Application of a disease-specific mapping function to estimate utility gains with effective treatment of schizophrenia

BACKGROUND Most tools for estimating utilities use clinical trial data from general health status models, such as the 36-Item Short-Form Health Survey (SF-36). A disease-specific model may be more appropriate. The objective of this study was to apply a disease-specific utility mapping function for schizophrenia to data from a large, 1-year, open-label study of long-acting risperidone and to compare its performance with an SF-36-based utility mapping function. METHODS Patients with schizophrenia or schizoaffective disorder by DSM-IV criteria received 25, 50, or 75 mg long-acting risperidone every 2 weeks for 12 months. The Positive and Negative Syndrome Scale (PANSS) and SF-36 were used to assess efficacy and health-related quality of life. Movement disorder severity was measured using the Extrapyramidal Symptom Rating Scale (ESRS); data concerning other common adverse effects (orthostatic hypotension, weight gain) were collected. Transforms were applied to estimate utilities. RESULTS A total of 474 patients completed the study. Long-acting risperidone treatment was associated with a utility gain of 0.051 using the disease-specific function. The estimated gain using an SF-36-based mapping function was smaller: 0.0285. Estimates of gains were only weakly correlated (r = 0.2). Because of differences in scaling and variance, the requisite sample size for a randomized trial to confirm observed effects is much smaller for the disease-specific mapping function (156 versus 672 total subjects). CONCLUSION Application of a disease-specific mapping function was feasible. Differences in scaling and precision suggest the clinically based mapping function has greater power than the SF-36-based measure to detect differences in utility.