When is an HIV infection prevented and when is it merely delayed?

HIV prevention programs are typically evaluated using behavioral outcomes. Mathematical models of HIV transmission can be used to translate these behavioral outcomes into estimates of the number of HIV infections averted. Usually, intervention effectiveness is evaluated over a brief assessment period and an infection is considered to be prevented if it does not occur during this period. This approach may overestimate intervention effectiveness if participants continue to engage in risk behaviors. Conversely, this strategy underestimates the true impact of interventions by assuming that behavioral changes persist only until the end of the intervention assessment period. In this article, the authors (a) suggest a simple framework for distinguishing between HIV infections that are truly prevented and those that are merely delayed, (b) illustrate how these outcomes can be estimated, (c) discuss strategies for extrapolating intervention effects beyond the assessment period, and (d) highlight the implications of these findings for HIV prevention decision making.