ORIGINAL RESEARCH CONTRIBUTION
Pediatric Emergency Department Crowding Is Associated With a Lower Likelihood of Hospital Admission Kenneth A. Michelson, MD, Michael C. Monuteaux, ScD, Anne M. Stack, MD, and Richard G. Bachur, MD
Abstract Objectives: Emergency department (ED) crowding may affect disposition decision-making. The objective was to measure the effect of ED crowding on probability of admission and return visit to the ED after discharge. Methods: The authors studied a historical cohort at a large pediatric ED over 40 months. Each patient was assigned a score on arrival based on the ED occupancy rate (the ratio of patients to beds). Patients were divided into quintiles by occupancy rate. The proportion admitted for each quintile was compared to the least crowded quintile adjusting for acuity, hospital occupancy, and time of arrival. The same analysis was performed for return visits to the ED within 48 hours. The analyses were repeated for the subsets of patients with asthma and with gastroenteritis and ⁄ or dehydration. Results: From the 40 months of historical data, 198,778 visits were analyzed. The adjusted odds ratio (aOR) for admission among the whole cohort was 0.85 (95% confidence interval [CI] = 0.81 to 0.89) comparing the highest to the lowest crowding quintiles (occupancy rate > 1.17 and < 0.54, respectively). For asthma patients, aOR = 0.93 (95% CI = 0.72 to 1.20), and for gastroenteritis patients, aOR = 0.87 (95% CI = 0.65 to 1.17). The aOR of return visits comparing the highest to the lowest crowding quintiles for all patients was aOR = 0.87 (95% CI = 0.79 to 0.97), for asthma patients was aOR = 1.52 (95% CI = 0.95 to 2.46), and for gastroenteritis patients was aOR = 0.83 (95% CI = 0.54 to 1.28). Conclusions: Increasing ED crowding is associated with a lower likelihood of hospital admission and lower frequency of return visits within 48 hours. ACADEMIC EMERGENCY MEDICINE 2012; 19:816–820 ª 2012 by the Society for Academic Emergency Medicine
D
ecisions regarding disposition of emergency department (ED) patients are complex. They are influenced by patient-specific factors (such as diagnosis, medical condition, existing care protocols, and medical resources) and ED provider-specific factors (such as knowledge, experience, or self-defined thresholds for admission).1–4 For patients with serious conditions, the decision is straightforward; however, many From the Division of Emergency Medicine, Children’s Hospital Boston, Boston, MA. Received December 22, 2011; revision received February 28, 2012; accepted March 1, 2012. This research was presented at the Pediatric Academic Societies meeting, Denver, CO, April 2011. The authors have no relevant financial information or potential conflicts of interest to disclose. Supervising Editor: Christopher R. Carpenter, MD, MSc. Address for correspondence and reprints: Kenneth A. Michelson, MD; e-mail:
[email protected].
816
ISSN 1069-6563 PII ISSN 1069-6563583
patients are not admitted to an intensive care unit (ICU) or surgical suite. In other cases, the decision is not based on objective, universally accepted criteria. Although to our knowledge not previously investigated, ED crowding may also influence decisions about a patient’s disposition. Among the known factors associated with ED crowding from studies in adults are increased morbidity, reduced quality, and decreased access.5–9 Increased ED length of stay (LOS) also is predictive of adverse outcomes, including higher mortality in the adult ICU.10 Pediatric studies about the effect of ED crowding on patient outcomes and disposition are limited. In one study, crowding severe enough to divert ambulances did not influence pediatric mortality, but other studies have shown that crowding adversely affects patient flow and is associated with suboptimal asthma and fracture care.11–14 Crowding may lead to changes in ED flow strategies or the behavior of providers; specifically, providers might adjust their threshold to admit or discharge
ª 2012 by the Society for Academic Emergency Medicine doi: 10.1111/j.1553-2712.2012.01390.x
ACADEMIC EMERGENCY MEDICINE • July 2012, Vol. 19, No. 7
•
www.aemj.org
patients in an effort to maintain ED flow. Incorporated into this assumption may be reduced time for observation, earlier decisions prior to full effect of therapeutics, or incomplete evaluations related to divided attention. Provider decision-making about patient disposition is not well studied. Investigating whether crowding influences provider decision-making is important to understand variations in care and to ensure patient safety. We sought to determine whether ED crowding influences the probability of admission and return ED visits within 48 hours. We hypothesized that crowding would increase the probability of admission. Because the entire ED population is very heterogeneous, we also chose to focus on two specific subgroups of patients, those with asthma or gastroenteritis. These two conditions were selected as common conditions that generally require a period of therapy and observation before a disposition decision is made and are therefore possibly sensitive to the effects of crowding on admission. METHODS Study Design We conducted a historical cohort study of all patient visits to a pediatric ED from April 2007 to August 2010. The institutional review board approved this study and waived informed consent for participants. Study Setting and Population We studied a large, high-acuity, urban, tertiary care pediatric ED with an annual volume of 58,000 visits. There are no observation beds in the ED, but there are eight additional urgent care rooms open during peak hours. The population is ethnically and socioeconomically diverse, composed of white (42%), black or African American (20%), Hispanic (21%), and other (17%) race ⁄ ethnicities. Forty-six percent of patients who visit the ED have private insurance, 43% have Medicaid, 8% are members of a health management organization, and 3% have no insurance. As a crude estimate of acuity, the ED has an admission rate of 17% to 19%, which has been stable for the past 5 years. Patients with asthma have an admission rate of 27%; those with gastroenteritis or dehydration have an admission rate of 30%. Study Protocol Every visit to the pediatric ED during the study period was eligible for inclusion in the study. We excluded visits with incomplete data. We also excluded those with a LOS longer than 24 hours from arrival to the ED, which in our system often resulted from a software failure or scheduled downtime of the tracking system, resulting in an invalid LOS. During the period of study, true boarders requiring ED LOS > 24 hours were limited to less than 20 patients per year and generally are children with primary psychiatric complaints awaiting placement, rather than medical admission. Each visit received an occupancy rate calculated at the arrival time, defined as the ratio of the number of patients to total beds in the ED.15 The numerator includes patients in beds and those in the waiting room. The denominator varied based on whether urgent care
817
rooms were open. An occupancy rate of 1 indicates there is one patient for each bed in the ED. In a previous study, six large adult EDs had occupancy rates of median, 0.8 to 1.1; 25th percentile, 0.6 to 0.8; and 75th percentile, 1.0 to 1.4.15 Possible confounders included hospital occupancy, the triage Emergency Severity Index (ESI) score assigned by a registered nurse at entry into the ED, and time of day.16 A single measure of hospital occupancy was available, and that was the census at midnight. Time of arrival was assigned as 6 AM to 12 PM, 12 PM to 6 PM, 6 PM to 12 AM, or 12 AM to 6 AM. We defined a return visit as a patient who presented for a second unplanned visit to the ED within 48 hours of the beginning of the index visit. Data were abstracted from an administrative database containing information on all ED visits. We defined an asthma subgroup with patients with International Classification of Diseases-9 (ICD-9) codes 493.0 through 493.9. We defined a gastroenteritis subgroup using patients with primary ICD-9 codes including those for gastroenteritis and for dehydration of 8.8 through 9.3, 558.0 through 558.9, 276.5 through 276.6, or 787.91. The primary author calculated occupancy rates for each patient through the following process. With arrival times and LOS data for each patient during the study period, a table was created with the number of patients in the ED at each minute. The number of patients at each minute was divided by the number of open beds in the ED to calculate an occupancy rate at that minute. The occupancy rate for each patient was then assigned by the time of arrival of the patient. The outcomes studied were hospital admission and return visits within 48 hours. These were calculated by dividing the number of patients admitted in the group by the number of patients in the group and similarly for return visits. Data Analysis The cohort of visits was divided into five equal-sized groups based on the quintiles of the assigned occupancy rate. We determined the mean LOS and 95% confidence interval (CI) for each quintile. The proportion admitted was calculated among the patients assigned to each quintile. For each outcome (admission and return visits), we compared the odds of the outcome for each of the four most crowded quintiles against the least crowded quintile using logistic regression, with adjustment for ESI score, hospital occupancy, and time of arrival. The use of quintiles allowed us to detect nonlinear relationships in the data. Statistical significance was defined as p < 0.05, and all tests were twotailed. This analysis was performed for the entire cohort and repeated within two subgroups: those with asthma and those with gastroenteritis. All statistics were computed using STATA 12.0 (Stata statistical software, Release 10.0, StataCorp., College Station, TX). RESULTS There were 200,901 visits eligible for inclusion in the study. A total of 794 patients (0.4%) were excluded for
818
Michelson et al.
having an LOS of more than 24 hours, and 1,329 patients (0.7%) were excluded for missing values for time of arrival, departure, or disposition. A total of 198,778 visits (98.9%) were analyzed. Table 1 shows the demographics of those included in the study. The mean age was 7 years, and the majority of patients (56%) had an ESI score of 3. Seventeen percent of patients in the cohort were admitted. From least to most crowded, the occupancy rate quintiles were defined as less than 0.54, 0.54 to 0.74, 0.75 to 0.93, 0.94 to 1.16, and 1.17 or greater. Median LOS and interquartile ranges (IQRs) in minutes among the whole cohort for each quintile from least to most crowded were 179 (IQR = 105 to 293), 191 (IQR = 119 to 292), 201 (IQR = 128 to 298), 214 (IQR = 141 to 312), and 239 (IQR = 165 to 339), respectively. Figure 1 depicts the percentage admitted and the percentage of patients with return visits within 48 hours for each crowding quintile across the entire sample, as well as within the asthma and gastroenteritis subgroups. Among the entire cohort, the patients in the two most crowded quintiles were less likely to be admitted than those in the least crowded quintile after adjusting for acuity, hospital occupancy, and time of arrival (adjusted odds ratios [aOR] = 0.89, 95% CI = 0.85 to 0.94; and aOR = 0.85, 95% CI = 0.81 to 0.89, respectively). No other significant differences were noted in the analysis of percentage admitted across quintiles in the entire cohort. Each of the four most crowded quintiles was less likely to return to the ED within 48 hours than the least crowded quintile, aOR = 0.89 (95% CI = 0.82 to 0.97), aOR = 0.91 (95% CI = 0.83 to 1.00), aOR = 0.90 (95% CI = 0.82 to 0.99), and aOR = 0.87 (95% CI = 0.790.97), in order of increasing crowding.
•
CROWDING AND ADMISSION PROBABILITY IN THE PEDIATRIC ED
Among the asthma and gastroenteritis subgroups, there were no significant differences in proportion admitted between the crowding quintiles, comparing the most crowded to the least crowded quintile: aOR = 0.93 (95% CI = 0.72 to 1.20) for asthma and aOR = 0.87 (95% CI = 0.65 to 1.17) for gastroenteritis. The proportion of return visits within 48 hours in the gastroenteritis subgroup also did not differ between crowding quintiles for the asthma subgroup (aOR = 1.52, 95% CI = 0.95 to 2.46, p = 0.08 comparing the most to the least crowded quintiles) or the gastroenteritis subgroup (aOR = 0.83, 95% CI = 0.54 to 1.28). DISCUSSION To the best of our knowledge, this is the first study to examine the association of pediatric ED crowding with the probability of admission. We had hypothesized that in a crowded ED, providers would have a more difficult time determining the best disposition for each patient, driving providers to have a lower threshold for admitting. This did not appear to be the case, and in fact in the most crowded times, the fraction of admitted patients decreased, even when adjusting for hospital occupancy, triage acuity, and time of day.17 This could potentially be explained by the fact that crowding was associated with longer ED LOS, which may lower the likelihood of admission given a tendency for acute illness in pediatrics to improve acutely during observation and treatment time.18 Providers could also preferentially discharge patients during times of crowding to free ED space for patients in the waiting room, knowing that discharging a patient usually requires less time than the admission process. This would be particularly true when the hospital is at high occupancy.
Table 1 Demographics of Patients Included in the Study All Patients N (%) Age 0–2 months 3–11 months 12–35 months 3–6 years 7–17 years ‡ 18 years ESI acuity score15 1 2 3 4 5 Arrival hour 6AM-12PM 12PM-6PM 6PM-12AM 12AM-6AM
Length of stay (minutes)* Return within 48 hours Admitted
198,778
Gastroenteritis With Dehydration
Asthma
4,199 (2.1)
6,488 (3.3)
9,729 20,144 39,423 40,157 74,903 14,409
(4.9) (10.1) (19.8) (20.2) (37.7) (7.2)
217 671 1,263 803 992 253
(5.2) (16.0) (30.1) (19.1) (23.6) (6.0)
9 348 1,693 1,951 2,200 286
(0.1) (5.4) (26.1) (30.1) (33.9) (4.4)
61 17,548 111,348 58,090 8,136
(0.0) (8.8) (56.0) (29.2) (4.1)
0 331 2,853 904 58
(0.0) (7.9) (67.9) (21.5) (1.4)
1 642 4,368 1,315 70
(0.0) (9.9) (67.3) (20.3) (1.1)
(19.1) (34.9) (36.0) (9.9) (131–308) (3.7) (17.8)
935 1,580 1,292 392 279 393 1,295
(22.3) (37.6) (30,8) (9.3) (191–380) (9.4) (30.8)
1,555 1,903 2,116 914 247 330 1,776
(24.0) (29.3) (32.6) (14.1) (177–328) (5.1) (27.4)
38,121 69,371 71,550 19,736 206 7,281 35,449
Values are reported as n (%) unless otherwise noted. ESI = emergency severity index; IQR = interquartile range. *Median (IQR).
ACADEMIC EMERGENCY MEDICINE • July 2012, Vol. 19, No. 7
•
www.aemj.org
819
studies suggest that progression of illness is the driver of return to the ED and subsequent admission, and this may prove true for asthma.14,19 We believe that closer study of asthma care in times of crowding warrants further study to determine whether asthma care results in a higher frequency of return and, if so, whether the driver is an inadequate period of observation after initial therapy or aspects of verbal or written discharge instructions or whether crowding leads to poor physician judgment. LIMITATIONS
Figure 1. Predicted percentage of patients admitted (top panel) and predicted percentage with return visits to the ED within 48 hours (bottom panel) adjusted for ESI acuity and hospital occupancy for each quintile of crowding. Estimates were adjusted for acuity, hospital occupancy, and time of arrival. Crowding increases from left to right. Error bars depict 95% CI of the mean. The bars represent all patients (dark gray), gastroenteritis (light gray), and asthma (white). ESI = Emergency Severity Index.
A potential worry given a lower admission rate in times of crowding is a counterbalanced higher probability of return visits. Potentially the amount of time spent assessing each patient or preparing him or her for discharge could be shortened during crowded conditions, leading to suboptimal care. However, in the full cohort, the proportion of return visits was lower for the patients in the most crowded quintile than for those in the least crowded. A family’s experience with crowding could decrease desire to return to the ED or increase use of primary care. A longer period of observation, or another unknown factor, may also be responsible for the decreased likelihood of return. The difference in likelihood of return visit did not meet the predetermined threshold for statistical significance; however, because it was close to that threshold, there is a greater likelihood of type II error. Prior
Interpretation of the percentage admitted across the entire study cohort is limited because of the heterogeneity of the patients. We did not adjust for time of year and case mix, which may bear heavily on this group. To focus the analysis, we selected two condition-specific subgroups. Additionally, the occupancy rate was assigned on arrival, which may be less determinant of the effect of crowding on decision-making than occupancy rate at the time of disposition decision. We may also have underestimated return visits if they were made to other institutions. Staffing during the day was also not included; however, prior studies have shown that the occupancy rate performs as well as scoring systems that take staffing into account.15 Finally, although we controlled for hospital occupancy using a midnight hospital census, the influence of hospital crowding is more dynamic, and therefore the adjustment is limited. The applicability of this study is also necessarily limited because the study was conducted at a single institution with a specific range of crowding and spectrum of acuity, although crowding at our institution is comparable to that at other large adult EDs.15 The ED occupancy rate is a convenient instrument for estimating crowding and has good predictive power for patients leaving without being seen and ambulance diversion in several adult EDs. As a pediatric ED, we have relatively few boarders and a low leaving without being seen rate and are never on ambulance diversion. The influence of crowding on physician behavior would likely be different based on institutional expectations related to wait times, thoroughness of ED evaluations prior to admission, and the capabilities of outpatient medical care in a particular region. Finally, outcomes in all groups may partly reflect local practices. CONCLUSIONS We found that ED crowding was associated with lower likelihood of admission at the most crowded times, independent of hospital occupancy. This effect could not be demonstrated for the specific subgroups of asthma or gastroenteritis. These results suggest that providers respond to crowding by changing clinical practice, with unpredictable effects. Understanding the factors that influence decisions regarding disposition is important for optimal and costeffective care. For some cases, the medical condition dictates the need for admission; however, as typical for pediatric medicine, exact admission criteria do not exist
820
Michelson et al.
for many routine illnesses, leading to considerable variability among ED providers. Drivers of admission other than crowding, such as the causes of interprovider variability, must be studied and understood in order to improve hospital and emergency care. References 1. Rogers AJ, Delgado CA, Simon HK. The effect of limited English proficiency on admission rates from a pediatric ED: stratification by triage acuity. Am J Emerg Med. 2004; 22:534–6. 2. Chamberlain JM, Patel KM, Pollack MM. Association of emergency department care factors with admission and discharge decisions for pediatric patients. J Pediatr. 2006; 149:644–9. 3. Mutrie D, Bailey SK, Malik S. Individual emergency physician admission rates: predictably unpredictable. CJEM. 2009; 11:149–55. 4. Jain S, Elon LK, Johnson BA, Frank G, Deguzman M. Physician practice variation in the pediatric emergency department and its impact on resource use and quality of care. Pediatr Emerg Care. 2010; 26:902–8. 5. Begley CE, Chang Y, Wood RC, Weltge A. Emergency department diversion and trauma mortality: evidence from Houston, Texas. J Trauma. 2004; 57:1260–5. 6. Burt CW, McCaig LF, Valverde RH. Analysis of ambulance transports and diversions among US emergency departments. Ann Emerg Med. 2006; 47:317–26. 7. Schull MJ, Vermeulen M, Slaughter G, Morrison L, Daly P. Emergency department crowding and thrombolysis delays in acute myocardial infarction. Ann Emerg Med. 2004; 44:577–85. 8. McCarthy ML, Zeger SL, Ding R, et al. Crowding delays treatment and lengthens emergency department length of stay, even among high-acuity patients. Ann Emerg Med. 2009; 54:511–3. 9. Richardson DB. Increase in patient mortality at 10 days associated with emergency department overcrowding. Med J Aust. 2006; 184:213–6.
•
CROWDING AND ADMISSION PROBABILITY IN THE PEDIATRIC ED
10. Chalfin 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;35:1477–83. 11. Shenoi RP, Ma L, Jones J, Frost M, Seo M, Begley CE. Ambulance diversion as a proxy for emergency department crowding: the effect on pediatric mortality in a metropolitan area. Acad Emerg Med. 2009; 16:116–23. 12. Timm NL, Ho ML, Luria JW. Pediatric emergency department overcrowding and impact on patient flow outcomes. Acad Emerg Med. 2008; 15:832–7. 13. Sills MR, Fairclough D, Ranade D, Kahn MG. Emergency department crowding is associated with decreased quality of care for children with acute asthma. Ann Emerg Med. 2011; 57:191–200. 14. Sills MR, Fairclough D, Ranade D, Kahn MG. Emergency department crowding is associated with decreased quality of care for children. Pediatr Emerg Care. 2011; 27:837–45. 15. McCarthy ML, Aronsky D, Jones ID, et al. The emergency department occupancy rate: a simple measure of emergency department crowding? Ann Emerg Med. 2008; 51:15–24. 16. Travers DA, Waller AE, Katznelson J, Agans R. Reliability and validity of the Emergency Severity Index for pediatric triage. Acad Emerg Med. 2009; 16:843–9. 17. Graham J, Aitken ME, Shirm S. Correlation of measures of patient acuity with measures of crowding in a pediatric emergency department. Pediatr Emerg Care. 2011; 27:706–9. 18. Roland D, Geelhoed G. Tertiary pediatric emergency department admission rates fall as waiting times increase. Pediatr Emerg Care. 2008; 24:125– 9. 19. Goldman RD, Kapoor A, Mehta S. Children admitted to the hospital after returning to the emergency department within 72 hours. Pediatr Emerg Care. 2011; 27:808–11.
