The Association Between Emergency Department Crowding and Adverse Cardiovascular Outcomes in Patients with Chest Pain

Jesse M. Pines, MD, MBA, MSCE, Charles V.Pollack Jr, MD, MA, Deborah B. Diercks, MD, Anna Marie Chang, MD, Frances S. Shofer, PhD, and Judd E. Hollander, MD

Abstract
Objectives: While emergency department (ED) crowding is a worldwide problem, few studies have demonstrated associations between crowding and outcomes. The authors examined whether ED crowding was associated with adverse cardiovascular outcomes in patients with chest pain syndromes (chest pain or related complaints of possible cardiac origin). Methods: A retrospective analysis was performed for patients 30 years of age with chest pain syndrome admitted to a tertiary care academic hospital from 1999 through 2006. The authors compared rates of inpatient adverse outcomes from ED triage to hospital discharge, dened as delayed acute myocardial infarction (AMI), heart failure, hypotension, dysrhythmias, and cardiac arrest, which occurred after ED arrival using ve separate crowding measures. Results: Among 4,574 patients, 251 (4%) patients developed adverse outcomes after ED arrival; 803 (18%) had documented acute coronary syndrome (ACS), and of those, 273 (34%) had AMI. Compared to less crowded times, ACS patients experienced more adverse outcomes at the highest waiting room census (odds ratio [OR] = 3.7, 95% condence interval [CI] = 1.3 to 11.0) and patient-hours (OR = 5.2, 95% CI = 2.0 to 13.6) and trended toward more adverse outcomes during time of high ED occupancy (OR = 3.1, 95% CI = 1.0 to 9.3). Adverse outcomes were not signicantly more frequent during times with the highest number of admitted patients (OR = 1.6, 95% CI = 0.6 to 4.1) or the highest trailing mean length of stay (LOS) for admitted patients transferred to inpatient beds within 6 hours (OR = 1.5, 95% CI = 0.5 to 4.0). Patients with non-ACS chest pain experienced more adverse outcomes during the highest waiting room census (OR = 3.5, 95% CI = 1.4 to 8.4) and patient-hours (OR = 4.3, 95% CI = 2.6 to 7.3), but not occupancy (OR = 1.8, 95% CI = 0.9 to 3.3), number of admitted patients (OR = 0.6, 95% CI 0.4 to 1.1), or trailing LOS for admitted patients (OR = 1.2, 95% CI = 0.6 to 2.0). Conclusions: There was an association between some measures of ED crowding and a higher risk of adverse cardiovascular outcomes in patients with both ACS-related and nonACS-related chest pain syndrome. ACADEMIC EMERGENCY MEDICINE 2009; 16:617625 2009 by the Society for Academic Emergency Medicine Keywords: crowding, overcrowding, adverse outcomes, cardiovascular, emergency department, health care

From the Department of Emergency Medicine, University of Pennsylvania School of Medicine (JMP, CVP, AMC, FSS, JEH), Philadelphia, PA; the Center for Clinical Epidemiology and Biostatistics, University of Pennsylvania School of Medicine (JMP), Philadelphia, PA; The Leonard Davis Institute of Health Economics, University of Pennsylvania (JMP), Philadelphia, PA; and the Department of Emergency Medicine, University of California at Davis (DBD), Sacramento, CA. Received December 18, 2008; revision received March 9, 2009; accepted March 28, 2009. This work was funded by the Thomas McCabe Fund at the University of Pennsylvania. Address for correspondence and reprints: Jesse M. Pines, MD, MBA, MSCE; e-mail: pinesjes@uphs.upenn.edu.

here were 1.6 million patients hospitalized in 2005 in the United States with a discharge diagnosis of acute coronary syndrome (ACS).1,2 Of those, 64% were admitted to the hospital through the emergency department (ED).1 Individual therapies for ACS recommended by American College of Cardiology American Heart Association (ACC AHA) guidelines reduce the relative likelihood of death or reinfarction by up to 50%, and the absolute reduction in adverse outcomes ranges from 6.1% to 1.2%.35 In addition, adherence to guideline-concordant care has been associated with improved outcomes in patients with ACS.68 Operational strategies, such as one-call activation of the cardiac

2009 by the Society for Academic Emergency Medicine doi: 10.1111/j.1553-2712.2009.00456.x

ISSN 1069-6563 PII ISSN 1069-6563583

617

618

Pines et al.

ED CROWDING AND ADVERSE OUTCOMES IN CHEST PAIN

catheterization laboratory, are associated with improved door-to-needle times for patients with ST-segment elevation myocardial infarction (STEMI).9,10 However, current unstable angina (UA) and non-STEMI guidelines do not address the operational issues that may affect the likelihood of timely application of evidence-based recommendations. There were 6 million patients who visited the ED in the United States in 2004 with a chief complaint of chest pain.11,12 A high proportion of these patients are exposed to crowded conditions in the ED, which is considered a critical public health problem by the Institute of Medicine.13,14 ED crowding, often a result of poor inpatient bed availability and the use of ED treatment areas for admitted patients,15 is associated with longer waits to be seen, delays in testing and critical treatments and lower patient satisfaction.1623 Crowding has been shown to have an adverse impact on in-hospital survival, particularly for patients awaiting transfer to intensive care unit beds.2427 Specically related to cardiovascular disease, patients with chest pain who present in temporal proximity to a severely injured patient that results in activation of the trauma team are more likely to experience in-hospital adverse events.28 In addition, longer ED lengths of stay (LOS) in patients with non-STEMI are associated with a lower likelihood of receiving guideline-concordant care.29 Neither of these latter two studies directly assessed a patient-level exposure to ED crowding (i.e., the actual level of crowding when they arrived), but the competition for ED and hospital resources has been proposed as an explanation for the results of both.30 We estimated the effect of patient-level exposures to ED crowding on the incidence of subsequent adverse cardiovascular outcomes in ED patients with chest pain syndrome in patients with ACS-related and non ACS-related chest pain syndrome. Secondarily, we determined if there was a reduced likelihood of aspirin and beta-blocker use in the ED or delay in the timing of electrocardiograms (ECG) and whether, if present, these may account for differences in adverse outcome rates in patients with ACS. METHODS Study Design We performed a retrospective analysis of data from an ongoing prospective cohort study of ED patients 30 years of age and older with chest pain syndrome. This study was determined to be exempt from informed consent requirements by the institutional review board. Study Setting and Population The study was performed at the Hospital of the University of Pennsylvania, a large, urban tertiary care teaching hospital with an ED census of 50,000 to 55,000 adult patients per year during the study period. The main ED has 25 individual patient rooms and 15 hallway treatment spaces. There is a separate eight-bed minor-care area. Hallway beds are utilized to evaluate patients when the standard rooms are lled to capacity. A clinical decision unit, which involves a protocol for the

low-risk patients with chest pain syndrome, opened in November 2005. Results of the prospective cohort study have been previously reported.3133 For study purposes, chest pain syndrome was dened as chest pain or related complaints of possible cardiac origin that prompted either the treating emergency physician (EP) to order an ECG or a nurse to perform an ECG by standing order. Related complaints may have included symptoms such as dyspnea in a diabetic patient, but did not include more general symptoms such as syncope or weakness. This group was enrolled as a convenience sample from September 1999 through March 2006. We performed queries in our ED information system and calculated the actual state of the ED when they arrived at triage (i.e., level of ED crowding). In our ED, when a patient with chest pain of possible cardiac origin presents in the triage area, an ECG is typically performed by the triage nurse or a paramedic. Patients are usually seen initially by resident physicians who present the case to attending physicians. All patients are seen and evaluated directly by an attending EP. If the decision is made to admit, a bed request is placed. For medicine or cardiology admissions, the admitting team then manages the patient while they are in the ED. Both ED patients and admitted patients are cared for by ED nursing staff. Study Protocol Data Collection and Processing. Patients age 30 years with chest pain syndromes who prompted ordering of an ECG by the treating EP or registered nurse and were ultimately admitted to the hospital were included in this analysis. STEMI patients were excluded from enrollment. Where there were repeated visits by the same patient, only the rst visit was used in the analysis. Trained research assistants were present in the ED for 16 hours per day, 7 days per week to enroll patients.34 Coverage during these times captures 85%95% of eligible patients. Clinical information was obtained directly from the treating physician while the patient was still in the ED. The treating physician recorded the patients demographic information, chest pain characteristics, and medical history at the time of the initial evaluation. Laboratory results, cardiac markers, electrocardiographic interpretation, Thrombolysis In Myocardial Infarction (TIMI) risk score, medications administered (aspirin and beta-blockers), timing of ECG, and ED disposition were collected in real time.35 The time of day of ED triage, day of week, and year of service were also recorded. Data were collected in accordance with standardized reporting guidelines.36,37 Admitted patients were followed daily during hospitalization for death or adverse cardiac outcomes (death, signicant dysrhythmias, heart failure, hypotension, recurrent myocardial infarction, or revascularization). Primary Outcome. The primary outcome was the development of an adverse cardiovascular outcome that was not present on ED arrival, but that occurred during their hospitalization. Timing of the event was recorded and categorized into the following subgroups: 24 hours and >24 hours to demonstrate how close

ACAD EMERG MED July 2009, Vol. 16, No. 7

www.aemj.org

619

each individual event occurred related to the ED visit. Each individual patient was classied as either having an adverse cardiovascular outcome or not. An adverse cardiovascular outcome was dened as any of the following: death, cardiac arrest, delayed myocardial infarction, development of congestive heart failure, ventricular tachycardia or ventricular brillation, supraventricular dysrhythmias, symptomatic bradycardia, or hypotension. Acute myocardial infarction (AMI) as a cause of the presenting symptoms was not considered an adverse outcome. Revascularization was not considered an adverse outcome because the need for revascularization is dened individually by a patients anatomy and is likely not related to the study question of ED crowding. The development of new heart failure was dened by clinician initiation of treatment. Specic dysrhythmias required documentation in the medical record separate from identication on rhythm strips without intervention or notation. Hypotension was dened as an episode of systolic blood pressure measured at <90 mm Hg. We presumed that cardiovascular outcomes were important in patients with ACS-related chest pain as they may be directly related to their primary reason for admission. In addition, for patients without ACS-related chest pain (which includes other diseases including more benign syndromes such as costochondritis and gastroesophageal reex disease and also more serious diseases such as pneumonia, aortic disease, pulmonary embolism, and other potentially life-threatening conditions), we presumed that the presence of cardiovascular outcomes was similarly an important outcome that was related to our research question. Although this group may not have ACS at arrival, it is still possible that they may develop adverse cardiovascular outcomes after ED arrival. We also evaluated the potential causes for these associations with three guideline recommendations: 1) treatment with aspirin in the ED, 2) treatment with beta-blockers in the ED for those with ACS, and 3) performance of an ECG within 10 minutes of ED arrival.3 If patients received aspirin in the out-of-hospital setting, it was marked as treated in the ED. Out-of-hospital ECGs were not performed during the study period. Acute myocardial infarction was considered to occur if the patient had prolonged chest pain and either elevation of the CK-MB enzyme two times the upper limit of normal and or cardiac troponin I greater than the upper limit of normal or standard ECG evolution of AMI. The diagnosis of UA was made if there was an attending physician diagnosis of UA, documented reversible ischemia on stress test, coronary artery occlusion of >70% in at least one vessel, or elevations in cardiac markers above laboratory normal but below threshold for diagnosis of AMI (CK-MB > 5 ng mL, < 10 ng mL; troponin I > 0.3 ng mL, < 2 ng mL using rst-generation assays). ACS was dened as a diagnosis of AMI or UA. Objective cardiovascular testing (cardiac catheterization, stress testing, or computed tomography coronary angiogram testing) was performed on 37% of admitted patients. ED Crowding Measures. Using our electronic medical record, a patient-level exposure to ED patient crowd-

ing was assigned to each patient at the time of triage. That is, the exposure was assigned to represent the precise operational state the ED was in at the time each patient arrived at the hospital. We chose ve simple measures of ED crowding: waiting room number, total patient care hours (dened as the sum of the hours for all patients presently in the main ED), ED occupancy in the main ED as a proportion of 40 beds (both patient rooms and hallway beds) as the denominator (ED occupancy excludes fast-track and trauma-bay patients), the average length of stay for patients who were moved to inpatient beds in the 6 hours prior to ED triage of the patient with chest pain (represents a trailing average for ED LOS for admitted patients), and number of admitted patients in the ED at that time. These measures were chosen because they have been shown to be associated with outcomes in previous studies of other patient populations.19,21 In addition, recent work has shown that simpler measures such as occupancy are equally predictive as more complex measures such as EDWIN to identify episodes of diversion and higher left without being seen rates.38 Elements of individual patient care timing, including patient triage time, room arrival time, and room departure time are automatically time-stamped at the time of care in the patient tracking system. This permits the calculation of crowding variables using Microsoft Access (Redmond, WA, 2003) queries and allows us to reconstruct the administrative state of the ED at any moment with regard to crowding variables. Information for each subject did not directly contribute to their crowding exposure. That is, study subjects were not included in any calculation of patient crowding variables. Data Analysis Data are reported with 95% condence intervals (CI), standard deviations (SD), and interquartile ranges (IQR). We compared subjects with missing data on primary and secondary outcomes to those without complete follow-up data. Logistic regression modeling was used to determine the effect of ED crowding variables on the outcomes, adjusting for patient-level confounders known to be associated with cardiovascular morbidity and mortality, including age, sex, race, time of day (as three time intervals to control for diurnal variation in cardiovascular outcomes: 00:0108:00, 08:0115:00, 15:0100:00), weekend weekday, year of service as a dummy variable (to control for secular trends in quality of care), and severity of illness (TIMI risk score 07).39 We also included a prior history of heart failure in the model because of the large numbers of patients who had heart failure as an adverse outcome. All variables were included in the adjusted model without interaction terms. Non-linearity (i.e., threshold effects) in the adjusted effect of crowding variables was tested by dividing crowding measures into quartiles as dummy variables. Adjusted odds ratios (ORs) and CIs represent bootstrapped estimates with replacement to validate the model. To help validate the regression models, global goodness of t was assessed using the Hosmer-Lemeshow test and area under the receiver operator curve statistics were calculated and compared to bootstrapped estimates. Because there was an ED

620

Pines et al.

ED CROWDING AND ADVERSE OUTCOMES IN CHEST PAIN

chest pain unit in the nal ve months of the study, we performed a sensitivity analysis excluding these data and assessed the changes in the coefcients for the primary independent variables (crowding measures). The analysis was conducted using Stata statistical software (Release 10.0, 2007, StataCorp, College Station, TX). RESULTS Baseline Characteristics, Follow-up Information, and Primary Outcomes Over the study period, 4,574 patients presented with a chest pain syndrome and were admitted to the hospital. The median length of hospital stay was 2 days (IQR = 13 days), the mean (SD) age was 56 (15) years, and 55% were female. Patient demographics and cardiac risk factors are listed in Table 1, categorized by those with documented ACS and those without ACS. Outcomes were available for 4,424 patients (97%). There were no differences in age, sex, or race for those with and without complete follow-up data. There were 803 (18%) patients with a nal diagnosis of ACS: 273 (6%) with AMI and 530 (12%) with UA. There were 251 patients (6%) who developed one or more adverse outcomes after ED arrival. A total of 96 patients (12%) with ACS and 155 (4%) with non-ACS chest pain experienced adverse outcomes. Table 2 shows the specic outcomes and their timing (24 hours vs. >24 hours) and ACS versus non-ACS.

Table 2 Adverse Cardiovascular Outcomes That Occurred During Hospitalization and Complication Timing (n = 251) Complication 24 hours >24 hours 20 59 9 20 18 21 13 2 13 2 8 4 1 7 Total 25 67 11 33 21 32 14 5 15 4 9 6 1 8

Heart failure ACS 5 Non-ACS 8 Hypotension ACS 2 Non-ACS 13 Supraventricular dysrhythmias ACS 3 Non-ACS 11 Delayed myocardial infarction ACS 1 Non-ACS 3 Cardiac arrest with CPR ACS 2 Non-ACS 2 Ventricular dysrhythmias ACS 1 Non-ACS 2 Bradydysrhythmias ACS 0 Non-ACS 1

ACS = acute coronary syndrome; CPR = cardiopulmonary resuscitation.

Table 1 Characteristics for 4,574 ED Patients Admitted to the Hospital with Documented ACS-related and NonACS-related Chest Pain Documented ACS (n = 803) Mean age (SD) Female sex Race Black or African American White Cardiac history Angina Coronary artery disease Heart failure Myocardial infarction Cardiac risk factors Hypertension Tobacco Hypercholesterolemia Diabetes mellitus Family history TIMI risk score 01 24 57 Percent receiving aspirin in the ED Percent receiving beta-blockers in the ED Percent receiving an ECG within 10 minutes of arrival 60 (13) 42 59 38 32 49 21 32 71 48 41 30 21 25 63 12 81 21 65 Non-ACS (n = 3,771) 56 (15) 57 70 26 15 23 15 15 60 41 24 23 20 55 42 3 71 8 65

ED Crowding Measures During the study period, median waiting room census was eight patients (IQR = 412). The median patient care hours were 98 (IQR = 63142), and the median occupancy rate was 60% (IQR 45%72.5%). The median number of admitted patients waiting for an inpatient bed was 9 (IQR = 512), and the trailing average LOS for admitted patients who were transferred to inpatient beds within 6 hours of triage time was 8 hours (IQR = 59 hours). Main Results Table 3 shows adjusted ORs for adverse cardiovascular outcomes by quartiles of waiting room census, total patient care hours, percent occupancy, number of admitted patients, and trailing LOS for admitted patients who were transferred to inpatient beds within the previous 6 hours. In an adjusted analysis of patients with documented ACS, a waiting room census of 12 was associated with a higher risk of cardiovascular outcomes (adjusted OR = 3.7, 95% CI = 1.3 to 11.0) compared to the lowest quartile. Patient care hours 142 (OR = 5.2, 95% CI = 2.0 to 13.6) and ED occupancy 72.5% (OR = 3.1, 95% CI = 1.0 to 9.3) were also associated with adverse cardiovascular outcomes compared to the lowest quartile. The number of admitted patients 12 and recent LOS for admitted patients 9 hours were not associated with a higher risk of adverse outcomes compared to the lowest quartile (OR = 1.6, 95% CI = 0.6 to 4.1; and OR = 1.5, 95% CI = 0.5 to 4.0, respectively). In the non-ACS cohort, patients were more likely to experience adverse outcomes when the waiting room had 12 patients compared to the lowest quartile (OR = 3.5, 95% CI = 1.48.4). Similarly, when the number of patient-hours was 142, patients

Values are reported as percent except where otherwise noted. ACS = acute coronary syndrome; ECG = electrocardiogram; TIMI = Thrombolysis in Myocardial Infarction.

ACAD EMERG MED July 2009, Vol. 16, No. 7

www.aemj.org

621

Table 3 Adjusted ORs for Any Adverse Cardiovascular Outcome for Patients With Chest Pain With Documented ACS and Non-ACS ED Crowding Measures ACS Cohort (n = 803) Waiting room number Patient care hours Number of admitted patients Occupancy Recent ED LOS for admitted patients* Non-ACS cohort (n = 3,371) Waiting room number Patient care hours Number of admitted patients Occupancy Recent ED LOS for admitted patients* 25th49th 1.1 1.3 1.3 0.9 1.0 1.6 2.7 0.9 0.9 0.9 (0.43.2) (0.53.5) (0.62.8) (0.32.8) (0.52.0) (0.83.4) (1.55.0) (0.51.5) (0.51.6) (0.51.5) 50th74th 0.9 1.4 1.3 1.3 0.6 2.3 2.2 0.8 1.2 0.8 (0.32.9) (0.54.3) (0.53.3) (0.44.4) (0.21.4) (1.15.0) (1.23.9) (0.41.3) (0.62.2) (0.51.3) 3.7 5.2 1.6 3.1 1.5 3.5 4.3 0.6 1.8 1.2 75th (1.311.0) (2.013.6) (0.64.1) (1.09.3) (0.54.0) (1.48.4) (2.67.3) (0.41.1) (0.93.3) (0.62.0)

Adjusted model included TIMI risk score, a history of heart failure, time of day in three categories, year of service, weekend versus weekday, age, race, and sex and all reported ORs are compared to the lowest quartile, which corresponds to the lowest level of ED crowding. Reported ORs represent a single crowding measure placed in each model. 95% CIs are represented in parentheses. ACS = acute coronary syndrome; LOS = length of stay; TIMI = Thrombolysis In Myocardial Infarction. *Refers to other patients in the ED (admitted patients who were transferred to inpatient beds within 6 hours of a patients ED triage time).

experienced higher rates of adverse outcomes compared to the lowest quartile (OR = 4.3, 95% CI = 2.6 to 7.3). Occupancy of 72.5%, number of admitted patients 12, and trailing ED LOS for admitted patients of 9 hours were not signicantly associated with a higher rate of adverse outcomes compared to the lowest quartiles (OR = 1.8, 95% CI = 0.9 to 3.3; OR = 0.6, 95% CI = 0.4 to 1.1; and OR = 1.2, 95% CI = 0.6 to 2.0, respectively). The majority of patients (74%) in the non-ACS cohort had a hospital discharge diagnosis of chest pain not otherwise specied; other diagnoses were less frequent (Table 4). These ndings cannot be explained by initial
Table 4 Final Hospital Discharge Diagnosis for 3,767 Patients Admitted to the Hospital with Chest Pain and Not Found to Have ACS Final hospital discharge diagnosis Chest pain not otherwise specied Other Congestive heart failure or cardiomyopathy Pneumonia Gastroesophageal reex disease or gastritis Pulmonary embolism or deep vein thrombosis Asthma chronic obstructive pulmonary disease bronchitis Atrial brillation Cancer (many types) Musculoskeletal chest pain or costochondritis Stable angina Other gastrointestinal illness Pericarditis, endocarditis, or myocarditis Aortic dissection Other infectious etiology Other arrhythmia Anxiety or palpitations Other lung disease Hypertension Gallbladder disease Pneumothorax No. 2,803 245 169 68 62 43 42 40 37 35 30 29 28 26 21 21 18 17 13 10 5 % 74.4 6.5 4.5 1.8 1.6 1.1 1.1 1.1 1.0 0.9 0.8 0.8 0.7 0.7 0.6 0.6 0.5 0.5 0.3 0.3 0.1

treatment differences. ED crowding was not associated with a reduced likelihood of receiving treatment with aspirin or beta-blockers in the ED in patients with ACS (Table 5). Likewise, there was no association between ECG performed within 10 minutes and ED crowding levels (Table 5). In the sensitivity analysis, removing the last 5 months of data when the observation unit was available from the cohort did not appreciably change the ORs for the ED crowding variables. DISCUSSION We found that a patient-level exposure to a crowded ED as measured by three of the ve crowding measures we tested was associated with higher rates of adverse inpatient cardiovascular outcomes in a large sample of ED patients with ACS-related chest pain syndrome and two out of ve measures in nonACS-related chest pain. Among patients with ACS, several crowding measures showed three to ve times higher rates of adverse outcomes during the highest levels of crowding. We thought that ED-based process measures of performance in patients with ACS may account for some of this effect because they have had a weak but statistically signicant effect on patient outcomes in previous studies.40,41 However, when we tested whether the study outcome was associated with the use of aspirin and beta-blockers, or differences in the timing of the ECG, we found that they did not modify our results, indicating that higher adverse outcome rates observed during crowded times are not explained by the speed or quality of these measurable ED-based interventions. This lack of effect may be explained by the fact that in our ED, ECGs are performed at triage. This is performed to ensure that they happen rapidly regardless of the crowding level. While the latest cardiology guidelines have downplayed the importance of beta-blockers, at the time of the study, the use of aspirin and beta-blockers was relatively standardized in chest pain patients.42

622

Pines et al.

ED CROWDING AND ADVERSE OUTCOMES IN CHEST PAIN

Table 5 Adjusted ORs for ED Treatments With Aspirin, Beta-blockers, and ECG Performed Within 10 Minutes for 803 Patients with ACS by Quartiles of ED Crowding ED Crowding Measures Treatment with aspirin in the ED Waiting room number Patient care hours Number of admitted patients Occupancy Recent ED LOS for admitted patients* Treatment with beta-blockers in the ED Waiting room number Patient care hours Number of admitted patients Occupancy Recent ED LOS for admitted patients* ECG performed within 10 minutes of ED arrival Waiting room number Patient care hours Number of admitted patients Occupancy Recent ED LOS for admitted patients* 25th49th 0.8 1.3 0.7 0.9 1.4 0.6 0.8 0.8 1.5 1.0 1.6 1.1 1.7 1.3 1.7 (0.41.7) (0.82.2) (0.41.3) (0.42.3) (0.72.7) (0.31.3) (0.41.5) (0.51.3) (0.73.1) (0.52.1) (0.92.8) (0.61.8) (1.02.8) (0.82.3) (0.52.1) 50th74th 0.8 1.1 1.0 1.3 1.3 0.9 0.9 1.1 1.5 1.0 1.5 1.0 1.8 1.5 1.0 (0.32.0) (0.52.4) (0.42.7) (0.63.0) (0.62.9) (0.41.8) (0.41.8) (0.71.8) (0.73.3) (0.52.3) (0.92.8) (0.61.8) (1.03.1) (0.92.5) (0.52.3) 1.2 1.1 1.6 1.4 1.4 1.2 1.3 1.2 1.7 1.7 1.5 1.4 1.5 1.5 1.7 75th (0.52.8) (0.52.3) (0.54.2) (0.63.3) (0.63.3) (0.62.6) (0.72.5) (0.71.9) (0.83.8) (0.73.1) (0.82.8) (0.82.3) (0.92.5) (0.82.9) (0.83.5)

Adjusted model included TIMI risk score, a history of heart failure, time of day in three categories, year of service, weekend versus weekday, age, race, and sex, and all reported ORs are compared to the lowest quartile, which corresponds to the lowest level of ED crowding. Reported ORs represent a single crowding measure placed in each model. 95% CIs are represented in parentheses. ACS = acute coronary syndrome; ECG = electrocardiogram; TIMI = Thrombolysis In Myocardial Infarction. *Refers to other patients in the ED (admitted patients who were transferred to inpatient beds within 6 hours of a patients ED triage time).

While we were not able to directly explain the association between crowding and higher rates of adverse outcomes through what was measured in this observational study, we have several hypotheses. During crowded times, quality of nursing care may be lower (i.e., higher nurse-to-patient ratios), which may draw ED resources away from chest pain patients. In addition, there may be worse coordination of care and less time to communicate with patients. There may be a higher number of risky handoffs and less staff availability to ensure that the complex process of care is conducted in a high-quality manner. Because we observed higher rates of adverse outcomes in patients with ACSrelated and nonACS-related chest pain, this suggests that the crowded environment may indicate a higher level of dysfunction and a higher error rate for a broad cohort of patients ill enough to require hospital admission. The majority of the adverse outcomes occurred greater than 24 hours after the patients arrived at the hospital. This suggests that the systematic dysfunction affects not only temporally proximal outcomes, but also more distal ones. It is possible that an early error may manifest later during the admission if care providers have missed an important element of history, missed the diagnosis, or spent less time ensuring a seamless hospital stay. Many of the adverse outcomes were related to heart failure, which may indicate that upfront medication errors, such as a delayed evaluation by the admitting team or delays in administering medications (such as restarting a patients furosemide), may explain these outcomes. Some measures of crowding were associated with adverse outcomes, while others were not. Because each crowding measure indicates a different type of supply

demand mismatch, this may allow us to pinpoint which types of system dysfunction from ED crowding may account for observed differences in adverse outcomes. A high waiting room census may indicate either that there is a long triage queue or that that there is no space to move patients into the ED, while occupancy indicates that the main ED itself is at capacity. Waiting room census and ED occupancy are independent predictors of worse pneumonia care and pain control.19,21 Very high waiting room census numbers may indicate resource saturation as ED staff try to expedite care of existing patients to make room for waiting ones. ED occupancy was a weaker predictor, possibly because a high occupancy ED may be functioning at capacity, not necessarily over capacity. Total patient care hours is an indirect measure of total ED workload, because it lengthens as individual patients stay for longer periods. These longer periods may be from boarding, but also may be from long waiting periods for computed tomography scans, magnetic resonance images, or consultant evaluation. High patient care hours demonstrated the largest effect in both populations studied, indicating that it may be the best general measure of systematic dysfunction and patient safety in our ED. Patient care hours is predictive of longer waiting times for both antibiotics and pain medications, which also demonstrate an association with proximal ED care.19,21 Because measures of ED crowding and workload were predictive, while boarding measures such as the number of admitted patients were not, this may indicate that the higher levels of adverse outcomes may be more attributable to ED-based dysfunction (i.e., the crowded environment itself), as opposed to dysfunction that is directly linked to boarding or hospital crowding. While

ACAD EMERG MED July 2009, Vol. 16, No. 7

www.aemj.org

623

boarding for intensive care unit patients has been associated with higher death rates for individual patients who receive critical care services in the ED, it is possible that exposure to boarding may not be predictive of higher adverse event rates in broader cohorts of patients.26 We were unable to measure in this study whether these adverse outcomes were associated with higher death rates. A recent study from our group found that ED crowding measures did not predict higher death rates in a broad cohort of older adults.43 It is possible that adverse outcomes associated with systematic dysfunction do not directly translate to higher death rates because they are either less severe or can be corrected after they are recognized. However, it is difcult to draw this conclusion because the present work studied a more homogenous group of patients. Higher death rates attributable to crowding may be diluted in broader cohorts of patients. This may suggest that certain patients or admission diagnoses may be more susceptible to adverse events from system dysfunction than others. The present study is one of the rst to demonstrate the effect of ED crowding on adverse outcomes that occur after initial ED evaluation. As organizations interested in the measurement of quality move forward in the use of performance measures for EDs, simple elements like waiting room number, total patient care hours, and ED occupancy appear to be sensitive indicators of both proximal ED quality (time to antibiotics in pneumonia and treatment of pain) and also distal outcomes for patients with ACS. The benet of simple measures is that they are widely generalizable to hospitals. They are also relatively easy to measure. Most importantly, if improved, they may lead to improved delivery of emergency care in the United States. However, future studies are needed to demonstrate improved outcomes as a result of reducing ED and hospital crowding. LIMITATIONS A central limitation is that this was a single-center study. This impacts the generalizability of these ndings, specically the operational cutoffs that were predictive of worse outcomes. In addition, because the level of crowding across the entire study may be lower than is seen at other centers, it may explain some of the inconsistency in the association where some measures of crowding were predictive of adverse outcomes while others were not. Therefore, the associations reported in this study may not be observed in other EDs with different processes of care for patients with chest pain syndrome or levels of crowding. In other centers with higher levels of crowding, these associations may be more dramatic. This data set represents 8 years of data. Therefore, there may have also been a longitudinal time-related effect that we were unable to adequately control for in the adjusted models. While we did control for year of service, because the level of crowding also increased over time, there remains the possibility that this did not adequately adjust for temporal factors related to both severity of illness and

crowding. Because we did not specically measure left without being seen rates, or capture chest pain patients who left prior to evaluation, which would be expected to be higher during times of crowding, this may have biased our sample. However, we may have underestimated the true impact of crowding because patients who leave without being seen may also have experienced poor outcomes because they did not receive ED testing or treatment. We also did not take into account medication administration prior to ED arrival or in the out-of-hospital setting. Because there was a low rate of aspirin use overall, even in the ACS cohort, it is likely that many did receive aspirin before arriving at the hospital. We think that this would not bias our estimates because there is no clear reason that arrival during crowding times would be associated with lower rates of out-of-hospital medication use. In an ecologic study, we found that door-to-needle times were not affected by overall levels of ED crowding in patients with STEMI.18 This may be because typically resources are directed toward critically ill patients regardless of how crowded the ED is. Therefore, systematic exclusion of these patients may bias the study toward showing a larger effect if indeed STEMI patients are unaffected by crowding. We were not able to adjust for stafng levels, handoffs, or other detailed elements of care such as when the admitting team saw the patient, actual time spent with the patient, or medication errors that may have helped delineate the specic cause for this association. In addition, because we could not directly explain this nding with measured variables, our results should be seen as hypothesis generating and needing conrmation. Finally, because we were unable to power this study for specic cardiovascular outcomes, we used a composite outcome. As a result, adverse outcomes in this study were not equivalentthat is, symptomatic bradycardia is clearly less severe than cardiac arrest; this reduced our ability to detect clinical outcomes that may lead to higher death rates. CONCLUSIONS At our institution, there was an association between higher rates of adverse cardiovascular outcomes and some measures of higher ED crowding. We could not explain this association by process measures such as the time to electrocardiogram acquisition or treatment with aspirin and beta-blockers.
The authors thank Chris Boedec for his help with this project. Dr. Pines had full access to all of the data in the study and takes responsibility for the integrity of the data analysis.

References 1. Healthcare Cost and Utilization Project (HCUP) Nationwide Inpatient Sample using ICD-9 410 and 411. Available at: http://hcupnet.ahrq.gov/HCUPnet. jsp. Accessed Nov 12, 2008. 2. Bertoni A, Bonds D, Thom G, Chen GJ, Goff DC Jr. Acute coronary syndrome national statistics: challenges in denitions. Am Heart J. 2005; 149:105561.

624

Pines et al.

ED CROWDING AND ADVERSE OUTCOMES IN CHEST PAIN

3. Anderson JL, Adams CD, Antman EM, et al. ACC AHA 2007 guidelines for the management of patients with unstable angina non-ST-elevation myocardial infarction: a report of the American College of Cardiology American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol. 2007; 50:e1e157. 4. Alexander KP, Roe MT, Chen AY, et al. Evolution in cardiovascular care for elderly patients with nonST-segment elevation acute coronary syndromes: results from the CRUSADE National Quality Improvement Initiative. J Am Coll Cardiol. 2005; 46:147987. 5. Peterson ED, Roe MT, Mulgund J, et al. Association between hospital process performance and outcomes among patients with acute coronary syndromes. JAMA 2006; 295:191220. 6. Eagle KA, Montoye CK, Riba AL. Guidelines-based standardized care is associated with substantially lower mortality in Medicare patients with acute myocardial infarction. J Am Coll Cardiol. 2005; 46:12428. 7. Rogers WJ, Canto JG, Lambrew CT, et al. Temporal trends in the treatment of over 1.5 million patients with myocardial infarction in the US from 1990 through 1999: the National Registry of Myocardial Infarction 1, 2 and 3. J Am Coll Cardiol. 2000; 36:205663. 8. Freemantle N, Cleland J, Young P, Mason J, Harrison J. Beta blockade after myocardial infarction: systematic review and meta regression analysis. BMJ 1999; 318:17307. 9. Bradley EH, Herrin J, Wang Y, et al. Strategies for reducing the door-to-balloon time in acute myocardial infarction. N Engl J Med. 2006; 355:230820. 10. Mehta RH, Newby LK, Patel Y, et al. The impact of emergency department structure and care processes in delivering care for non-ST-segment elevation acute coronary syndromes. Am Heart J. 2006; 152:64860. 11. National Hospital Ambulatory Medical Care Survey: 2004 emergency department summary. Advance data from vital and health statistics; No. 372. Hyattsville, MD: National Center for Health Statistics, 2006. 12. Brown DW, Xie J, Mensah GA. Electrocardiographic recording and timeliness of clinician evaluation in the emergency department in patients presenting with chest pain. Am J Cardiol. 2007; 99:11158. 13. Derlet R, Richards J, Kravitz R. Frequent overcrowding in US emergency departments. Acad Emerg Med 2001; 8:1515. 14. Institute of Medicine. Hospital-Based Emergency Care: At the Breaking Point. 2006. Available at: http://www.nap.edu/catalog/11621.html#toc. Accessed Nov. 11, 2008 15. Falvo T, Grove L, Stachura R, et al. The opportunity loss of boarding admitted patients in the emergency department. Acad Emerg Med. 2007; 14:3327. 16. Derlet RW, Richards JR. Overcrowding in the nations emergency departments: complex causes and disturbing effects. Ann Emerg Med. 2000; 35:638.

17. Olshaker JS, Rathlev NK. Emergency department overcrowding and ambulance diversion: the impact and potential solutions of extended boarding of admitted patients in the emergency department. J Emerg Med. 2006; 30:3516. 18. Pines JM, Hollander JE, Localio AR, Metlay JP. The association between emergency department crowding and hospital performance on antibiotic timing for pneumonia and percutaneous intervention for myocardial infarction. Acad Emerg Med. 2006; 13:8738. 19. Pines JM, Localio AR, Hollander JE, et al. The impact of emergency department crowding measures on time to antibiotics for patients with community-acquired pneumonia. Ann Emerg Med. 2007; 50:5106. 20. Schull MJ, Vermeulen M, Slaughter G, Morrison L, Daly P. Emergency department crowding and thrombolysis delays in acute myocardial infarction. Ann Emerg Med 2004; 44:57785. 21. Pines JM, Hollander JE. ED crowding is associated with delays and non-treatment for patients with severe pain. Ann Emerg Med 2008; 51:15. 22. Bernstein SL, Asplin BR. Emergency department crowding: old problem, new solutions. Emerg Med Clin North Am. 2006; 24:82137. 23. Pines JM, Iyer S, Disbot M, Hollander JE, Shofer FS, Datner EM. The effect of emergency department crowding on patient satisfaction for admitted patients. Acad Emerg Med. 2008; 15:82531. 24. Sprivulis PC, Da Silva JA, Jacobs IG, Frazer AR, Jelinek GA. The association between hospital overcrowding and mortality among patients admitted via Western Australian emergency departments. Med J Aust. 2006; 184:20812. 25. Richardson DB. Increase in patient mortality at 10 days associated with emergency department overcrowding. Med J Aust. 2006; 184:2136. 26. Chaln DB, Trzeciak S, Likourezos A, Baumann BM, Dellinger RP, DELAY-ED study group. Impact of delayed transfer of critically ill patients from the emergency department to the intensive care unit. Crit Care Med. 2007; 35:147783. 27. Carr BG, Kaye AJ, Wiebe DJ, Gracias VH, Schwab CW, Reilly PM. Emergency department length of stay: a major risk factor for pneumonia in intubated blunt trauma patients. J Trauma. 2007; 63:912. 28. Fishman PE, Shofer FS, Robey JL, et al. The impact of trauma activations on the care of emergency department patients with potential acute coronary syndromes. Ann Emerg Med. 2006; 48:34753. 29. Diercks DB, Roe MT, Chen AY, et al. Prolonged emergency department stays of non-ST-segmentelevation myocardial infarction patients are associated with worse adherence to the American College of Cardiology American Heart Association Guidelines for management and increased adverse events. Ann Emerg Med. 2007; 50:48996. 30. Hollander JE, Pines JM. The emergency department crowding paradox: the longer you stay, the less care you get. Ann Emerg Med. 2007; 50:4979. 31. Baxt WG, Shofer FS, Sites FD, Hollander JE. A neural network for the prediction of cardiac ischemia

ACAD EMERG MED July 2009, Vol. 16, No. 7

www.aemj.org

625

32.

33.

34.

35.

36.

37.

in patients presenting to the emergency department with chest pain. Ann Emerg Med. 2002; 40:57583. Hollander JE, Sites FD, Pollack CV Jr, Shofer FS. Lack of utility of telemetry monitoring for identication of cardiac death and life threatening ventricular dysrhythmias in low risk patients with chest pain. Ann Emerg Med. 2004; 43:716. Chase M, Robey JL, Zogby KE, Sease KL, Shofer FS, Hollander JE. Prospective validation of the Thrombolysis in Myocardial Infarction risk score in the emergency department chest pain patient population. Ann Emerg Med. 2006; 48:2529. Hollander JE, Singer AJ. An innovative strategy for conducting clinical research the academic associate program. Acad Emerg Med. 2002; 9:1347. Lyon R, Morris AC, Caesar D, Gray S, Gray A. Chest pain presenting to the emergency department-to stratify risk with GRACE or TIMI? Resuscitation 2007; 74:903. Hollander JE, Blomkalns AL, Brogan GX, et al., Multidisciplinary Standardized Reporting Criteria Task Force Standardized reporting guidelines for studies evaluating risk stratication of ED patients with potential acute coronary syndromes. Acad Emerg Med. 2004; 11:133140. ACC Writing Committee for Acute Coronary Syndromes Clinical Data Standards. American College of Cardiology key data elements and denitions for measuring the clinical management and outcomes

38.

39.

40.

41.

42.

43.

of patients with acute coronary syndromes. J Am Coll Cardiol 2001; 30:21149. McCarthy ML, Aronsky D, Jones ID, et al. The emergency department occupancy rate: a simple measure of emergency department crowding? Ann Emerg Med. 2008; 51:1524. Peberdy MA, Ornato JP, Larkin GL, et al. National Registry of Cardiopulmonary Resuscitation Investigators. Survival from in-hospital cardiac arrest during nights and weekends. JAMA 2008; 299:78592. Bradley EH, Herrin J, Elbel B, et al. Hospital quality for acute myocardial infarction: correlation among process measures and relationship with short-term mortality. JAMA 2006; 296:728. Werner RM, Bradlow ET. Relationship between Medicares hospital compare performance measures and mortality rates. JAMA 2006; 296:2694702. Pollack CV Jr, Antman EM, Hollander JE, American College of Cardiology; American Heart Association. 2007 focused update to the ACC AHA guidelines for the management of patients with ST-segment elevation myocardial infarction: implications for emergency department practice. Ann Emerg Med. 2008; 52:34455. Pines JM, Synnestvedt M, Weiner MG, et al. Prolonged ED length of stay is associated with longer hospital stays, higher costs and workload in hospitalized older adults [abstract]. Acad Emerg Med. 2008; 15:S128.