Assessing the utility of five domains in SF-12 health status questionnaire in an AIDS clinical trial

Objective: To assess a shortened quality-of-life (QoL) measurement tool in a population with advanced HIV infection. Design: Five domains (seven items) in a 12-item questionnaire (SF-12) were compared with those same domains in a 39-item questionnaire (SF-39). Data were collected using SF-39 in a randomized clinical trial for the prevention of cytomegalovirus disease. Methods: The performance of SF-12 relative to SF-39 was evaluated within each domain by comparing QoL scores at baseline and over time, assessing the reliability and validity for both instruments, assessing item consistency and discrimination within instruments, and implementing event-time analyses that quantified dependence of the hazard for death and progression of disease (POD) on baseline values. Results: Baseline measures are similar for both instruments, with high correlation within each domain. The slopes over time for the SF-12 and SF-39 domains are also similar. Both the SF-12 and SF-39 domains have satisfactory reliabilities and perfect discrimination. The hazard ratios for death and POD are similar for both instruments within a domain. All SF-12 and most SF-39 domains are highly predictive for death but are not highly predictive for POD. Conclusions: For the domains considered, SF-12 is a reasonable and effective replacement for SF-39 in studies of patients with advanced HIV disease. SF-12 reduces item redundancy and the burden of data requirements for both investigators and patients; consequently, it may improve compliance with form completion.