RESUMO
CONTEXT: In 2003, the Centers for Disease Control and Prevention launched the Advancing HIV Prevention project to implement new strategies for diagnosing human immunodeficiency virus (HIV) infections outside medical settings and prevent new infections by working with HIV-infected persons and their partners. OBJECTIVES: : To assess the cost and effectiveness of a social network strategy to identify new HIV diagnoses among minority populations. DESIGN, SETTINGS, AND PARTICIPANTS: Four community-based organizations (CBOs) in Boston, Philadelphia, and Washington, District of Columbia, implemented a social network strategy for HIV counseling and testing from October 2003 to December 2005. We used standardized cost collection forms to collect program costs attributable to staff time, travel, incentives, test kits, testing supplies, office space, equipment, and utilities. The CBOs used the networks of high-risk and HIV-infected persons (recruiters) who referred their partners and associates for HIV counseling and testing. We obtained HIV-testing outcomes from project databases. MAIN OUTCOME MEASURES: Number of HIV tests, number of new HIV-diagnoses notified, total program cost, cost per person tested, cost per person notified of new HIV diagnosis. RESULTS: Two CBOs, both based in Philadelphia, identified 25 and 17 recruiters on average annually and tested 136 and 330 network associates, respectively. Among those tested, 12 and 13 associates were notified of new HIV diagnoses (seropositivity: 9.8%, 4.4%). CBOs in Boston, Massachusetts, and Washington, District of Columbia, identified 26 and 24 recruiters per year on average and tested 228 and 123 network associates. Among those tested, 12 and 11 associates were notified of new HIV diagnoses (seropositivity: 5.1%, 8.7%). The cost per associate notified of a new HIV diagnosis was $11 578 and $12 135 in Philadelphia, and $16 437 and $16 101 in Boston, Massachusetts, and Washington, District of Columbia. CONCLUSIONS: The cost of notifying someone with a new HIV diagnosis using social networks varied across sites. Our analysis provides useful information for program planning and evaluation.