Article

ED crowding is associated with inpatient mortality among critically ill patients admitted via the ED: post hoc analysis from a retrospective study

a b s t r a c t

Background: Adverse effects of emergency department (ED) crowding among critically ill patients are not well known. Objectives: We evaluated the association between ED crowding and Inpatient mortality among critically ill patients ad- mitted via the ED, and analyzed subsets of patients according to admission diagnosis.

Methods: We performed a post hoc analysis using data from a previous retrospective study. We enrolled admitted pa- tients via the ED with an initial systolic blood pressure of 90 mm Hg or lower when presenting to the ED. The ED oc- cupancy ratio was used as a measure of crowding. The primary outcome was inpatient mortality. Multivariable logistic regression models adjusted for potential confounding variables were constructed for the entire cohort and for subsets according to admission diagnosis (infection, cardiac and vascular disease, trauma, gastrointestinal bleeding, and other factors).

Results: A total of 1801 patients were enrolled, with a mortality rate of 14.6% (262 patients).

The mortality rate by ED occupancy ratio quartile was 9.7% for the first quartile, 15.9% for the second quartile, 18.2% for the third quartile, and 14.4% for the fourth quartile. This resulted in adjusted odds ratios of 1.95, 2.51, and 1.93 and corresponding 95% confidence intervals of 1.23-3.12, 1.58-3.99, and 1.21-3.09 for the second, third, and fourth quartiles, respectively, compared with the first quartile. The effect of ED crowding was highest in the trauma subset, followed by the infection subset, whereas ED crowding did not appear to have any effect on the cardiac and vascular disease subsets.

Conclusion: Emergency department crowding was associated with increased inpatient mortality among critically ill pa- tients admitted via the ED.

(C) 2015

Introduction

Background

Emergency department (ED) crowding has been a problem in health care systems worldwide, and concerns about ED crowding are increas- ing [1,2]. Previous studies have shown an association between ED crowding and an unfavorable outcome in general ED patients [3,4] and admitted patients [5,6].

However, to the best of our knowledge, no previous research has di- rectly evaluated the association between ED crowding and critically ill

? Source of support: None.

?? All authors have no interest of conflict.

? Author contributions: conception and design: S.J., T.J.; analysis and interpretation: S.J.,

B.P.; drafting the manuscript for important intellectual content: S.J.; review: J.B.L, Y.H.J, J.Y.

* Corresponding author. Department of Emergency Medicine, Research Institute of Clin- ical Medicine of Chonbuk National University, 567 Baekje-daero, deokjin-gu, Jeonju-si, Jeollabuk-do 561-756, Republic of Korea.

patients. Thus far, there have been some studies that have shown an as- sociation between delayed admission to the intensive care unit (ICU) and poor outcome for critically ill patients [7-9]. However, these are out of the scope of ED crowding. Because delayed admission, which is nearly same as prolongED boarding time, is attributed to Hospital crowding, it remains uncertain whether delayed admission accom- panies ED crowding. For example, if there is no available bed in an ICU and ward (full hospital occupancy), then the critically ill patient has to be boarded in the ED where there are sometimes few patients in the ED (low ED occupancy). In other words, Prolonged ED stay is limited in its use as an ED crowding measurement.

Recently, the authors reported the harmful effect of ED crowding on early mortality among general ED patients [10]. Using this data set, we evaluated the effect of ED crowding on critically ill patients admitted via the ED. Associations between ED crowding and procedural times were also investigated.

If there is an association between ED crowding and outcome of crit- ically ill patients, then medical teams are more likely to review their ED

http://dx.doi.org/10.1016/j.ajem.2015.08.004

0735-6757/(C) 2015

process and hospital administrators may modify the hospital policy. Furthermore, determining how ED crowding affects patients with spe- cific diagnoses could help staff managers to focus their limited resources on specific conditions.

For critically ill patients, the timely completion of procedures, such as advanced airway management, vasopressor or inotropic use, central line catheterization, and blood transfusions, are very important. Thus, investigations of the association between ED crowding and these proce- dures in critically ill patients may reveal the mechanism of ED crowding.

Objectives

The purpose of the present study was to explore the potential effect of ED crowding on critically ill patients admitted via the ED. In addition, we explored the effect of ED crowding on mortality in subsets of pa- tients categorized according to the specific disease condition including infection, cardiac and vascular disease, trauma, gastrointestinal bleed- ing, and others factors. Finally, we investigated the association between ED crowding and the procedure time for intubation, vasopressor or ino- tropic use, central line catheterization, and blood transfusions.

Materials and methods

Study design and participants

This study is a post hoc analysis using data from a previous retro- spective cohort study of general ED patients seen between January 1, 2009, and December 31, 2010. Of the approximately 70000 patients seen in that time frame, 54410 were 15 years or older. From this data set, we selected patients who were admitted to the study hospital via the ED and had systolic blood pressure (SBP) at or less than 90 mm Hg (n = 1846); these patients were designated as critically ill patients. Among the patients, we excluded patients who were cardiac arrest pa- tients upon arrival to the ED (n = 45); thus, 1801 patients were in- cluded in the analysis.

This study was approved by the institutional review board of the study hospital, and a waiver for informed consent was obtained. The study hos- pital is a 1000-bed urban academic tertiary care hospital equipped with 42 licensed ED beds during the study period. The study hospital’s annual consensus was 35 000. There were 3 shifts in the study hospital, from 7 AM to 3 PM (day), 3 PM to 11 PM (evening), and 11 PM to 7 AM (night). The same number of Emergency doctors (1 board-certificated emergency medicine physician and 3 emergency medicine residents) and emergency medical technicians (n = 2) worked in each shift. There were 9 nurses during the day, 10 during the evening, and 8 during the night shifts.

Data collection and processing

Baseline variables were extracted from previously constructed data including age; sex; use of emergency medical service (EMS); transferred cases; weekend and holiday visits; shifts; triage acuity; vital sings upon ED arrival, such as the mean blood pressure, pulse rate, respiratory rate, body temperature, and mental status using the AVPU scale (alert, ver- bal, painful, unresponsive scale); ICU admission; surgical intervention; ED occupancy ratio (EDOR); ED length of stay (ED LOS); hospital LOS; date and time of ED arrival; and survival status upon discharge.

In addition, we obtained data on comorbidities (malignancy, liver cirrhosis, and chronic kidney disease), admission diagnostic classifica- tion (infection, cardiac and vascular disease, trauma, gastrointestinal bleeding, and others), and date and time of the initiation of various pro- cedures (intubation, vasopressor or inotropic use, successful central line, and any blood transfusion) performed during the ED stay via med- ical record reviews by trained abstractors following the guidelines rec- ommended by Gilbert et al [11]. The procedural time was calculated from the date, hour, and minute of ED arrival for each procedure.

We used EDOR as a measure of ED crowding. The EDOR is a ratio of the total number of ED patients to the number of licensed ED beds. In a previous study, a computer program was used to calculate the EDOR for every patient based on the patient’s ED arrival and discharge time.

Outcome measures

The primary study outcome was inpatient mortality. Mortality cate- gorized according to admission diagnosis (infection, cardiac and vascu- lar disease, trauma, gastrointestinal bleeding, and others) was set as secondary outcomes.

Primary data analysis

All continuous data were presented as the mean and SD. In addition, we used interquartile ranges to show more accurate data distributions. All discrete data were presented as both counts and percentages. Logis- tic regression analyses results were presented as odds ratios (ORs) with a 95% confidence interval (CI). Statistical significance was defined as a 2- sided P b .05.

The Kruskal-Wallis test was used to compare nonparametric variables for each EDOR quartile, and 1-way analysis of variance (ANOVA) with a Scheffe, Bonferroni, Sidak posttest was used for parametric variables.

Logistic regression analyses were performed to control for poten- tially confounding factors. Logistic regression analyses, which are gener- ally applied to predict whether the patient experiences an outcome based on observed characteristics and provide unbiased results adjusted for other covariates, were used to present ORs with a 95% CI. The EDOR variable was used as the quartile. The variables included age, sex, EMS transport, transferred case, day of the week (weekend or holiday vs nonweekend), shift (day, evening, or night), triage acuity (immediate, emergent, urgent, semiurgent, and nonurgent), visit cause (injury or noninjury), comorbidities (malignancy, liver cirrhosis, or chronic kid- ney disease), surgical intervention, mean arterial pressure, pulse rate, respiratory rate, body temperature, mental status (AVPU scale), whether to admit (ED discharge, ward admission, or ICU admission) and ED LOS (by quartile).

We calculated the mean time from ED arrival to the initiation of varioUS procedures and estimated whether the mean time differed according to ED crowding. The EDOR was analyzed by quartile as categorical variables.

All analyses were performed using STATA 11.1 (StataCorp LP, College Station, Texas) and SAS 9.1 (SAS Institute Inc, Cary, North Carolina).

Results

Study subjects characteristics

We enrolled 1846 admitted patients via the ED who presented with SBP at or less than 90 mm Hg from a prior data set. A total of 1801 patients were included in the analysis after excluding 45 cardiac arrest patients upon arrival to the ED. There were minimal missing data on covariates (7 missing EMS use, 2 missing transferred cases, 6 missing triage level).

Table 1 presents baseline characteristics for the entire cohort and each quartile. Most of the demographic variables were not significantly different among quartiles, except visits during the weekends and holi- days. Physiological variables, admission diagnosis, and the disposition of patients also showed no significant differences. Scheffe, Bonferroni, and Sidak post hoc analyses also revealed no between-group differ- ences. At the time of admission, 552 patients (30.6%) were diagnosed as having infectious diseases, 214 (11.9%) patients as having cardiac and vascular disease, 306 (17.0%) patients as having traumatic injuries, and 235 (13.0%) patients as having gastrointestinal bleeding. The mean (SD) EDOR was 1.26 (0.27; interquartile range, 1.07-1.40). Importantly, the ED LOS and total hospital LOS were not different among quartiles.

There were 262 (14.6%) deaths. The mortality by EDOR quartiles for all patients and for subsets categorized by admission diagnosis are

Table 1

Baseline characteristics of enrolled patients

Total

1st quartile

2nd quartile

3rd quartile

4th quartile

P

No.

1801 (100)

477 (100)

427 (100)

453 (100)

444 (100)

Age (y)

59.6 +- 17.8

59.0 +- 18.6

59.3 +- 17.5

59.3 +- 17.4

60.1 +- 17.7

.32

Sex (% of male)

1010 (56.1)

266 (55.8)

231 (54.1)

248 (54.8)

265 (59.7)

.34

EMS transport (%)

401 (22.4)

127 (26.7)

97 (22.8)

95 (21.1)

82 (18.5)

.03

Transferred (%)

838 (46.6)

220 (46.3)

190 (44.5)

220 (48.6)

208 (46.9)

.69

Visit during weekend and holiday (%)

543 (30.1)

191 (40.0)

150 (35.1)

109 (24.1)

93 (21.0)

b.01

Shift (%)

Day (7 AM-3 PM)

767 (42.6)

158 (33.1)

187 (43.8)

209 (46.1)

213 (48.0)

b.01

Evening (3 PM-11 PM)

713 (39.6)

188 (39.4)

161 (37.7)

186 (41.1)

178 (40.1)

.78

Night (11 PM-7 AM)

321 (17.8)

131 (27.5)

79 (18.5)

58 (12.8)

53 (11.9)

b.01

Triage acuity (%)

Immediate

28 (1.6)

7 (1.5)

11 (2.6)

6 (1.3)

4 (0.9)

.23

Emergent

52 (2.9)

11 (2.3)

13 (3.1)

12 (2.7)

16 (3.6)

.68

Urgent

823 (45.7)

209 (44.0)

198 (46.6)

215 (47.6)

201 (45.4)

.72

Semiurgent

656 (36.4)

180 (37.9)

151 (35.5)

157 (34.7)

168 (37.9)

.67

Nonurgent

236 (13.1)

68 (14.3)

52 (12.2)

62 (13.7)

54 (12.2)

.72

Missing

6 (0.3)

2 (0.4)

2 (0.5)

1 (0.2)

1 (0.2)

Malignancy (%)

382 (21.2)

100 (21.0)

94 (22.0)

93 (20.5)

95 (21.4)

.96

Liver cirrhosis (%)

156 (8.7)

39 (8.2)

38 (8.9)

40 (8.8)

39 (8.8)

.98

Chronic kidney disease (%)

87 (4.8)

20 (4.2)

20 (4.7)

24 (5.3)

23 (5.2)

.86

MBP (mm Hg)

59.4 +- 16.2

59.4 +- 16.2

59.1 +- 17.4

60.2 +- 14.3

58.8 +- 16.9

.61

Pulse rate (beats/min)

89.1 +- 24.6

89.8 +- 24.8

89.4 +- 25.6

89.2 +- 24.1

87.9 +- 23.9

.70

Respiratory rate (beats/min)

20.3 +- 3.7

20.1 +- 3.4

20.5 +- 3.9

20.4 +- 3.8

20.2 +- 3.5

.18

Body temperature (?C)

36.5 +- 1.2

36.5 +- 0.9

36.5 +- 1.9

36.6 +- 0.9

36.6 +- 0.9

.52

Mental status (%)

Alert

1.524 (84.6)

391 (82.0)

369 (86.4)

385 (85.0)

379 (85.4)

.28

Verbal

121 (6.7)

32 (6.7)

27 (6.3)

34 (7.5)

28 (6.3)

.88

Pain

93 (5.2)

31 (6.5)

22 (5.2)

21 (4.6)

19 (4.3)

.44

Unresponsive

63 (3.5)

23 (4.8)

9 (2.1)

13 (2.9)

18 (4.1)

.12

Admission diagnosis (%) Infection

552 (30.6)

138 (28.9)

133 (31.2)

149 (32.9)

132 (29.7)

.58

Cardiac and vascular disease

214 (11.9)

47 (9.9)

63 (14.8)

51 (11.3)

53 (11.9)

.14

Trauma

306 (17.0)

91 (19.1)

72 (16.9)

76 (16.8)

67 (15.1)

.45

Gastrointestinal bleeding

235 (13.0)

71 (14.9)

48 (11.2)

63 (13.9)

53 (11.9)

.33

Others

494 (27.4)

130 (27.3)

111 (26.0)

114 (25.2)

139 (31.3)

.17

ICU admission (%)

533 (29.6)

138 (28.9)

141 (33.0)

125 (27.6)

129 (29.1)

.33

Surgical intervention (%)

228 (12.7)

64 (13.4)

54 (12.7)

57 (12.6)

53 (11.9)

.93

EDOR

1.26 +- 0.27

0.94 +- 0.11

1.17 +- 0.05

1.34 +- 0.05

1.61 +- 0.14

b.01

ED LOS of all patients (h)

24.5 +- 27.0

23.2 +- 26.4

24.1 +- 29.5

24.6 +- 25.2

26.0 +- 27.1

.45

Total hospital LOS (day)

15.9 +- 21.9

17.0 +- 26.0

14.9 +- 20.9

15.1 +- 19.3

16.4 +- 20.3

.40

In-hospital mortality cases (%)

262 (14.55)

47 (9.9)

70 (16.4)

79 (17.4)

66 (14.9)

b.01

Abbreviation: MBP, mean blood pressure.

One-way ANOVA test was used for continuous variables, and Kruskal-Wallis rank test was used for categorical variables.

shown in Fig. 1. For the entire cohort, the mortality was 9.7% for the first quartile of EDOR, 15.9% for the second quartile, 18.2% for the third quar- tile, and 14.4% for the fourth quartile. Mortality showed a stepwise in- crease until the third quartile, but it decreased in the last quartile. Mortality was significantly higher in the second and third quartiles compared with the first quartile. There were 91 deaths (16.5%) in the in- fection subset, 26 deaths (12.1%) in the cardiac and vascular diseases subset, 41 deaths (13.3%) in the trauma subset, 16 deaths (6.8%) in the gastrointestinal bleeding subset, and 88 deaths (17.8%) in the “other” subset. Among the infection subset, mortality was significantly higher in the second and fourth quartiles compared with the first quartile. Among the trauma subset, mortality was significantly higher in the third quartile than in the first quartile. For each subset, mortality generally increased by the EDOR until the third or fourth quartiles, except for the gastrointestinal bleeding subset, which did not demonstrate a pattern.

The results of the logistic regression analysis are presented in Table 2. The unadjusted model revealed that the EDOR in the second, third, and fourth quartiles was associated with an increased inpatient death com- pared with the first quartile. After controlling for potential confounders, the EDOR in the second, third, and fourth quartiles remained a signifi- cant factor for increased inpatient death (adjusted ORs [AORs], 1.95 [95% CI, 1.23-3.12], for the second quartile; 2.51 [1.58-3.99] for the third quartile; and 1.93 [1.21-3.09] for the fourth quartile). Age, concur- rent malignancy, mean arterial pressure, mental status, admission diag- nosis, and ICU admission were also significant factors after controlling

for covariates. The ED LOS was not associated with inpatient mortality after controlling for confounding factors.

A multivariable logistic regression model was constructed according to each admission diagnosis (infection, cardiac and vascular disease, trauma, gastrointestinal bleeding, and others; Table 3). However, the lo- gistic regression model could not be constructed for the gastrointestinal bleeding subset due to the small sample size (16/235 patients [6.8%]). Emergency department crowding was significantly associated with in- patient death in the trauma subset (AORs, 4.72 [95% CI, 1.17-19.07], for the second quartile, and 5.64 [1.46-21.70] for the third quartile). In the infection subset, the AOR was 3.38 (95% CI, 1.33-8.56) for the second quartile, 2.78 (95% CI, 1.10-7.07) for the third quartile, and 4.37 (95% CI, 1.73-11.05) for the fourth quartile. In the “other” subset, ED crowding was significantly associated with inpatient mortality for the second quartile (AOR, 2.47 [95% CI, 1.07-5.74]). However, among the cardiac and vascular disease subset, there was no significant association be- tween ED crowding and inpatient mortality. The ED LOS was not associ- ated with inpatient mortality in any of the subsets.

The procedural time according to ED crowding is shown in Fig. 2. In

the trauma subset, the time to central line catheterization was signifi- cantly higher for EDOR in the fourth quartile compared with the first quartile. In the gastrointestinal bleeding subset, the time to transfusion was significantly higher for the EDOR in the fourth quartile compared with the first quartile. Except for these instances, there were no signifi- cant differences or specific patterns among the EDOR quartiles.

Fig. 1. Mortality for the total patient cohort and each subset according to admission diagnosis. The numbers in parentheses indicate the number of deaths and the number of patients in the relevant quartile. The Kruskal-Wallis test and 1-way ANOVA were used when appropriate. *P b .05 and **P b .01 when compared with the first quartile.

Regression analysis also showed no association between the EDOR and procedural time, regardless of whether EDOR was considered a contin- uous or categorical variable.

Discussion

In the present study, ED crowding measured using the EDOR was as- sociated with increased inpatient death among critically ill patients ad- mitted via the ED. Emergency department crowding had the greatest effect on the trauma subset, followed by the infection subset, and did not appear to affect the cardiac and vascular disease subset. Intubation, vasopressor or inotropic use, central line catheterization, and blood transfusion were also not associated with ED crowding for the entire co- hort, but did affect specific subsets.

Several studies have previously reported that ED crowding is associ-

ated with increased mortality in general ED patients [3,4]. However, this finding did not indicate that each ED patient was equally affected by ED crowding. Rather, ED crowding may have a greater effect on patients under specific conditions. To the best of our knowledge, this is the first study to demonstrate a direct association between ED crowding and inpatient mortality among critically ill ED patients. The OR for ED crowding in this study was 1.95 (95% CI, 1.23-3.12) for the second quar- tile, 2.51 (95% CI, 1.58-3.99) for the third quartile, and 1.93 (95% CI, 1.21- 3.09) for the fourth quartile. These ORs were greater than those we ob- served in a previous study focused on the general ED population, with ORs ranging from 1.13 to 1.20 [10]. A previous study also reported an OR of 1.05 (95% CI, 1.02-1.08) for inpatient death among patients who were admitted via the ED, but were not necessarily critically ill [3].

A subgroup analysis showed that susceptibility to ED crowding varied according to diagnosis. Trauma patients were most affected by ED crowding, followed by patients with infections. For the trauma subset, the AOR was 5.64 (95% CI, 1.46-21.7) for the EDOR in the third quartile. Furthermore, for the infection subset, the AOR was 4.37 (95% CI, 1.73- 11.05) for the EDOR in the fourth quartile. Assuming that patients were treated the same across all quartiles, this finding suggests that nearly 70% of the deaths might have been prevented by addressing ED crowding. Our results were consistent with those obtained in previous studies. Begley et al [12] reported a higher mortality rate for severe trauma pa- tients who were transferred from another hospital. Intriguingly, no

previous studies have focused on infectious patients in the ED. However, ED crowding was associated with higher mortality in pneumonia pa- tients, which consisted of a substantial portion of the infection subset [13]. Interestingly, cardiac and vascular disease patients were not signifi- cantly affected by ED crowding, although patients in this subset may suffer acute myocardial infarction or other health conditions that re- quire urgent attention. We assume that ED crowding may not affect this subset because there are well-established treatment guidelines that clearly state the need for timely treatment. However, this study did not analyze ED crowding and time to thrombolysis, although some previous studies have reported that ED crowding prolonged time to thrombolysis in Acute myocardial infarction patients [14,15], whereas

another study reported opposite findings [16].

In the present study, mortality in the entire cohort increased until the third quartile and then decreased in the fourth quartile. Harmful ef- fects of ED crowding, as measured by AORs, showed a similar incremen- tal increase until the third quartile before decreasing in the fourth quartile. This trend was also observed in the subset analyses. In a previ- ous study [10], the overall mortality for all ED patients was 3.27% for the first quartile, 3.72% for the second quartile, 3.67% for the third quartile, and 3.90% for the fourth quartile. The overall mortality in a previous study was calculated as the sum of inpatient mortality (death after hos- pital admission) and ED mortality (death during ED stay). Thus, we as- sumed that mortality would be considerably high in patients who were not admitted during an ED stay when the EDOR was in the fourth quar- tile. Emergency department physicians and policy makers should be aware of this possibility.

In addition, we found that ED LOS was not statistically different ac- cording to EDOR quartiles. This finding supports our hypothesis that prolonged ED stay has limited use as an ED crowding measurement be- cause ED LOS reflects hospital crowding and not ED crowding itself. Thus, EDOR, which was applied in this study as an index of ED crowding, is more effective in estimating ED crowding as an important factor for inpatient mortality among critically ill patients admitted via the ED.

Researchers previously described how ED crowding affects patient outcome by focusing on time to treatment. Time to first antibiotics was delayed in patients with pneumonia [12,13,15], time to thromboly- sis was delayed in patients with acute myocardial infarction [14,17], and time to resuscitation bundle was delayed in patients with severe sepsis

Table 2

Logistic regression analysis for inpatient mortality

Unadjusted OR (95% CI, P)

AOR (95% CI, P)

Recently, Hong et al [20] reported an association between ED crowding and delay in resuscitation effort among patients who entered the resuscitation room and underwent resuscitative procedures. They found that delays in resuscitation efforts were associated with higher

EDOR 1.54 (0.94-2.51, .086) 1.89 (1.06-3.39, .032)

1st quartile (b 1.07) Reference Reference

2nd quartile (1.07-1.26) 1.79 (1.21-2.66, .004) 1.95 (1.23-3.12, .005)

3rd quartile (1.26-1.40) 1.93 (1.31-2.85, .001) 2.51 (1.58-3.99, b.001)

4th quartile (>= 1.40) 1.60 (1.07-2.38, .021) 1.93 (1.21-3.09, .006)

Age

1.01 (1.00-1.02, .003)

1.01 (1.00-1.02, .058)

Male sex

1.30 (0.99-1.69, .059)

1.17 (0.85-1.62, .326)

EMS transport

1.23 (0.91-1.67, .176)

1.06 (0.68-1.64, .796)

Transferred

1.07 (0.83-1.40, .596)

0.86 (0.59-1.25, .429)

Visit during weekend and

0.98 (0.74-1.30, .885)

1.16 (0.82-1.63, .401)

holiday Shift

Day Reference Reference

Evening 1.05 (0.79-1.39, .749) 1.01 (0.72-1.43, .947)

Night 0.57 (0.37-0.87, .009) 0.60 (0.36-1.01, .056)

Triage acuity

Immediate 0.49 (0.16-1.47, .202) 2.56 (0.70-9.42, .157)

Emergent 0.18 (0.05-0.60, .005) 0.32 (0.08-1.25, .101)

Urgent 0.39 (0.27-0.55, b.001) 0.92 (0.56-1.49, .722)

Semiurgent 0.52 (0.36-0.75, b.001) 0.82 (0.52-1.30, .398)

Nonurgent Reference Reference

Malignancy 2.70 (2.04-3.57, b.001) 4.57 (3.07-6.80, b.001)

Liver cirrhosis 1.26 (0.81-1.94, .307) 1.65 (0.92-2.96, .095)

Chronic kidney disease 1.57 (0.92-2.69, .098) 1.45 (0.74-2.84, .275)

Mean arterial pressure 0.97 (0.97-0.98, b.001) 0.99 (0.98-1.00, .002)

Pulse rate 1.02 (1.01-1.02, b.001) 1.01 (1.01-1.02, .095)

Respiratory rate 1.05 (1.01-1.09, .007) 1.04 (1.00-1.07, .067)

Body temperature 0.94 (0.86-1.03, .202) 0.91 (0.83-1.01, .083)

Mental status

Alert Reference Reference

Verbal 2.87 (1.87-4.41, b.001) 2.25 (1.33-3.79, .002)

Pain 3.47 (2.18-5.53, b.001) 2.08 (1.13-3.83, .019)

Unresponsiveness 4.71 (2.77-8.03, b.001) 2.57 (1.25-5.28, .010)

Admission diagnosis

Infection 0.91 (0.66-1.26, .569) 0.78 (0.52-1.18, .236)

Cardiac and vascular disease 0.64 (0.40-1.02, .061) 0.29 (0.16-0.52, b.001)

Trauma 0.71 (0.48-1.07, .100) 0.38 (0.21-0.69, .002)

Gastrointestinal bleeding 0.34 (0.19-0.59, b.001) 0.50 (0.26-0.95, .034) Others Reference Reference

ICU admission 5.41 (4.10-7.14, b.001) 6.76 (4.57-10.01, b.001)

Surgical intervention 1.25 (0.86-1.81, .242) 1.70 (0.99-2.90, .055) ED LOS

1st quartile (b 5.9 h)

Reference

Reference

2nd quartile (5.9-24.5 h)

0.65 (0.46-0.93, .017)

0.70 (0.44-1.10, .126)

3rd quartile (24.5-31.6 h)

0.54 (0.38-0.79, .001)

0.71 (0.44-1.15, .164)

4th quartile (>= 31.6 h)

0.60 (0.42-0.86, .006)

0.76 (0.46-1.24, .266)

and septic shock [18]. However, although delayed time to treatment is an important factor, it does not account for all of the mechanisms of ED crowding. In the present study, we investigated the association between ED crowding and Time to interventions such as intubation, vasopressor or inotropic use, central line catheterization, and blood transfusion. There was no significant time delay, except for the time to central line catheterization for the fourth quartile EDOR in the trauma subset and the time to transfusion for fourth quartile EDOR in the gastrointestinal bleeding subset. These results were similar to those obtained by other studies showing that ED crowding was not associated with time to treat- ment. Time to thrombolysis was not delayed in patients with acute stroke

[19] or in patients with acute myocardial infarction [16].

in-hospital mortality. In that study, patients who received delayed re- suscitation were initially staying in other places of the ED and not in the resuscitation room. In addition, the study reported that immediate and comprehensive resuscitation effort was delivered after a patient had entered the resuscitation room. Thus, delayed resuscitation effort in this study indicates that recognition of critically ill patients in the ED was delayed under ED crowding. These findings suggest that ED crowding has an adverse effect on the ability to monitor critically ill pa- tients during their ED stay.

In addition to the issue of adequate monitoring, there are other po- tential explanations, such as time to diagnosis, time to patient being ex- amined by the physician, and adherence to the recommended guidelines for a given intervention. Finally, there are explanations that can be difficult to quantify but should also be considered, such as quality of treatment. However, all of the reasons have not yet been uncovered, and thus, further investigations are needed.

Thus, we performed the present study to determine the harmful ef- fect of ED crowding on outcome among critically ill ED patients. We found that critically ill patients are remarkably susceptible to ED crowding, and thus, hospital policies should include guidelines for the management of critically ill patient under ED crowding. With regard to the hospital policy, implementation of a time rule to admission for critically ill patients and practical criteria for escalation policy are con- siderable. Triage using a specific tool based on physiological and labora- tory findings may be helpful because Nurse assessment alone can be inaccurate for High-acuity patients [21]. In terms of national policy, the government should make an effort to provide facilities that meet the de- mands of critically ill patients. These demands have been increasing [22,23], but there is still an insufficient inpatient capacity in the ED [24,25] and ICU [26]. This issue cannot be improved by individual hospi- tals and must be addressed by governmental policy.

Limitations

This study has several limitations. It is a secondary analysis of previ- ously collected data, and we attempted to minimize bias by adjusting for important confounding factors which were not included in the previ- ous analysis, such as concurrent disease status and admission diagnosis [10]. Despite our efforts, confounding variables may not be revealed or may be excluded, which affects the outcome. For instance, we did not ob- tain information about whether a patient was bedridden, which could be a potential confounder. However, we attempted to minimize other con- founders by adjusting for demographics, concurrent disease status, phys- iological variables, admission diagnosis, and patient destination.

Second, for the subset analysis, there were only a small number of pa- tients who were included in each quartile according to admission cate- gory. This was particularly true for the gastrointestinal bleeding subset, which was too small population to perform a logistic regression analysis. When examining the relationship between the EDOR and procedural time, the sample size was less than 30 patients for most of the quartiles of each disease subset. This finding may result in lack of power to detect statistical significance between quartiles.

Table 3

Logistic regression analysis of subgroup divided according to admission diagnosis

Infection (n = 527)

Cardiac disease (n = 214)

Trauma (n = 306)

Others (n = 494)

Mortality (%)

91 (16.5)

26 (12.15)

41 (13.4)

88 (17.81)

1st quartile

Reference

Reference

Reference

Reference

2nd quartile

3.38 (1.33-8.56, .010)

1.56 (0.29-8.58, .606)

4.72 (1.17-19.07, .029)

1.64 (0.71-3.81, .249)

3rd quartile

2.78 (1.10-7.07, .031)

3.69 (0.64-21.35, .144)

5.64 (1.46-21.7, .012)

2.47 (1.07-5.74, .035)

4th quartile

4.37 (1.73-11.05, .002)

2.85 (0.47-17.27, .255)

3.78 (0.85-16.80, .081)

0.89 (0.38-2.10, .791)

Data were presented as AOR (95% CI, P value).

Fig. 2. Procedural time for the total patient cohort and for each subset according to admission diagnosis. The numbers in the bars indicate the number of patients per relevant quartile. The Kruskal-Wallis test and 1-way ANOVA were used when appropriate. *P b .05 when compared with the first quartile.

Third, this study included admitted ED patients who presented to the ED with an SBP at or less than 90 mm Hg. However, such inclusion criteria may not be representative of critically ill patients. Patients with Postural hypotension [27], postprandial hypotension [28], and chronic hypotension [29] can all be included in the present study as crit- ically ill patients. Thus, it may be necessary to obtain a consensus of practical definition about critically ill ED patients rather than using blood pressure alone.

Fourth, this study collected data on the procedure time for intubations, vasopressor-inotropic use, central line catheterization, and transfusions. We did not collect data about the administration of antibiotics because it is specific to infections and not applicable to all admission diagnoses. There were many other procedures, such as the insertion of a temporary pace maker, correction of acid imbalance, and correction of stable ventric- ular tachycardia, which were not included for the same reason. In addi- tion, the procedure time did not reflect ED crowding alone but also reflected factors, such as patient irritability, time needed to explain the procedure, hesitation of the patient or guardians, unexpected clinical de- terioration, atypical presentation, interphysician dissension, misjudgment of the physician, and failure of the previous procedures.

Fifth, we did not obtain data about hospital occupancy or complications during the hospital stay, both of which could affect inpatient mortality. A statistical model, including these parameters and other hospital ward fac- tors, could provide more insight but would be extremely difficult due to the complexity of acquiring and processing such a vast array of data.

Sixth, the data used in this study were collected from a single hospital.

Thus, further research based on multicenter data is warranted.

Conclusions

In conclusion, we found that ED crowding is associated with in- creased inpatient mortality among critically ill ED patients admitted via the ED. Harmful effects of ED crowding were most prominent in the trauma subset, followed by the infection subset, and were not ob- served in the cardiac and vascular disease subset. Emergency depart- ment crowding generally did not have a major effect on procedural time including intubation, vasopressor or inotropic use, central line catheterization, and blood transfusions.

Article summary

Why is this topic important?

Emergency department crowding is associated with patient out- come in general ED patients. However, until recently, adverse effects of ED crowding among critically ill patients have not been well known.

What does this study attempt to show?

This study investigated the harmful effect of ED crowding on inpa- tient mortality among critically ill patients admitted via the ED and also evaluated data by subset according to admission diagnosis. The as- sociation between ED crowding and procedural time to intubation, vasopressor-inotropic use, central line catheterization, and blood trans- fusion was also investigated.

What are the key findings?

This study demonstrated that ED crowding is associated with increased inpatient mortality and that ED crowding has the greatest effect on the trauma subset and infection subset. However, ED crowding did not appear to have any effect on the cardiac and vascular disease subset. Furthermore, ED crowding generally did not have a major effect on procedural time.

How is patient care impacted?

This study highlights the adverse effects of ED crowding on critically ill patients and will hopefully encourage physicians and policy makers to address these issues.

Acknowledgments

Conflict of interests: None Funding and support: None. Author contributions

S.J. and T.J. designed this study. S.J. supervised the overall data collec- tion process, had full access to all of the data in the study, and takes re- sponsibility for the integrity of the data. B.P. performed the data analysis. S.J. wrote the initial draft of the manuscript. All authors pro- vided a substantial review and feedback on the final version of the arti- cle. T.J. takes responsibility for the manuscript as a whole.

All authors have read and approved the submitted manuscript. This manuscript has not been submitted or published elsewhere in whole or in part, except as an abstract (if relevant).

References

  1. Stead LG, Jain A, Decker WW. Emergency department over-crowding: a global per- spective. Int J Emerg Med 2009;2:133-4.
  2. Pines JM, Hilton JA, Weber EJ, Alkemade AJ, Al Shabanah H, Anderson PD, et al. Inter- national perspectives on emergency department crowding. Acad Emerg Med 2011; 18:1358-70.
  3. Richardson DB. Increase in patient mortality at 10 days associated with emergency department overcrowding. Med J Aust 2006;184:213-6.
  4. Guttmann A, Schull MJ, Vermeulen MJ, et al. Association between Waiting times and short term mortality and hospital admission after departure from emergency depart- ment: population based cohort study from Ontario, Canada. BMJ 2011;342:d2983. http://dx.doi.org/10.1136/bmj.d2983.
  5. Sprivulis PC, Da Silva JA, Jacobs IG, et al. The association between hospital over- crowding and mortality among patients admitted via Western Australian emer- gency departments. Med J Aust 2006;184:208-12.
  6. Sun BC, Hsia RY, Weiss RE, et al. Effect of emergency department crowding on out- comes of admitted patients. Ann Emerg Med 2013;61:605-611.e6.
  7. Cardoso LT, Grion CM, Matsuo T, et al. Impact of delayed admission to intensive care units on mortality of critically ill patients: a cohort study. Crit Care 2011;15:R28. http://dx.doi.org/10.1186/cc9975.
  8. Chalfin DB, Trzeciak S, Likourezos A, et al. Impact of Delayed transfer of critically ill patients from the emergency department to the intensive care unit. Crit Care Med 2007;35:1477-83.
  9. Rincon F, Mayer SA, Rivolta J, et al. Impact of delayed transfer of critically ill stroke pa- tients from the emergency department to the neuro-ICU. Neurocrit Care 2010;13:75-81.
  10. Jo S, Jin YH, Lee JB, et al. Emergency department occupancy ratio is associated with increased early mortality. J Emerg Med 2014;46:241-9.
  11. Gilbert EH, Lowenstein SR, Koziol-McLain J, et al. Chart reviews in emergency med- icine research: where are the methods? Ann Emerg Med 1996;27:305-8.
  12. Begley CE, Chang Y, Wood RC, et al. Emergency department diversion and trauma mortality: evidence from Houston, Texas. J Trauma 2004;57:1260-5.
  13. Jo S, Kim K, Lee JH, et al. Emergency department crowding is associated with 28-day mortality in community-acquired pneumonia patients. J Infect 2012;64:268-75.
  14. Schull MJ, Vermeulen M, Slaughter G, et al. Emergency department crowding and thrombolysis delays in acute myocardial infarction. Ann Emerg Med 2004;44:577-85.
  15. Pines JM, Hollander JE, Localio AR, et al. The association between emergency depart- ment crowding and hospital performance on antibiotic timing for pneumonia and percutaneous intervention for myocardial infarction. Acad Emerg Med 2006;13: 873-8.
  16. Harris B, Bai JC, Kulstad EB. Crowding does not adversely affect time to percutaneous coronary intervention for acute myocardial infarction in a community emergency department. Ann Emerg Med 2012;59:13-7.
  17. Schull MJ, Morrison LJ, Vermeulen M, et al. Emergency department overcrowding and Ambulance transport delays for patients with chest pain. CMAJ 2003;168: 277-83.
  18. Shin TG, Jo IJ, Choi DJ, et al. The adverse effect of emergency department crowding on compliance with the resuscitation bundle in the management of severe sepsis and septic shock. Crit Care 2013;17:R224. http://dx.doi.org/10.1186/cc13047.
  19. Chatterjee P, Cucchiara BL, Lazarciuc N, et al. Emergency department crowding and time to care in patients with acute stroke. Stroke 2011;42:1074-80.
  20. Hong KJ, Shin SD, Song KJ, et al. Association between ED crowding and delay in re- suscitation effort. Am J Emerg Med 2013;31:509-15.
  21. O’Connor E, Gatien M, Weir C, et al. Evaluating the effect of emergency department crowding on triage destination. Int J Emerg Med 2014;7:16. http://dx.doi.org/10. 1186/1865-1380-7-16.
  22. Lambe S, Washington DL, Fink A, et al. Trends in the use and capacity of California’s emergency departments, 1990-1999. Ann Emerg Med 2002;39:389-96.
  23. Angus DC, Kelley MA, Schmitz RJ, et al. Committee on Manpower for Pulmonary and Critical Care Societies (COMPACCS). Caring for the critically ill patient. Current and projected workforce requirements for care of the critically ill and patients with pul- monary disease: can we meet the requirements of an aging population? JAMA 2000;284:2762-70.
  24. Gordon JA, Billings J, Asplin BR, et al. Safety net research in emergency medicine: proceedings of the Academic Emergency Medicine Consensus Conference on “The Unraveling Safety Net”. Acad Emerg Med 2001;8:1024-9.
  25. Richardson LD, Asplin BR, Lowe RA. Emergency department crowding as a health policy issue: past development, future directions. Ann Emerg Med 2002;40:388-93.
  26. Varon J, Fromm Jr RE, Levine RL. Emergency department procedures and length of stay for critically ill Medical patients. Ann Emerg Med 1994;23:546-9.
  27. Sathyapalan T, Aye MM, Atkin SL. Postural hypotension. BMJ 2011;342:d3128. http://dx.doi.org/10.1136/bmj.d3128.
  28. Rayner CK, Horowitz M. Physiology of the ageing gut. Curr Opin Clin Nutr Metab Care 2013;16:33-8.
  29. Akahoshi M, Hida A, Imaizumi M, Soda M, Maeda R, Ichimaru S, et al. Basic charac- teristics of chronic hypotension cases: a longitudinal follow-up study from 1958 through 1999. Hypertens Res 2006;29:1-7.

Leave a Reply

Your email address will not be published. Required fields are marked *