Cost-threshold analyses of the National AIDS Demonstration Research HIV prevention interventions

Objective: The goal of the multisite National AIDS Demonstration Research ( NADR) program was to reduce the sexual and drug injection-related HIV risks of out-of-treatment injection drug users and their sex partners. Previous analyses have established that the NADR interventions were effective at cha nging participants' risky behaviors. This study was to determine whether th e NADR program also was cost-effective. Methods: Data from eight NADR study sites were included in the analysis. A mathematical model was used to translate reported sexual and injection-rela ted behavior changes into an estimate of the number of infections prevented by the NADR interventions and then to calculate the corresponding savings in averted HIV/AIDS medical care costs and quality-adjusted years of life, assuming United Stares values for these parameters. Because cost data were not collected in the original NADR evaluation, the savings in averted medic al care costs were compared with the cost of implementing a similar interve ntion program for injection drug users. Results: The eight NADR interventions prevented approximately 129 infection s among 6629 participants and their partners. Overall, the NADR program wou ld be cost saving (i.e. provide net economic savings) ii it cost less than US$2107 per person and would be cast-effective ii it cost less than US$10 2 64 per person. Both of these estimates are considerably larger than the US$ 273 per person cost of the comparison intervention. There was substantial c ross-site variability. Conclusions: The results of this analysis strongly suggest that the NADR in terventions were cost-saving overall and were; at the very least, cost-effe ctive at all eight sites. In the United States and other developed counties , investments in HIV-prevention interventions such as these have the potent ial to save substantial economic resources by averting HIV-related medical care expenses among injection drug users. (C) 2000 Lippincott Williams & Wi lkins.