The Committee reviewed peer-reviewed publications reporting evaluations of HIV prevention programs. It has proven challenging to interpret such studies in terms of the relative reduction in the rate of new HIV infections implied by the programs examined. Yet, it is precisely this information about new infections that is required to sensibly consider the overall impact of alternative plans for distributing HIV prevention dollars.
To develop data for this analysis, we reviewed in detail a subset of HIV prevention studies with an eye toward estimating the relative reduction in exposure to HIV risk as a result of prevention interventions aimed at men who have sex with men, injection drug users, and women at high risk for heterosexual transmission (see references at the end of this appendix for a listing of the studies reviewed). These three risk groups were selected to coincide with the risk group classification employed by Holmberg in estimating HIV incidence (Holmberg, 1996). Prevention efficacy, defined as the percentage reduction in new HIV infections, was estimated under the assumption that HIV incidence is proportional to the product of HIV prevalence and risky exposure rates (or “contact rates,” as defined in the mathematical epidemiology literature; see Anderson and May, 1991, and Kahn, 1996, for an example). There is little question that more work along these lines is needed in order to better understand the value of HIV prevention interventions.
The Committee’s review of these studies also revealed that there were the high rates of subject attrition from the interventions evaluated. This observation led us to question the ability of many prevention programs to effectively reach and retain program participants over time. An intervention might be effective for program participants, but unable to reach more than, say, 10 percent of the total population. While many discussions of HIV prevention assume, in fact, that all persons at risk can be contacted, the Committee felt it was important to build into the analysis the recognition that not all of those at risk can be reached by interventions. Because no data exist on program reach, the Committee made some assumptions in the analysis to account for this fact. We assumed: 50 percent in the base case; 75 percent in the optimistic case; and 25 percent in the pessimistic case.
The Committee notes that very few evaluations of HIV prevention programs report the costs of providing such services, though there are