Emergency department occupancy is useful as a simple real-time measure of crowding.

INTRODUCTION

Emergency department (ED) crowding compromises patient outcomes. Existing crowding measures are complex and difficult to use in real-time. This study evaluated readily available single flow variables as crowding measures.

METHODS

Over 2 weeks in a tertiary Canadian ED, we recorded the following potential crowding measures during 168 consecutive two-hour study intervals: total ED patients (census), patients in beds, patients in waiting rooms, patients in treatment areas awaiting MD assessment; number of inpatients boarding, and ED occupancy. We also calculated four complex crowding scores-NEDOCS, EDWIN, ICMED, and a local modification of NEDOCS. We performed ROC analyses to assess the predictive validity of these measures against a reference standard of physician perception of crowding.

RESULTS

We gathered data for 144 (63.9%) of 168 study intervals. ED census correlated strongly with crowding (AUC = 0.82, 95% CI 0.76-0.89), as did ED occupancy (AUC = 0.75, 95% CI 0.66-0.83). Their performance was similar to NEDOCS (AUC = 0.80) and to the local modification of NEDOCS (AUC = 0.83).

CONCLUSION

ED occupancy as a single measure has similar predictive accuracy to complex crowding scores and is easily generalizable to diverse emergency departments. Real-time tracking of this simple indicator could be used to prompt investigation and implementation of crowding interventions.