Association of emergency department crowding with inpatient outcomes
Hsuan C, Segel JE, Hsia RY, et al. Health Serv Res. Published online September 26, 2022. doi:10.1111/1475-6773.14076
The key idea underpinning this article is that ED crowding is not a localized problem to be endured by ED patients and staff. Instead, crowding may have important implications for patients in the hospital regardless of whether they were admitted through the ED, and, hospital leadership should view it as such.
This article assessed whether the relationship between ED crowding on the day of hospital discharge is related to a variety important of outcomes for hospitalized patients, specifically (1) inpatient length of stay, (2) ED revisitation within 3 days, (3) readmission within 30 days, and (4) inpatient mortality.
Conceptually, ED crowding may induce inpatient providers to discharge patients more quickly, thus decreasing the length of stay (LOS), but could in turn increase the revisitation rate if the patients were not ready for discharge. Alternatively, busy days may be extremely challenging for hospitals, thus resulting in staff experiencing stress in trying determine whom to discharge, case managers being overextended, and physicians being too occupied with taking care of new admissions to attend to discharges. Consequently, the LOS is prolonged, but the effect on revisitation rates may not be predictable, and the situation might differ depending on whether the inpatient admission is elective vs unscheduled (ED).
The data in this study come from California’s Healthcare Access and Information (formerly OSHPD) data set from 2015 to 2017. This system requires detailed submissions of all inpatients and ED encounters by hospital, regardless of payer status.
The sample was restricted to adult patients and excluded some of the very small hospitals in California. The key outcome variables were (1) LOS, (2) revisitation rates to the ED, (3) hospital readmission, and (4) patient mortality; the key predictor was ED crowding on the day of discharge. Crowding in this study was fairly crudely defined as the ED census on that day, rather than other more direct measures of crowding (such as NEDCOS scores), because the data were not available. Crowding was defined in relation to normal conditions. For example, if a hospital had an average ED census of 100, any value above 100 was considered more crowded. The authors generated 50th percentile, 75th percentile, and 90th percentile crowding levels for each hospital and used these as the predictor.
Their statistical models were adjusted for hospital size, teaching status, and other observable variables, and they used a variety of reasonable econometric techniques and sensitivity analyses to validate the robustness of the findings.
Ultimately, data on 5,784,253 inpatient admissions from 307 hospitals through the study period were analyzed.
The top-line findings were that hospital LOS increased with increasing crowding. Although the increase was not large, a dose-response relationship was observed, in which LOS increased by approximately 1% as the crowding increased. Interestingly, higher ED census was associated with a slightly lower readmission rate at 30 days but no change in the ED revisitation rate. A higher risk of mortality was observed on crowded than uncrowded days.
The effects markedly differed depending on whether the admission was elective vs unscheduled. The longer LOS effect was driven by elective admissions (ie, patients with elective admission waited longer to be discharged), whereas the higher mortality and lower readmission effects were concentrated in patients with unscheduled admissions.
Why ED crowding on any given day would be associated with a higher risk of death is unclear. The finding that readmissions were lower when the ED was crowded seems somewhat odd. In addition, the effect sizes are fairly small overall, possibly because only crowding on the day of discharge was examined. The effects of a crowded ED may accumulate over days at a hospital, such that a patient exposed to multiple crowded days might be more affected than a patient exposed for only 1 day. A stronger strategy might have been to examine the cumulative effect of crowding rather than that on the discharge day. The findings may simply highlight that ED crowding stresses the entire hospital and affects the outcomes of patients who are already admitted to the hospital; consequently, hospital leaders must understand this effect and develop genuine interest in strategies to mitigate it.
EDITOR’S COMMENTARY: This is a huge study examining the relationship between ED crowding and patient outcomes. The findings suggest that ED crowding has small but measurable effects on a variety of patient outcomes including inpatient LOS and in-patient mortality, which seemingly should motivate hospital leaders to help relieve this problem.