 2011, and thereby a
concordantly proportional change in the annual frequency of
CI: 1.47–3.13]), had a higher hazard rate as compared to the shock cases (Fig. 5). Moreover, we hypothesize a possible
reference (non-septic shock) within 0 to 7 days. Conditional upon change in ED personal behavior concerning monitoring
surviving 8 days or more, only septic shock was a significant patients and increased awareness during 2009 to 2011 due to
etiologic independent predictor with an HR ¼ 1.66 (95% CI: the implementation of the ADAPT triage algorithm in 2009
1.14–2.42) as compared to the reference (Table 2). (especially among the older individuals).
Exploring mortality proportions revealed 90-day mortalities
of septic and cardiogenic shock of 56.2% and 52.3%, respec-
DISCUSSION
tively. These findings are in line with rates reported for septic
In this population-based cohort study, we explored the (26) and cardiogenic shock (30), although the later has
etiological characteristics of undifferentiated shock at presen- decreased (29). Non-septic shock had the lowest 90-day mor-
tation in the ED. Hypovolemic and septic shock were the most tality (22.6%). This group of patients was younger (Fig. 3) and
common etiologies followed by non-septic and cardiogenic consisted of patients suffering mainly of disorders related to
shock, whereas obstructive shock was a rare condition. glucose homeostasis, poisoning, epilepsy, and use of alcohol
The published studies investigating the etiology of shock in the (see Appendix 4, http://links.lww.com/SHK/A525). Finally, we
ED have been limited (3) while the frequency in the post-ED found the underlying etiology an independent predictor of
period has gained more attention. One-third of the ICU popula- mortality, especially in patients suffering septic shock.
tion has been reported to suffer shock (20). In a large randomized
trial by De Backer et al. (21), septic shock occurred in 62% of the Study strengths and limitations
patients, cardiogenic shock in 16%, hypovolemic shock in 16%, The strengths of this study were the large sample size and the
obstructive shock in 2%, whereas other types of distributive shock accurate linkage between healthcare registries. Due to the
counted 4%. As these studies often include patients from different Danish public healthcare system, it was possible to identify
settings, both the ED and local wards within the hospital, the
estimates could be confounded by the premise of patients sur-
viving to the ICU. These estimates are therefore a reflection of
this selection making extrapolation to the ED difficult. In ED
populations with undifferentiated symptomatic persistent hypo-
tension, Jones et al. (22) identified 43% to suffer septic shock,
15% cardiovascular shock, 28% severe dehydration, and 7%
toxicological causes. Less than 1% suffered anaphylaxis (22).
Accordingly, Arendts et al. (23) described septic shock in 28%,
cardiogenic in 27%, trauma in 14%, non-traumatic hemorrhage in
6%, and acute respiratory failure in 6%. Lastly, in a cohort of
hypotensive shock patients, Kheng et al. (24) found hypovolemic
shock 36%, septic shock 33%, cardiogenic shock 29%, and
anaphylactic shock in 2%. Altogether the studies suggest septic,
cardiovascular and hypovolemic shock to be common causes of
shock. These findings are in line with our results.
FIG. 5. Annual proportion of shock patients based on the annual
Trend analysis of IRs across etiologies in our study did not overall ED visits of adult patients (age 18 years). ED indicates emer-
exhibit significant trends, except for an increasing trend of gency department.
SHOCK JANUARY 2019 ETIOLOGY OF SHOCK IN THE EMERGENCY DEPARTMENT 65

all patients in the population-based registries, allowing com- TABLE 3. Cohort characteristics stratified on systolic blood pressure
thresholds
plete follow-up. Moreover, it was possible to follow each
individual patient event throughout the study period. The blood Systolic blood pressure threshold
pressure measurements were registered prospectively and as a >90 mm Hg SBP
routine documentation in the ED population. We used the first 100 mm Hg 90 mm Hg 100 mm Hg
contact with shock within the study period to minimize bias Variable n (%) n (%) n (%)
from repeated measurements. Furthermore, we excluded Male (1), Female (0)
patient with residency outside the catchment area and a previ- 0 723 (46.6) 427 (45.2) 296 (48.7)
ously reported admission with SBP  100 mm Hg in the years 1 830 (53.4) 518 (54.8) (312 (51.3)
1998 to 1999 to avoid possible overestimation of the incidence. Total 1,553 (100.0) 945 (100.0) 608 (100.0)
CCI (%)
We derived a definition for etiological characteristics of
0 477 (30.7) 283 (29.9) 194 (31.9)
shock based on hospital discharge data (see appendix 1, 1–2 589 (37.9%) 352 (37.2) 237 (39.0)
http://links.lww.com/SHK/A522), with a case definition that >2 487 (31.4%) 310 (32.8) 177 (29.1)
required both a diagnostic code indicating the etiology (e.g., Total 1,553 (100.0) 945 (100.0) 608 (100.0)
‘‘pneumococcal pneumonia’’ ¼ septic, ‘‘Myocardial infarc- Number of organ dysfunctions
1 1,160 (74.7%) 665 (70.4) 495 (81.4)
tion’’ ¼ cardiogenic) plus biochemical variables or SI  1 at
2 311 (20.0%) 218 (23.1) 93 (15.3)
presentation indicating organ failure. However, metabolic fail- 3þ 82 (5.3%) 62 (6.6) 20 (3.3)
ure was not included, as arterial punctures were not systemati- Total 1,553 (100.0) 945 (100.0) 608 (100.0)
cally collected and registered. Moreover, respiratory Etiology
frequencies and Glasgow Coma Scale were not consistently Nonseptic 363 (23.4) 194 (20.5) 169 (27.8)
Septic 423 (27.2) 254 (26.9) 169 (27.8)
registered in the electronic records, whereby organ failures
Cardiogenic 217 (14.0) 133 (14.1) 84 (13.8)
related to the respiratory system, and failure of the central Hypovolaemic 479 (30.8) 314 (33.2) 165 (27.1)
nervous system were not included. Due to the design of the Obstructive 14 (0.9) 10 (1.1) 4 (0.7)
study and data availability (electronic administrative data), we Other 57 (3.7) 40 (4.2) 17 (2.8)
are unfortunately not able to add the variables mentioned Total 1,553 (100.0) 945 (100.0) 608 (100.0)
Overall 7-day mortality
(arterial punctures, respiratory frequencies, use of vasopres-
Alive 1,191 (76.7) 683 (72.3) 508 (83.6)
sors, mechanical ventilation, etc.). Including a higher number Dead 362 (23.3) 262 (27.7) 100 (16.5)
of organ failures would not only increase the IR but could also Total 1,553 (100.0) 945 (100.0) 608 (100.0)
have a possible impact on the mortality estimates. Moreover, Overall 90-day mortality
we used SBP  100 mm Hg based on increasing evidence Alive 915 (58.9) 513 (54.3) 402 (66.1)
Dead 638 (41.1) 432 (45.7) 206 (33.9)
supporting a higher threshold (9, 31–33). Traditionally, hypo-
Total 1,553 (100.0) 945 (100.0) 608 (100.0)
tension is defined as SBP  90 mm Hg. Using the traditional
CCI indicates Charlson Comorbidity Index; SBP, systolic blood pressure.
definition would exclude 608 (39.2%) patients (90 > mm Hg
SBP  100 mm Hg) with shock from our cohort of which 6.4%
(100/1,553) died within 7 days and 13.3% (206/1,553) died the ED. In some instances, the clinician have omitted to order
within 90 days. Details of patients with SBP < 90 mm Hg are further investigations despite the fact that the patient had an
given in Table 3. We used discharge diagnosis information to SBP  100 mm Hg. Tables 4 and 5 aim to describe these
calculate the CCI, which means that for a comorbidity to be patients. The tables show that most of these patients are 70 years
recognized it had to require hospitalization. Comorbidities such or older, and that they are discharged directly from the ED. The
as diabetes could therefore be under reported. ethical question of whether or not it is acceptable to omit further
Importantly, we acknowledge the limitation in our estimates investigations of elderly patients is highly relevant and is
of cardiogenic shock as these patients are likely to be under- probably the explanation of why so many patients had no
estimated due to possible referral bias by the emergency further blood test ordered. The patients with missing laboratory
medical service operating in the prehospital setting during values are a result of clinical decisions between 2000 and 2011.
the period of observation. Moreover, the etiology of shock is This is a culture that changes with time and geography, and the
not always solely restricted to one type, but can overlap, due to data have to be interpreted as such.
the underlying heterogeneity, and patients admitted with one It is therefore possible that our local triage algorithm does not
type of shock can develop other types of shock (1). easily translate to other acute health care systems outside our
The study was a single-center, retrospective study from an ED, which limit the generalizability of the results. 3.5%
academic hospital. The large proportion of ED patients who (n ¼ 22) died in the ED upon arrival or shortly after (majority
were not eligible for study inclusion because SBP was not suffered cardiac arrest) explaining the rather excessive 24-h
measured at all (n ¼ 273,794) may be seen as an important mortality of ‘‘discharged’’ ED patients and the lack of blood
limitation. However, most of these patients suffered minor test performed in these patients (Tables 4 and 5).
complaints and based on a clinical judgment vital parameters Also the differences between our estimates reported here and
were not measured. results from the United States and other centers in Europe could
The data we have access to were limited to the type of reflect different ways of defining etiologies of shock, place of
observation and investigations that the clinician found relevant setting and variability in demographics and health systems.
to order—and register—for the specific patient who arrived at Moreover, the use of different algorithms of case identification
66 SHOCK VOL. 51, No. 1 HOLLER ET AL.

TABLE 4. Characteristics of hypotensive patients (non-shock) without blood tests performed (<24 h)
Overall mortality, n (%)
Specialty/Department* n (%) Age (mean) SBP (mean) Heart rate (mean) 24 h† 7 days 90 days

Emergency Department 627 (91.3) 54 (31–73) 94 (90–97) 72 (63–81) 22 (3.7) 29 (5.1) 57 (9.3)
Cardiology 4 (0.6) 77 (55–84) 85 (74–97) 61 (50–80) 4 (100.0) 4 (100.0) 4 (100.0)
Endocrinology 1 (0.2) 50 (50–50) 92 (92–92) 80 (80–80) 0 (0.0) 0 (0.0) 0 (0.0)
General Internal Medicine 11 (1.6) 43 (36–86) 98 (89–99) 78 (62–82) 0 (0.0) 1 (3.7) 2 (18.2)
General Surgery 3 (0.4) 61 (31–85) 94 (87–99) 78 (56–84) 0 (0.0) 0 (0.0) 0 (0.0)
Geriatriology 1 (0.2) 79 (79–79) 87 (87–87) 70 (70–70) 1 (100.0) 1 (100.0) 1 (100.0)
Heart, Pulmonary and vascular Surgery 4 (0.6) 50 (29–71) 95 (87–97) 60 (56–67) 2 (50.0) 2 (50.0) 2 (50.0)
Nephrology 1 (0.2) 41 (41–41) 88 (88–88) 49 (49–49) 0 (0.0) 0 (0.0) 0 (0.0)
Neurosurgery 2 (0.3) 79 (78–79) 90 (82–97) 75 (58–92) 1 (50.0) 2 (100.0) 2 (100.0)
Orthopedic Surgery 33 (4.8) 49 (34–73) 93 (89–97) 70 (64–78) 1 (3.0) 2 (6.1) 5 (15.2)
Total 687 (100.0) 55 (34–74) 94 (90–97) 72 (63–80) 31 (4.5) 41 (6.2) 73 (10.6)
*Two missing variables.

These 22 patients died while they were in the ED due to severe time dependent diseases.
ED indicates emergency department; SBP, systolic blood pressure.

TABLE 5. Mortality characteristics stratified on age of unadmitted patients in the ED


Overall mortality, n (%)
Age n (%) 24 h* 7 days 90 days

18–29 137 (21.9) 1 (0.7) 1 (0.7) 1 (0.7)


30–39 78 (12.4) 1 (1.3) 1 (1.3) 1 (1.3)
40–49 70 (11.2) 0 (0.0) 0 (0.0) 1 (1.4)
50–59 86 (13.2) 1 (1.2) 2 (2.9) 4 (4.7)
60–69 69 (11.0) 2 (2.9) 2 (2.9) 6 (8.7)
70–79 85 (13.6) 6 (7.1) 6 (7.1) 15 (17.7)
80þ 102 (16.3) 11 (10.8) 17 (16.7) 29 (28.4)
Total 627 (100.0) 22 (3.5) 29 (4.6) 57 (9.1)
*The patients died while they were in the emergency department due to severe time-dependent diseases.
ED indicates emergency department.

could explain the difference in estimates (e.g., sepsis-specific REFERENCES


discharge codes or discharge codes of infection and organ 1. Vincent JL, De Backer D: Circulatory shock. N Engl J Med 369(18):1726–1734,
dysfunction (34)). 2013.
Finally, the increasing incidence of septic shock could be, in 2. Vincent JL, Ince C, Bakker J: Clinical review: circulatory shock—an update: a
tribute to Professor Max Harry Weil. Crit Care 16(6):239, 2012.
part, a consequence of the initiation and implementation of the 3. Holler JG, Bech CN, Henriksen DP, Mikkelsen S, Pedersen C, Lassen AT:
Surviving Sepsis Campaign in 2004 (35) (Denmark, 2006) as Nontraumatic hypotension and shock in the emergency department and the
focus has increased on this condition. prehospital setting, prevalence, etiology, and mortality: a systematic review.
PloS One 10(3):e0119331, 2015.
4. Holler JG, Henriksen DP, Mikkelsen S, Rasmussen LM, Pedersen C, Lassen AT:
Shock in the emergency department: a 12 year population based cohort study.
CONCLUSION
Scand J Trauma Resusc Emerg Med 24:87, 2016.
Hypovolemic (30.8%) and septic shock (27.2%) are common 5. Accident Analysis Group, Odense University Hospital, 5000 Odense C, Denmark.
6. Mikkelsen S, Kruger AJ, Zwisler ST, Brochner AC: Outcome following physi-
etiological characteristics of shock at presentation in the ED, cian supervised prehospital resuscitation: a retrospective study. BMJ Open
followed by non-septic (23.4%) and cardiogenic shock (14.0%), 5(1):e006167, 2015.
all increasing by age. Obstructive shock is a rare condition 7. Lindberg SO, Lerche la Cour J, Folkestad L, Hallas P, Brabrand M: The use of
triage in Danish emergency departments. Danish Med Bull 58(10):A4301, 2011.
(0.9%). Septic shock increased during the study period. The
8. Turchin A, Kolatkar NS, Grant RW, Makhni EC, Pendergrass ML, Einbinder JS:
7-, and 90-day mortality of septic, non-septic, cardiogenic, and Using regular expressions to abstract blood pressure and treatment intensifica-
hypovolemic shock was 30.3%/56.2%, 12.7%/22.6%, 34.6%/ tion information from the text of physician notes. J Am Med Informat Assoc
52.3%, and 19.2%/36.8%, respectively. Independent etiologies 13(6):691–695, 2006.
9. Kristensen AK, Holler JG, Mikkelsen S, Hallas J, Lassen A: Systolic blood
of mortality were primarily septic and cardiogenic shock within 0 pressure and short-term mortality in the emergency department and prehospital
to 7 days and septic shock within 8 to 90 days. setting: a hospital-based cohort study. Crit Care 19(1):158, 2015.
10. Schmidt M, Pedersen L, Sorensen HT: The Danish Civil Registration System as
a tool in epidemiology. Eur J Epidemiol 29(8):541–549, 2014.
ACKNOWLEDGMENTS 11. Lynge E, Sandegaard JL, Rebolj M: The Danish National Patient Register. Scand
The authors would like to thank Niels Ibsgaard Agerbek, MSCS, for professional J Public Health 39(7 Suppl):30–33, 2011.
help with all data extraction, as well as the Injury Prevention Group, Department of 12. Andersen TF, Madsen M, Jorgensen J, Mellemkjoer L, Olsen JH, The Danish
Orthopedic Surgery, Odense University Hospital, Odense, Denmark for help with National Hospital Register. A valuable source of data for modern health
aggregated data on annual ED visits during the period 2000 to 2011. sciences. Danish Med Bull 46(3):263–268, 1999.
SHOCK JANUARY 2019 ETIOLOGY OF SHOCK IN THE EMERGENCY DEPARTMENT 67
13. Rady MY, Smithline HA, Blake H, Nowak R, Rivers E: A comparison of the 25. Martin GS: Sepsis, severe sepsis and septic shock: changes in incidence,
shock index and conventional vital signs to identify acute, critical illness in the pathogens and outcomes. Expert Rev Anti Infect Ther 10(6):701–706, 2012.
emergency department. Ann Emerg Med 24(4):685–690, 1994. 26. Esteban A, Frutos-Vivar F, Ferguson ND, Peñuelas O, Lorente JA, Gordo F,
14. Rady MY, Nightingale P, Little RA, Edwards JD: Shock index: a re-evaluation in Honrubia T, Algora A, Bustos A, Garcı́a G, et al.: Sepsis incidence and outcome:
acute circulatory failure. Resuscitation 23(3):227–234, 1992. contrasting the intensive care unit with the hospital ward. Crit Care Med
15. Weil MH, Shubin H: Proposed reclassification of shock states with special 35(5):1284–1289, 2007.
reference to distributive defects. Adv Exp Med Biol 23(0):13–23, 1971. 27. Yeh RW, Sidney S, Chandra M, Sorel M, Selby JV, Go AS: Population trends in
16. Charlson ME, Pompei P, Ales KL, MacKenzie CR: A new method of classifying the incidence and outcomes of acute myocardial infarction. N Engl J Med
prognostic comorbidity in longitudinal studies: development and validation. J 362(23):2155–2165, 2010.
Chronic Dis 40(5):373–383, 1987. 28. Schmidt M, Jacobsen JB, Lash TL, Botker HE, Sorensen HT: 25 year trends in
17. http://www.statistikbanken.dk/FOLK1, http://www.statistikbanken.dk/BEF6, first time hospitalisation for acute myocardial infarction, subsequent short and
http://www.statistikbanken.dk/BEF607. Accessed January 2016. long term mortality, and the prognostic impact of sex and comorbidity: a Danish
18. Vandenbroucke JP, von Elm E, Altman DG, Gøtzsche PC, Mulrow CD, Pocock nationwide cohort study. BMJ 344:e356, 2012.
SJ, Poole C, Schlesselman JJ, Egger M, STROBE Initiative. Strengthening the 29. Kolte D, Khera S, Aronow WS, Mujib M, Palaniswamy C, Sule S, Jain D, Gotsis
Reporting of Observational Studies in Epidemiology (STROBE): explanation W, Ahmed A, Frishman WH, et al.: Trends in incidence, management, and
and elaboration. Epidemiology 18(6):805–835, 2007. outcomes of cardiogenic shock complicating ST-elevation myocardial infarction
19. Landis JR, Koch GG: The measurement of observer agreement for categorical in the United States. J Am Heart Assoc 3(1):e000590, 2014.
data. Biometrics 33(1):159–174, 1977. 30. Babaev A, Frederick PD, Pasta DJ, Every N, Sichrovsky T, Hochman JS: Trends
20. Sakr Y, Reinhart K, Vincent JL, Sprung CL, Moreno R, Ranieri VM, De Backer in management and outcomes of patients with acute myocardial infarction
D, Payen D: Does dopamine administration in shock influence outcome? Results complicated by cardiogenic shock. JAMA 294(4):448–454, 2005.
of the Sepsis Occurrence in Acutely Ill Patients (SOAP) Study. Crit Care Med 31. Jones AE, Yiannibas V, Johnson C, Kline JA: Emergency department hypoten-
34(3):589–597, 2006. sion predicts sudden unexpected in-hospital mortality: a prospective cohort
21. De Backer D, Biston P, Devriendt J, Madl C, Chochrad D, Aldecoa C, Brasseur A, study. Chest 130(4):941–946, 2006.
Defrance P, Gottignies P, Vincent JL, SOAP II Investigators. Comparison of dopamine 32. Seymour CW, Cooke CR, Heckbert SR, Copass MK, Yealy DM, Spertus JA, Rea
and norepinephrine in the treatment of shock. N Engl J Med 362(9):779–789, 2010. TD: Prehospital systolic blood pressure thresholds: a community-based out-
22. Jones AE, Tayal VS, Sullivan DM, Kline JA: Randomized, controlled trial of comes study. Acad Emerg Med 20(6):597–604, 2013.
immediate versus delayed goal-directed ultrasound to identify the cause of 33. Eastridge BJ, Salinas J, McManus JG, Blackburn L, Bugler EM, Cooke WH,
nontraumatic hypotension in emergency department patients. Crit Care Med Convertino VA, Wade CE, Holcomb JB: Hypotension begins at 110 mm Hg: redefining
32(8):1703–1708, 2004. ‘‘hypotension’’ with data. J Trauma 63(2):291–297, 2007. discussion 7-9.
23. Arendts G, Stone SF, Fatovich DM, van Eeden P, MacDonald E, Brown SG: 34. Henriksen DP, Laursen CB, Jensen TG, Hallas J, Pedersen C, Lassen AT:
Critical illness in the emergency department: lessons learnt from the first Incidence rate of community-acquired sepsis among hospitalized acute medical
12 months of enrolments in the Critical Illness and Shock Study. Emerg Med patients-a population-based survey. Crit Care Med 43(1):13–21, 2015.
Australas 24(1):31–36, 2012. 35. Dellinger RP, Carlet JM, Masur H, Gerlach H, Calandra T, Cohen J, Gea-
24. Kheng CP, Rahman NH: The use of end-tidal carbon dioxide monitoring in Banacloche J, Keh D, Marshall JC, Parker MM, et al.: Surviving Sepsis
patients with hypotension in the emergency department. Int J Emerg Med Campaign guidelines for management of severe sepsis and septic shock. Crit
5(1):31, 2012. Care Med 32(3):858–873, 2004.

You might also like