Comparison of Health State Utilities Using Community and Patient Preference Weights Derived from a Survey of Patients with HIV/AIDS

Abstract

Objectives. The authors compare health state utilities derived from a national survey of patients with HIV/AIDS to represent community-based preferences with utilities derived from the same survey representing patient preferences; explore the relationships between these utilities and the dimensions of the SF-6D health state classification; and examine the implications of differences in the source of utilities for a cost-effectiveness analysis of early treatment of patients with HIV/AIDS. Methods. The authors derived community-based standard gamble (SG) utilities using an algorithm developed for the SF-6D health state classification system. The authors derived patient SG utilities from HIV/AIDS patient rating scale self-assessments using a power transformation. Data were from the HIV Cost and Services Utilization Study, a probability sample of 2864 HIV-infected adults receiving care in the United States in 1996. Results. Patient SG utilities were higher than community SG utilities by 4% to 9% (0.979 vs.0.937, 0.910 vs. 0.841, and 0.845 vs. 0.778; P < 0.001 for all comparisons). In regression analyses, patient SG utilities were less influenced by physical functioning, pain, and mental health dimensions of the SF-6D. The base case results of a cost-effectiveness analysis comparing early antiretroviral therapy to deferred therapy were unaffected by the choice between community ($20,100 per quality-adjusted life year) and patient ($18,400 per quality-adjusted life year) perspectives. The impact of the choice of utilities remained small in sensitivity analyses that varied the influence of treatment side effects on utilities and the initial symptom status of patients. Conclusion. There are differences between community and patient utilities for HIV/AIDS health states, although even when treatment side effects are important, these differences may not affect cost-effectiveness ratios.