Estimating the impact of sexually transmitted HIV prevention interventions: A comparison of the proportion of unprotected acts to a regression-based approach

The proportion of times that participants used condoms during the recall period is the measure most often used to assess the impact of an intervention designed to prevent the transmission of HIV through sexual contact. Here, it is argued that researchers, service providers, and policy makers should consider using a regression-based approach to control for differences in sexual activity and assess the impact of their programs. Unlike the proportion, a regression-based summary measure is defined when individuals report that they did not have sex during the recall period, and varies among individuals who report different levels of sexual activity. More importantly, a regression-based approach lends itself to studying relationships between the determinants of condom use and the determinants of sexual activity.