The Effect of Emergency Department Crowding on Analgesia in Patients with Back Pain in Two Hospitals

Authors

  • Jesse M. Pines MD, MBA, MSCE,

    1. From the Department of Emergency Medicine (JMP), George Washington University School of Medicine and Department of Health Policy, George Washington School of Public Health and Health Sciences, Washington D.C.; Department of Emergency Medicine (FSS, JAI, SBA, AMM) and the Center for Clinical Epidemiology and Biostatistics (JMP), University of Pennsylvania School of Medicine, Philadelphia, PA.
    Search for more papers by this author
  • Frances S. Shofer PhD,

    1. From the Department of Emergency Medicine (JMP), George Washington University School of Medicine and Department of Health Policy, George Washington School of Public Health and Health Sciences, Washington D.C.; Department of Emergency Medicine (FSS, JAI, SBA, AMM) and the Center for Clinical Epidemiology and Biostatistics (JMP), University of Pennsylvania School of Medicine, Philadelphia, PA.
    Search for more papers by this author
  • Joshua A. Isserman MS,

    1. From the Department of Emergency Medicine (JMP), George Washington University School of Medicine and Department of Health Policy, George Washington School of Public Health and Health Sciences, Washington D.C.; Department of Emergency Medicine (FSS, JAI, SBA, AMM) and the Center for Clinical Epidemiology and Biostatistics (JMP), University of Pennsylvania School of Medicine, Philadelphia, PA.
    Search for more papers by this author
  • Stephanie B. Abbuhl MD,

    1. From the Department of Emergency Medicine (JMP), George Washington University School of Medicine and Department of Health Policy, George Washington School of Public Health and Health Sciences, Washington D.C.; Department of Emergency Medicine (FSS, JAI, SBA, AMM) and the Center for Clinical Epidemiology and Biostatistics (JMP), University of Pennsylvania School of Medicine, Philadelphia, PA.
    Search for more papers by this author
  • Angela M. Mills MD

    1. From the Department of Emergency Medicine (JMP), George Washington University School of Medicine and Department of Health Policy, George Washington School of Public Health and Health Sciences, Washington D.C.; Department of Emergency Medicine (FSS, JAI, SBA, AMM) and the Center for Clinical Epidemiology and Biostatistics (JMP), University of Pennsylvania School of Medicine, Philadelphia, PA.
    Search for more papers by this author

  • Presented at the Society for Academic Emergency Medicine annual meeting, New Orleans, LA, May 2009.

Address for correspondence and reprints: Jesse M. Pines, MD, MBA, MSCE; e-mail: jesse.pines@gmail.com.

Abstract

Objectives:  The authors assessed the association between measures of emergency department (ED) crowding and treatment with analgesia and delays to analgesia in ED patients with back pain.

Methods:  This was a retrospective cohort study of nonpregnant patients who presented to two EDs (an academic ED and a community ED in the same health system) from July 1, 2003, to February 28, 2007, with a chief complaint of “back pain.” Each patient had four validated crowding measures assigned at triage. Main outcomes were the use of analgesia and delays in time to receiving analgesia. Delays were defined as greater than 1 hour to receive any analgesia from the triage time and from the room placement time. The Cochrane-Armitage test for trend, the Cuzick test for trend, and relative risk (RR) regression were used to test the effects of crowding on outcomes.

Results:  A total of 5,616 patients with back pain presented to the two EDs over the study period (mean ± SD age = 44 ± 17 years, 57% female, 62% black or African American). Of those, 4,425 (79%) received any analgesia while in the ED. A total of 3,589 (81%) experienced a delay greater than 1 hour from triage to analgesia, and 2,985 (67%) experienced a delay more than 1 hour from room placement to analgesia. When hospitals were analyzed separately, a higher proportion of patients experienced delays at the academic site compared with the community site for triage to analgesia (87% vs. 74%) and room to analgesia (71% vs. 63%; both p < 0.001). All ED crowding measures were associated with a higher likelihood for delays in both outcomes. At the academic site, patients were more likely to receive analgesia at the highest waiting room numbers. There were no other differences in ED crowding and likelihood of receiving medications in the ED at the two sites. These associations persisted in the adjusted analysis after controlling for potential confounders of analgesia administration.

Conclusions:  As ED crowding increases, there is a higher likelihood of delays in administration of pain medication in patients with back pain. Analgesia administration was not related to three measures of ED crowding; however, patients were actually more likely to receive analgesics when the waiting room was at peak levels in the academic ED.

ACADEMIC EMERGENCY MEDICINE 2010; 17:276–283 © 2010 by the Society for Academic Emergency Medicine

In 2005 there were 2.4 million emergency department (ED) visits with a diagnosis of back pain in the United States.1 Emergency care for patients with symptoms of back pain typically involves 1) ruling out serious causes such as spinal cord compression, fractures, and other medical or traumatic causes of back pain; 2) attempting to identify the most likely cause, such as muscle strain or lumbago; and 3) symptomatic relief. After ED evaluation, only a small percentage of back pain cases are true medical emergencies requiring urgent intervention.2 Therefore, the focus in most cases is the provision of symptomatic relief in the form of pain medications, recommending outpatient regimens aimed at alleviating symptoms, and ensuring appropriate follow-up. Therefore, the timely use of analgesics, if desired, is an important measure of ED quality. However, studies have identified “oligoanalgesia,” or the undertreatment of acute painful conditions, as a significant issue in U.S. EDs.3–6

Several recent studies have identified a relationship between ED crowding and oligoanalgesia, where higher levels of crowding have been associated with both delays in treatment and lack of treatment.7–10 Specifically, higher crowding levels have been associated with worse pain control in patients with hip fracture or abdominal pain and in general ED populations. No studies, to our knowledge, have assessed the association between ED crowding and the use of analgesia or analgesia timing in patients with back pain. In addition, no studies have compared associations between ED crowding and the quality of pain management in more than one hospital.

We sought to study the association between ED crowding and the use of, and delays in, analgesia in patients with back pain in two EDs. We hypothesized that during times of higher ED crowding, patients with back pain would be less likely to receive analgesia and would be more likely to experience longer times to analgesia administration. In addition, we thought that the same measures of crowding would have similar effects on the use of and time to analgesia in the two EDs.

Methods

Study Design

We performed a retrospective analysis of all patients with a chief complaint of “back pain” as determined at triage from July 1, 2003, to February 28, 2007, in two EDs in the same health system that share the same ED-based electronic medical chart. For this investigation, we received an exemption from informed consent requirements for human subjects research by the institutional review board.

Study Setting and Population

The study was conducted at two urban, inner-city EDs. One ED is an academic, tertiary care ED with a 4-year emergency medicine residency training program and with approximately 57,000 annual adult visits, 25 treatment rooms, 15 hallway spaces, a separate eight-bed fast track, and a three-bed trauma bay. The second ED is a community ED in the same health system that sees both adult and pediatric cases, with approximately 35,000 annual visits with 20 treatment rooms, six hallway spaces, and three fast-track spaces. The community ED is staffed primarily by attending physicians. However, residents do rotate occasionally through the community ED. There was no change in size in either ED during the study period, nor was volume considerably different by study year.

All adult patients, 18 years of age or older, who visited either ED during the study period with back pain assigned by the triage nurse as the chief complaint were eligible for inclusion. Patients were excluded if they left without being seen, were pregnant, or had no documented pain score.

Study Protocol

The computerized medical record and order entry system EMTRAC (University of Pennsylvania, Philadelphia, PA) was used to identify patients in both EDs. Both hospitals use EMTRAC as their ED information system. The tracking, order entry, and charting systems are run identically at each site.

At triage, nurses can select from a predefined list of chief complaints, one of which is back pain. There are no other predefined complaints related directly to patients with symptoms of back pain. Patients with back pain are sometimes triaged with other predefined chief complaints; for example, if they have a complaint of back pain and fever, the triage nurse might choose “fever” as the chief complaint. However, only patients with a back pain chief complaint were used for this particular study because we thought that this would identify the most homogeneous population to test study aims. Patients who were seen by nurse practitioners were excluded because they sometimes use a different documentation system for medications. In addition, nurse practitioners see patients in a separate part of the academic ED (a fast-track area), which is less affected by levels of crowding in the main ED.

The following additional variables were extracted: age, sex, race, triage pain score, triage level, the administration of any initial pain medication, and the time of the administration. Triage pain scores were obtained by verbal report from the patient by the nurse who used an 11-point scale (0–10), from “no pain” (0) to “worst pain of my life” (10). Three classes of severity were created: low (0–3), moderate (4–7), and severe (8–10). Reassessments for pain were not available for this analysis, nor were additional doses of pain medications. All opioid or nonopioid pain medications were considered “analgesia.” Triage levels were assessed by the nurse on a four-point scale (1–4) based on the urgency of the complaint, where “1” indicated most emergent cases and “4” indicated least urgent.

EMTRAC was used to assign the following ED crowding measures at triage to each patient encounter: ED occupancy rate (percentage of ED beds filled including hallway spaces); total patient-care hours (defined as the arithmetic sum of the hours of all patients in the ED, including the waiting room, and excluding trauma and fast-track patients); number of patients in the waiting room; and number of admitted patients boarding in the ED (for whom a bed request had been entered). The calculation of the crowding variables was performed through Microsoft Access (Microsoft Corp., Redmond, WA) using queries that reconstructed the state of the ED at the time of triage for each patient. These crowding indicators have been associated with reduced quality of care in several other studies.8,10–13

The main study outcomes were whether patients received analgesia in the form of a pain medication during their ED visit, and if they did, the time to administration of this medication. If they received more than one pain medication, only the time to the first administration was considered. For time to medication administration, we used two outcomes to signify a delay: 1) greater than 1 hour from ED triage to analgesia administration and 2) greater than 1 hour from room placement to analgesia administration. These outcomes do overlap and were correlated (with a correlation coefficient = 0.68). However, they are analyzed separately because each is designed to reflect a different type of delay. Time from ED triage to analgesia encompasses the overall delay that the patient experiences and includes waiting room time. Time from room to analgesia is a waiting time that is independent of room availability and is more of a measure of doctor and nurse availability for evaluation, medication ordering, and administration. For patients receiving pain medication at the same time as documented triage, the time to pain medication was equal to zero. This was the case for 23 observations. Patients who did not receive any analgesia were excluded from the delays analysis. One-hour delays to analgesia have been used as a standard in several prior studies.7–10 The 1-hour cutoff was also used to permit use of binary regression techniques (relative risk [RR] regression) to allow for an interpretable and comparable risk ratio, compared with using a log transformation of time.

Data Analysis

Data are presented as means ± standard deviation (±SD) or frequencies with percentages. Times from triage or room placement to analgesia are presented as medians with interquartile ranges (IQRs) separately and for both hospitals combined. Because the level of crowding was considerably different at the two hospitals (the academic site had a higher overall level of crowding than the community site), each of the four crowding measures was divided into quartiles for each individual hospital. Fisher’s exact test and the Mann-Whitney U-test were used to determine if there were differences in the use of analgesia, time to analgesia, and delays in analgesia between the two EDs. We then tested if there was any difference in the use of analgesia among crowding quartiles in each hospital using a Cochrane-Armitage test for trend. We also tested if significant trends existed in time from triage to medication and time from room placement to medication using the Cuzick test for trend across ordered quartiles at each hospital.

Multivariable analysis was then performed separately for each hospital to assess the adjusted effects of crowding on the use of and delays (>1 hour) in analgesia from triage to analgesia and room to analgesia. For these analyses, the primary independent variables were crowding quartiles at each individual hospital. To calculate RR, a generalized linear model with a log link, Gaussian error, and robust estimates of the standard errors of the model coefficients were used. This model controlled for age, sex, black or African American race, triage class (as an indicator variable compared to the most severe triage score), time of day (7am–3pm, 3pm–11pm, 11pm–7am), study year, hospital, whether the patient was admitted to the hospital, and severity of pain (low, moderate, and severe). This list of confounders was determined a priori, and there were no stepwise techniques used to select variables. Given the large number of outcomes, the study was adequately powered for multivariable analysis. Data for these analyses are presented as RRs of the three outcomes with 95% confidence intervals (CIs), compared to the lowest quartile of crowding. There were very little missing data, so no imputation was performed. Model fit was assessed using the Akaike’s information criterion. All analyses were performed using Stata statistical software (Version 10, StataCorp, College Station, TX). To adjust for multiple statistical tests performed on the same data, the Bonferroni correction was used. An n of four was selected for the correction because there were four separate crowding quartiles tested. Therefore, a probability of <0.0125 was considered statistically significant.

Results

A total of 5,616 patients presented to the two EDs with a chief complaint of back pain during the study period. Patients were primarily male (57%), black or African American (62%), and young (mean ± sd age = 44 ± 17 years). The median pain score was 8 (IQR = 6–10; Table 1). Analgesia was administered to 79% of patients at both hospitals combined. Median triage time to analgesia was 130 minutes (IQR = 73 to 217 minutes), and 81% of these patients experienced a delay in analgesia from triage of >1 hour. Median room placement time to analgesia was 86 minutes (IQR = 51–135 minutes), and there was a delay in analgesia of >1 hour for 67% of these patients (Table 1). Over the study period, crowding levels as measured by waiting room number, patient-hours, and number of admitted patients were considerably higher at the academic hospital, while overall occupancy was similar. At the academic hospital, the use of analgesia was higher (82% vs. 74%, p < 0.001), delays were longer from triage to medication (147 minutes vs. 106 minutes, p < 0.001), and from room to medication (93 minutes vs. 77 minutes, p < 0.001), and the proportion of patients experiencing delays from triage (87% vs. 74%, p < 0.001) and room (71% vs. 63%, p < 0.001) was higher (Table 2).

Table 1. 
Demographic Characteristics of Patients Presenting to the ED With Back Pain at Two Inner-city Hospitals (n = 5,616)
Variable 
  1. Data are reported as n (%) unless otherwise stated.

  2. IQR = interquartile range.

  3. *Totals to 5,613.

Age, yr (mean ± SD)44 ± 17
Female3,188 (57)
Black or African American3,496 (62)
Pain score (median, IQR)6 (8–10)
Triage pain score
 Low (0–3)249 (4)
 Moderate (4–7)1,788 (32)
 Severe (8–10)3,579 (64)
Triage class*
 1 (emergent)42 (1)
 2823 (15)
 33,130 (56)
 4 (nonemergent)1,618 (29)
Time of day
 7:01am–3:00pm1,137 (20)
 3:01pm–11:00pm2,483 (44)
 11:01pm–7:00am1,996 (36)
Admitted to the hospital561 (10)
Received any analgesia4,425 (79)
Triage to first pain medication, minutes (median, IQR)130 (73–217)
Delay of >1 hour from triage to analgesia3,589 (81)
Room to first pain medication, minutes (median, IQR)86 (51–135)
Delay of >1 hour from room to analgesia2,985 (67)
Table 2. 
Crowding Levels, Administration of Any Analgesia, and Time to Analgesia in Back Pain Patients Presenting to Two Hospital EDs (n = 5,616)
 Academic Tertiary Care Hospital (n = 3,226)Community Teaching Hospital (n = 2,390)
  1. IQR = interquartile range.

Mean crowding scores (median, IQR)
 Admitted number  patients 10 (7–14)4 (2–6)
 Occupancy65% (50–80)68% (42–84)
 Patient-hours 108 (72–150) 38 (21–61)
 Waiting room number,  patients 7 (4–12)0 (0–3)
Outcome measures
 Any analgesia, n (%)2,650 (82)1,775 (74)
 Triage to analgesia,  minutes (median, IQR)147 (88–243)106 (59–175)
 Delay of >1 hour from  triage to analgesia, n (%)2,287 (87)1,302 (74)
 Room to analgesia, minutes (median, IQR)93 (55–142)77 (45–124)
  Delay of >1 hour from room to analgesia, n (%)1,879 (71)1,106 (63)

Emergency department crowding measures were not associated with the likelihood of receiving analgesia in the community hospital. There were significant differences in the likelihood of receiving pain medication at higher waiting room numbers at the academic hospital, where patients were actually more likely to be treated with analgesia at higher waiting room quartiles. Among those who received treatment, all the ED crowding measures were associated with a higher likelihood of delay in both time from triage to analgesia, and time from room placement to analgesia (p ≤ 0.001 for all measures; Tables 3 and 4).

Table 3. 
Proportion of Patients Receiving Any Analgesia and Median Time to Analgesia by Crowding Measure and Quartile in the Academic Tertiary Care ED (n = 3,226)
 Any Analgesia, %Triage to Analgesia (Minutes)Room to Analgesia (Minutes)
  1. Reported p-values for any analgesia represent a Fisher’s exact test. Reported p-values for triage and room to analgesia represent a nonparametric test for trend. p < 0.0125 is considered significant to account for multiple comparisons.

Admitted patients
 Quartile 182.512699
 Quartile 281.7158112
 Quartile 381.3192115
 Quartile 483.4253136
 p-value0.77<0.001<0.001
Occupancy rates
 Quartile 181.411594
 Quartile 281.6161114
 Quartile 381.4192117
 Quartile 484.4252133
 p-value0.35<0.001<0.001
Patient-hours
 Quartile 180.612799
 Quartile 282.0158112
 Quartile 384.3198120
 Quartile 481.7231125
 p-value0.26<0.001<0.001
Waiting room numbers
 Quartile 174.610093
 Quartile 278.912997
 Quartile 380.2173115
 Quartile 482.3242125
 p-value<0.001<0.001<0.001
Table 4. 
Proportion of Patients Receiving Any Analgesia and Median Time to Analgesia by Crowding Measure and Quartile in the Community Teaching Hospital ED (n = 2,390)
 Any Analgesia, %Triage to Analgesia (Minutes)Room to Analgesia (Minutes)
  1. Reported p-values for any analgesia represent a Fisher’s exact test. Reported p-values for triage and room to analgesia represent a nonparametric test for trend. p < 0.0125 is considered significant to account for multiple comparisons.

  2. *Represents combination of Quartiles 1 and 2.

Admitted patients
 Quartile 175.29890
 Quartile 276.7128104
 Quartile 374.3150103
 Quartile 472.0203107
 p-value0.35<0.0010.001
Occupancy rate
 Quartile 173.98078
 Quartile 275.2115101
 Quartile 374.3156111
 Quartile 474.0206110
 p-value0.95<0.001<0.001
Patient hours
 Quartile 174.98885
 Quartile 275.211698
 Quartile 372.5155103
 Quartile 474.5186111
 p-value0.71<0.001<0.001
Waiting room number
 Quartile 175.1*95*91*
 Quartile 2   
 Quartile 370.8133102
 Quartile 476.0219111
 p-value0.09<0.001<0.001

When data were analyzed separately for each hospital and adjusted for potential confounders, we found no association between the number of admitted patients, occupancy, or patient-hours and the likelihood of being treated with analgesia. The adjusted analysis for delays in treatment demonstrated that all crowding factors were associated with a higher likelihood of delays in analgesia when comparing the lowest quartiles of crowding to the second, third, and fourth quartiles for each variable. ED crowding measures had a greater impact on time from triage to analgesia at the community teaching site (Figures 1 and 2).

Figure 1.

 Adjusted RR of no analgesia or a delay in analgesia of >1 hour from triage time and room placement time for ED patients with back pain seen at the academic site (n = 3,226). Each value is an adjusted RR of the outcome compared to the lowest level of crowding with 95% CIs. All models account for the following covariates: age, sex, black or African-American race, hospital, time of day, pain category (low, moderate, severe), admission status, year of service, and triage class (1–4). RR = relative risk.

Figure 2.

 Adjusted RR of no analgesia or a delay in analgesia of >1 hour from triage time and room placement time for ED patients with back pain at the community site (n = 2,390). Each value is an adjusted RR of the outcome compared to the lowest level of crowding with 95% CIs. All models account for the following covariates: age, sex, black or African-American race, hospital, time of day, pain category (low, moderate, severe), admission status, year of service, and triage class (1–4). The reference for waiting room number is Quartiles 1 and 2 combined because both are equal to zero patients. RR = relative risk.

Discussion

We found that higher crowding levels in the ED were independently associated with delays in analgesia in adults with back pain, both from the time of triage and from the time of room placement. This confirms the findings of previous studies that demonstrate an association between ED crowding and delays in analgesia.7–10 While there were differences in the waits that patients experienced at the two EDs, there was a similar linear trend across all crowding measures for delays from triage to analgesia. Room to analgesia time followed a similar trend at both sites, with relatively linear associations between crowding levels and delays. However, for occupancy and number of admitted patients, there was a ceiling effect at the second quartile at the community site, which may indicate that the community ED is staffed better to attend to analgesic needs during the highest levels of occupancy and boarding. In addition, ED crowding measures appeared to have a greater effect on time from triage to analgesia at the community site, which may reflect the fact that it has lower overall capacity and fewer staff and may be less able to flex to the needs of greater capacity.

We recently published a similar paper from a prospective cohort study of patients with abdominal pain at the same academic ED and employed a similar methodology and measured identical outcomes relating to analgesia and delays in pain medication.10 There were a number of striking differences between the two patient populations. Patients with back pain were more likely than patients with abdominal pain to experience clinically significant delays in the administration of first pain medication. Median values for both triage to medication and room to medication for back pain ranged from 1.5 to 2 hours, while patients with abdominal pain are treated more promptly, with median values ranging from 1 to 1.5 hours. Differences in triage levels, waiting room times, or severity of illness may account for some of these differences. We also found that crowding appeared to have a greater effect on whether patients with abdominal pain experienced delays in treatment compared with back pain. During crowded times, providers might have a lower threshold to quickly evaluate and treat a patient with back pain than abdominal pain, which may require a longer evaluation.

For most crowding measures, ED crowding was not associated with a difference in the likelihood of being treated. However, at the highest waiting room numbers at the academic hospital, patients were, somewhat surprisingly, more likely to be treated. This may reflect the fact that patients who present to the ED for pain control are more likely to stay for care, even when the waiting room levels are very high. It makes sense that waiting room numbers would have affected this decision compared to other crowding measures because patients can see the number of other patients in the waiting room, while they do not have access to other measures of crowding. This effect persisted in the adjusted analysis after controlling for pain scores. Because pain scores obtained at triage do not perfectly predict the desire for ED-based analgesia, it may not adjust for the desire for analgesic medications.14 This effect was only seen at the academic site, which may reflect higher levels of crowding and longer waits. In addition, it is also possible that longer waits in and of themselves may make patients more frustrated and inclined to demand treatment for their pain. Left without being seen rates are also higher at the academic site and may account for some of this observed relationship. Another possible explanation for this effect is that because most cases of back pain are not caused by serious medical causes requiring emergency treatment such as surgery or other interventions, during crowded times clinicians may opt to just focus on symptom control and may spend less time evaluating patients. However, because there were no direct observations of clinicians in this study, we were unable to test these hypotheses.

This study also represents one of the first two-hospital studies that directly compares the effects of crowding on the same outcomes. Interestingly, the same levels of crowding had different effects on the probability of a delay in treatment at the two sites. A recent paper demonstrated similar findings in four EDs where the same level of occupancy had different effects on patient throughput times.15 ED occupancy is currently one of the leading contenders as a generalizable measure of ED crowding.16 Because the same levels of occupancy have different effects on both throughput and patient-oriented outcomes, it is possible that ED occupancy may not be an ideal measure of crowding that will generalize across multiple settings.17

Limitations

There were no reassessments of pain included in this analysis. Because pain can vary over time, we were unable to test how any evolution of pain may have influenced the decision to use analgesia or if crowding affected the likelihood of reassessment. Desire for analgesia was also not measured in this study, which may influence which patients decide to stay during crowded times and may also influence which patients get treated. Several populations of patients were excluded from the analysis, including those who left without being seen, those seen by nurse practitioners, and pregnant patients. Eliminating the “left without being seen” patients may have biased our study toward not finding an association between ED crowding and analgesia administration. Left without being seen rates are higher during times of higher crowding, and because left without being seen patients were systematically excluded, and by definition go untreated, crowding may be associated with a higher risk of nontreatment than was reported in this study. Excluding patients seen by nurse practitioners (approximately one-third of total patients with back pain from both EDs) likely biased our study toward showing longer delays in analgesia for patients seen in the main ED. This is because these patients would not have had to compete for resources with a more heterogeneous group of critically ill and non–critically ill patients, as opposed to fast track where turnaround times are traditionally more rapid. This study was also performed at only two urban institutions, which may limit the generalizability of these results to other settings.

Conclusions

In this study, ED crowding did not reduce the likelihood of receiving pain medication in a large cohort of patients with back pain in two EDs. However, crowding was associated with delays both from triage to analgesia and from room placement to analgesia at both EDs. At the academic site, patients who came to the ED at the highest waiting room numbers were, surprisingly, more likely to be treated. The relationship between crowding and delays was somewhat different in the two settings, which may indicate that static measures of crowding, such as ED occupancy, may not generalize because of their variable relationship with quality of care.

Ancillary