Validation of the short form of the International Crowding Measure in Emergency Departments: an international study

Abstract

Objective There is little consensus on the best way to measure emergency department (ED) crowding. We have previously developed a consensus-based measure, the International Crowding Measure in Emergency Departments. We aimed to externally validate a short form of the International Crowding Measure in Emergency Department (sICMED) against emergency physician’s perceptions of crowding and danger. Methods We performed an observational validation study in seven EDs in five different countries. We recorded sICMED observations and the most senior available emergency physician’s perceptions of crowding and danger at the same time. We performed a times series regression model. Results A total of 397 measurements were analysed. The sICMED showed moderate positive correlations with emergency physician’s perceptions of crowding, r = 0.4110, P < 0.05) and safety (r = 0.4566, P < 0.05). There was considerable variation in the performance of the sICMED between different EDs. The sICMED was only slightly better than measuring occupancy or ED boarding time. Conclusion The sICMED has moderate face validity at predicting clinician’s concerns about crowding and safety, but the strength of this validity varies between different EDs and different countries.

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

“…There is no gold standard measurement of ED crowding so, like multiple previous studies, we used ED physician perception [4][5][6]. Physician perception is pragmatic and reflects observations on time, occupancy and workload, although we found it had only moderate interobserver reliability [8].…”

“…Research assistants documented six single measures, including ED census (total patients), patients in beds, patients in the waiting room, patients in treatment areas awaiting physician assessment, number of inpatients boarding, and ED occupancy, defined as ED census divided by staffed ED beds (42 during the day, 24 overnight). Research assistants also compiled three multivariable crowding scores: EDWIN (ED Work Index), NEDOCS (National ED Overcrowding Score), and ICMED (International Crowding Measure in EDs) [4][5][6], as well as our local saturation calculator (SAT) based on NEDOCS. Crowding score descriptions are available in the Methods Supplement.…”

“…Our reference standard for crowding was physician perception, assessed using a 10 cm visual analogue scale (VAS) (0 = not crowded; 10 = extreme crowding), by asking the in-charge physician the following question: How crowded do you think this ED is at the moment? [6] Physicians were unaware of other measures being concurrently tracked. The team lead nurse independently provided a VAS crowding assessment for each relevant interval, and we calculated intraclass correlation coefficient to assure interrater reliability [7].…”

“…Indicators of crowding should be simple, easily available in real-time, and maintain reliability and accuracy across many EDs [3]. We examined the predictive validity of six single measures, as well as four complex crowding scores [4][5][6], for identifying ED crowding at a tertiary Canadian hospital. and performed receiver operating characteristic (ROC) curve analysis to examine predictive validity.…”