TY - JOUR T1 - Comparison of the International Crowding Measure in Emergency Departments (ICMED) and the National Emergency Department Overcrowding Score (NEDOCS) to measure emergency department crowding: pilot study JF - Emergency Medicine Journal JO - Emerg Med J SP - 307 LP - 312 DO - 10.1136/emermed-2014-203616 VL - 33 IS - 5 AU - Adrian Boyle AU - Gary Abel AU - Pramin Raut AU - Richard Austin AU - Vijayasankar Dhakshinamoorthy AU - Ravi Ayyamuthu AU - Iona Murdoch AU - Joel Burton Y1 - 2016/05/01 UR - http://emj.bmj.com/content/33/5/307.abstract N2 - Introduction There is uncertainty about the best way to measure emergency department crowding. We have previously developed a consensus-based measure of crowding, the International Crowding Measure in Emergency Departments (ICMED). We aimed to obtain pilot data to evaluate the ability of a shortened form of the ICMED, the sICMED, to predict senior emergency department clinicians’ concerns about crowding and danger compared with a very well-studied measure of emergency department crowding, the National Emergency Department Overcrowding Score (NEDOCS).Methods We collected real-time observations of the sICMED and NEDOCS and compared these with clinicians’ perceptions of crowding and danger on a visual analogue scale. Data were collected in four emergency departments in the East of England. Associations were explored using simple regression, random intercept models and models accounting for correlation between adjacent time points.Results We conducted 82 h of observation in 10 observation sets. Naive modelling suggested strong associations between sICMED and NEDOCS and clinician perceptions of crowding and danger. Further modelling showed that, due to clustering, the association between sICMED and danger persisted, but the association between these two measures and perception of crowding was no longer statistically significant.Conclusions Both sICMED and NEDOCS can be collected easily in a variety of English hospitals. Further studies are required but initial results suggest both scores may have potential use for assessing crowding variation at long timescales, but are less sensitive to hour-by-hour variation. Correlation in time is an important methodological consideration which, if ignored, may lead to erroneous conclusions. Future studies should account for such correlation in both design and analysis. ER -