Occupancy rates and emergency department work index scores correlate with leaving without being seen

TWO CROWDING METRICS ARE OFTEN USED TO MEASURE EMERGENCY DEPARTMENT (ED) CROWDING: the occupancy rate and the emergency department work index (EDWIN) score. To evaluate these metrics for applicability in our community ED, we sought to measure their correlation with the number of patients who left without being seen (LWBS) and determine if either, or both, correlated with our daily LWBS rate. We hypothesized a statistically significant positive correlation between the number of patients who LWBS and both crowding metrics.

We performed a retrospective observational study by reviewing data on all patients who LWBS from December 1, 2007, to February 29, 2008. Occupancy rates and EDWIN scores were obtained through our electronic patient tracking board. We identified LWBS status by searching the final disposition entered into our electronic medical record. We measured the correlation between each crowding metric averaged over each 24-hour day and the number of patients who LWBS per 24-hour day using Spearman's rank correlation, and created receiver operator characteristic (ROC) curves to quantify the discriminatory power of occupancy rate and EDWIN score for predicting more than two patients per day who LWBS.

We identified 1,193 patients who LWBS during the study period, including patients who registered but then left the waiting room (733), as well as those who left before: registration (71), triage (75), seeing a physician (260), or final disposition (54). The number of patients who LWBS per day ranged from one to 30, with a mean of 13 and median of 11 (IQR 6 to 19). The daily number of patients who LWBS showed a positive correlation with the average daily occupancy rate (Spearman's rho = 0.771, p = 0.01) and with average daily EDWIN score (Spearman's rho = 0.67, p< .001). Area under the ROC curve for occupancy rate was .97 (95% CI .93 to 1.0) and for EDWIN score was .94 (95% CI .89 to 1.0).

Average daily occupancy rates and EDWIN scores both correlate positively with, and have excellent discriminatory power for, the number of patients who LWBS in our ED; however, the scale of our EDWIN scores differs from that obtained at other institutions. For studies of crowding, occupancy rate may be the more useful metric due to its ease of calculation.