set priorities for allocating resources across various health promotion and disease prevention programs. Nonetheless, these estimates demonstrate that multiple models lead to the conclusion that needle exchanges reduce the number of new HIV infections at a cost per infection averted far below the $119,000 lifetime cost of treating an HIV-infected person.
In order for cost-effectiveness analysis to be a useful tool in guiding the setting of priorities for the use of scarce financial resource allocations for AIDS prevention, certain critical information is needed:
measures of program effectiveness on meaningful health outcome measures,
estimates of costs, as well as discounting and variable cost rates, and
a set of independent AIDS prevention programs from which to choose, each having an expected degree of effectiveness and an associated cost.
Once these elements are known, policy makers will be able to maximize program effectiveness by allocating financial resources to programs according to their respective cost-effectiveness ratio. Numerous uncertainties surround these elements, such as the parameter estimates that should be included in the models, the tenability of the underlying assumptions used to derive those estimates, the necessary level of precision, and whether collateral program effects should be included (e.g., on crime, sexually transmitted diseases, drug use). Although these issues must be carefully addressed in cost-effectiveness analyses of AIDS prevention programs, they are beyond the scope of this report. For a more detailed treatment of benefit-cost analysis of AIDS prevention programs, the interested reader is referred to a previous National Research Council report (Turner et al., 1989:471-499).
Moreover, Kaplan (1995, in press) in a recent paper provides various models for estimating the cost and benefits of HIV prevention programs and uses data from the New Haven needle exchange to illustrate how such models can assist policy makers in making resource allocation decisions. In particular, the proposed models do provide some guidance for addressing pragmatic issues, such as how a given needle exchange program, in a specific epidemiologic environment, can optimize its cost-effectiveness ratio (i.e., cost per infection averted), within a budget constraint. The author also depicts how these ratios are a function of critical program parameters (e.g., number of program participants, number of needles exchanged per participant, HIV incidence in population being served).