Defining Hospital Catchment Areas Using Multiscale Community Detection: A Case Study for Planned Orthopaedic Care in England
BackgroundThe English National Health Service 5-year Forward View emphasises the importance of integration of hospital and community services. Understanding the population a hospital serves is critical to formulating strategies for community engagement and determining their accountability for populations. Existing methods to define catchment areas are unable to adapt to dilute health care markets in urban areas where populations may interact with several different hospitals. Formulating catchment areas which permit the inclusion of more than one hospital based upon patient behaviour allows for collaboration between hospitals to reach out into the communities they collectively share.\n\nMethodThe proportion of presentations from all census Middle Super Output Areas (MSOAs) to every hospital trust providing orthopaedic care in England were calculated. The cosine similarity of all MSOAs to one another was computed from these proportions. Multiscale community detection was applied to planned orthopaedic surgical admissions in England from 1st April 2011 to 31st March 2015. Stable community configurations were identified and the proportion of patients presenting to hospitals located within the catchment area in which they resided was calculated. The performance of these catchment areas was compared to conventional methods for assigning mutually exclusive catchment areas.\n\nResults2,602,066 planned orthopaedic surgical admissions were identified for patients resident in 6,791 MSOAs in England attending 140 different hospital trusts. Markov multiscale community detection revealed five stable catchment area configurations consisting of 127, 51, 26, 15 and 11 catchment areas. Between 78% (127 catchments) and 93% (11 catchments) of clinical presentations were to hospitals within a patients allocated catchment area compared to 76% for the \"first past the post\" method.\n\nConclusionsMultiscale community detection is a novel and effective, data-driven method for defining mutually exclusive, collectively exhaustive catchment areas in secondary care. In urban areas with dilute healthcare markets, the model favours collaboration between hospitals in covering a clearly delineated but shared catchment, and thereby produces simplified and more representative catchment areas.