Constructing health state preference values from descriptive quality of life outcomes: mission impossible?

Abstract

Descriptive quality of life questionnaires are commonly administered in clinical trials, to evaluate outcomes from the patient's perspective alongside conventional clinical measures. When expressed in single index form as health state preference values (HSPVs), quality of life information is also relevant to economic evaluations. By combining HSPVs with survival information, quality adjusted life years (QALYs) may be derived for cost-utility analysis. Although HSPVs are rarely measured prospectively in cancer clinical trials, the UK Medical Research Council Cancer Therapy Committee recommends the routine administration of two specific quality of life questionnaires: the Rotterdam Symptom Checklist and the Hospital Anxiety Depression Scale. This study explores two potential methods for secondary derivation of HSPVs from these instruments, using data gathered in a clinical trial of two forms of radiotherapy for non-small cell cancer of the bronchus. The first method, secondary mapping to existing utility scales, was found to be infeasible from the above questionnaires. The second method used factor analysis to summarize the descriptive quality of life data collected through the questionnaires. This revealed five distinct factors prevalent in the trial population. Using these factors, simplified health state scenarios were developed from which direct measurement of HSPVs was feasible. As the resulting HSPVs and any QALYs that may be derived from them are cancer specific, their potential value in informing resource allocation would be limited to decisions within oncology services.