Abstract
BACKGROUND: Frequent users (FUs) of emergency departments (EDs) attend repeatedly, placing a disproportionate burden on healthcare systems. Although known to be heterogeneous, there is limited international evidence characterising FU subpopulations or examining how healthcare costs and outcomes differ across groups. Advancing this understanding is important for developing tailored interventions to meet diverse care needs. METHODS: FUs were defined as individuals with ≥5 ED attendances/year. We used two large UK datasets: Hospital Episode Statistics (HES, 2016-2019) and the Centre for Urgent and Emergency Care database (CUREd, 2017-2020). Together, these included over 148 000 FUs from 5 million ED users. Latent class analysis (LCA) was used to identify FU subgroups based on attendance patterns, healthcare use and diagnostic characteristics. RESULTS: We identified three consistent subgroups (HES and CUREd): (1) low-severity FUs (n=23 034, 43.2%; n=7081, 32.7%); (2) high-intensity FUs with mental health and neurological needs (n=6288, 11.8%; n=3456, 15.9%); (3) older FUs with chronic illness and high inpatient use (n=24 028, 45.0%; n=11 139, 51.4%). Subgroups differed substantially in healthcare utilisation, costs and mortality. A fourth class varied across datasets: in HES, it showed moderate morbidity and complex needs; in CUREd, high morbidity and high-intensity ED use. DISCUSSION: This is the first FU study to apply LCA across large-scale, multiyear ED datasets, identifying a potentially universal subgroup structure. Current services focus on a narrow subset of high-intensity users. Additional tailored strategies are needed to address the full spectrum of FU needs.