Multimorbidity indices improve the prediction of consultation costs in primary care in the UK


INTRODUCTION: As the population ages, the number of people with multiple problems is likely to increase. There is growing recognition of the importance of multimorbidity as an entity in itself and a belief that management of patients with multiple problems may require resources over and above those needed to treat each condition individually. This has implications for budgetary management in primary care; however no standard measure of multimorbidity for this purpose has been established. A number of indices of multimorbidity have been derived with the main aims of describing prevalence and predicting outcomes. These have focused predominantly on secondary care in the United States. In this study we aim to investigate the relationship between multimorbidity and consultation costs in primary care in the UK. This is the first study we are aware of which investigates the relationship between multimorbidity and costs in this setting. METHODS: We used data on a stratified sample of 85,709 individuals aged over 18 years from 182 practices in the General Practice Research Database (GPRD). We used all historic patient diagnoses to measure patient-level multimorbidity using: QOF chronic disease count; Charlson Index score; count of Expanded Diagnostic Clusters (EDCs) identified by the John Hopkins ACG System; and average number of drugs prescribed per year over a 2-year period. We estimated patient-level cost of primary care consultations over the subsequent 12 months. We related cost to age, sex, deprivation and multimorbidity using Generalised Linear Models (GLMs), and assessed model performance using a variety of fit statistics including a deviance-based R-squared measure. RESULTS: The model including age, sex, deprivation and practice ID alone explained 11% of observed consultation costs. Inclusion of the number of prescribed drugs, count of EDCs, QOF disease count, or Charlson Index score increased this to 27%, 21%, 18%, and 14% respectively. All models suggest a reasonably linear relationship between consultation costs and the number of chronic conditions. CONCLUSIONS: Multimorbidity indices improve the prediction of future consultation costs. These indices can be constructed easily using routinely recorded General Practice data, and therefore use of these indices could help to improve budgetary management in primary care.

Proceedings of the 40th Society for Academic Primary Care Annual Scientific Meeting
Bristol, UK