Consequences of HIV prevention interventions and programs: spectrum, selection, and quality of outcome measures

The outcome measures employed in an HIV prevention intervention study should match the research and policy questions at hand. If the question is 'did the intervention work to prevent HIV infection?', then seroincidence data may be insufficient. However, if the question is 'why did the intervention work?', then more detailed behavioral data are necessary (and sometimes behavior change itself is the real goal of an intervention study). Given the wide range of questions asked by HIV prevention policy makers, funders and researchers, a spectrum of outcome measures is needed across HIV prevention intervention studies. These include measures of behavioral determinants, HIV-related risk behaviors, HIV incidence (and other biologic markers), morbidity, mortality, and cost-effectiveness factors (such as cost per quality-adjusted life year saved). In this paper, we review the range of outcome measures used and needed in these intervention studies. Particular attention is paid to the psychometric properties of self-reported behavior change measures of sexual behavior and substance use. Additional emphasis is placed on the role of cost-effectiveness measures in intervention studies. A general framework is proposed for conceptualizing the array of outcome measure possible for any given HIV prevention intervention study.