Cluster analysis to identify targeted interventions for highly multimorbid patients in a one-million population in England.

Whole population segmentation can be a valuable asset to help understand the general distribution of health needs and healthcare utilisation within a population. In the one million resident health system in and around Bristol (UK), existing work has revealed a five-segment model of the adult population in which, with worsening health, segments halve in size and double in per-person spend. The top segment contains approximately 3% of the population but consumes over one-fifth of healthcare spend. With poor health outcomes and large costs, this high-multimorbidity cohort presents an opportunity to improve the 'value' extracted from available healthcare resources. Clinical management interventions considered for application to this segment should, however, appreciate its high heterogeneity with respect to health need, healthcare usage, and demographic and social characteristics. To support the identification of appropriately tailored interventions, we further partition this segment using cluster analysis (specifically, hierarchical density-based clustering following dimensionality reduction). This yields six clusters which, based on their defining features, are named: Older without dementia (n = 1,243); Cancer (n = 1,055); Younger and complex (n = 2,199); Respiratory (n = 4,743); Cardiovascular (n = 6,140); Dementia (n = 4,579). Following clinical review, each cluster is provided with a value-based objective, to improve outcomes and/or reduce costs. Additionally, a targeted set of possible clinical management interventions is suggested. This paper helps to address a deficit within the current literature of clinically contextualised clustering studies of highly multimorbid individuals.