Emergency department (ED) overcrowding has been a pain point in hospitals across the globe. "Frequent flyers," who visited the ED at a much higher rate than average, account for almost one third of ED visits even though they represent only a small proportion of all ED patients. In this study, we used data-mining methods to cluster ED frequent flyers at a large academic medical center in the US. The objective was to identify distinct types of frequent flyers, and the common characteristics associated with each type. The results show that the frequent flyers at the ED have three subgroups each exhibiting distinct characteristics: (1) the elderly with chronic health conditions, (2) middle-aged males with unhealthy behavior, and (3) adult females who are generally healthy. These findings may inform targeted interventional strategies for patients of each subgroup, who likely have distinct reasons for visiting the ED frequently, to reduce ED overcrowding.
Keywords: Cluster Analysis; Data Mining; Hospital Emergency Service.