Comparing Methods to Identify which Adult Emergency Department Visits Might be Avoided: A Retrospective Analysis of the MIMIC-IV-ED Database

BackgroundThere are multiple ways to define and identify potentially avoidable emergency department visits, making it difficult to estimate their frequency accurately. In this study, we compared the proportions and characteristics of potentially avoidable ED visits identified using three commonly used algorithms. MethodsWe analyzed the publicly available Medical Information Mart for Intensive Care IV Emergency Department dataset (MIMIC-ED) to estimate the proportions and characteristics of potentially avoidable ED visits identified using i) the eventual discharge diagnosis (Diagnosis), ii) the use of hospital resources (Resources), and iii) the triage acuity score assigned to the patient during emergency department triage (Triage-Acuity). We found that the proportions and characteristics of potentially avoidable ED visits were affected by the algorithm used to identify them. ResultsThe proportion of visits identified as potentially avoidable differed significantly between algorithms (2 - 21%), and few visits (<2%) were identified as potentially avoidable by all three algorithms. Presenting complaints, discharge diagnoses, and patient demographics were all similarly affected. ConclusionMethods for identifying potentially avoidable ED visits are not interchangeable. Choosing the appropriate definition and classification method will require researchers to carefully consider the types of visits and the characteristics of patients they wish to identify.