Predictors of mental health service costs for representative cases of psychosis in south London

Abstract

Background. Increasingly, evaluations of mental health services include an economic component, although often only summary statistics such as the mean or median are reported. Measures of variation are often limited to the standard error or standard deviation, though costs are rarely normally distributed and vary substantially between patients. The aim of this study is to identify factors that can explain variations in the cost of mental health services for epidemiologically representative cases of psychosis.

Methods. Cases with ICD-10 diagnoses of functional psychosis were drawn from a sample that included all identified cases of psychosis in two geographically defined sectors in Camberwell, south London. Mental health service use was measured and costed. A predictive model was developed using multiple regression analyses, in which patient characteristics and previous service use indicators were used as predictor variables. Services were measured using the Client Service Receipt Interview.

Results. Among the 147 cases included, the amount of cost variation explained by the model was 31·5%. The most significant predictive factor was social functioning, which was highly negatively associated with cost.

Conclusions. Current mental health service use can be predicted to a reasonable extent by previous service use and patient characteristics, especially the degree of social disability. Identification of such factors can aid the effective allocation of scarce resources. In particular, subgroups of patients who use most resources can be identified and targeted by mental health services.