Objective: To understand the impact of differences in participation rates between infected and uninfected individuals on estimates of the cost-effectiveness of HIV screening.
Methods: Costs per infection detected are modeled as function of both prevalence and serostatus-dependent testing rates. Data from national surveillance surveys, seroprevalence studies, and other sources are employed to suggest the magnitude of results.
Results: Differential participation produces a near-doubling in the estimated cost per infection identified. This result is sensitive to assumptions regarding the benefits of screening for seronegatives.
Conclusions: Voluntary HIV screening programs may incur prohibitive costs by over-recruiting people at little risk of infection. Failure to account for differential participation can result in over-optimistic cost-effectiveness estimates. However, the relevance of this result--and the significance of both prevalence and participation as cost drivers--is overwhelmed by what is assumed about the benefits conferred to uninfected people by HIV screening.