Elizabeth Williams ^1,2* Syaribah N Brice ^1,2 Dave Price ²

• ¹ Cardiff University, Cardiff, United Kingdom
• ² Cardiff and Vale University Health Board, Cardiff, United Kingdom

Objective: Emergency department (ED) frequent attenders (FA) have been the subject of discussion in many countries. This group of patients have contributed to the high expenses of health services and strained capacity in the department. Studies related to ED FAs aim to describe the characteristics of patients such as demographic and socioeconomic factors. The analysis may explore the relationship between these factors and multiple patient visits. However, the definition used for classifying patients varies across studies. While most studies used frequency of attendance to define the FA, the derivation of the frequency is not clear. We propose a mathematical methodology to define the time interval between ED returns for classifying FAs. K-means clustering and the Elbow method were used to identify suitable FA definitions. Recursive clustering on the smallest time interval cluster created a new, smaller cluster and formal FA definition. Results: Applied to a case study dataset of approximately 336,000 ED attendances, this framework can consistently and effectively identify FAs across EDs. Based on our data, a FA is defined as a patient with three or more attendances within sequential 21-day periods. This study introduces a standardised framework for defining ED FAs, providing a consistent and effective means of identification across different EDs. Furthermore, the methodology can be used to identify patients who are at risk of becoming a FA. This allows for the implementation of targeted interventions aimed at reducing the number of future attendances.