The 131st Annual Meeting (November 15-19, 2003) of APHA

Abstract #61295

Constantine, E. Frangakis, PhD¹, Ravi Varadhan, PhD¹, Steffanie A. Strathdee, PhD², Mahboobeh Safaeian, MPH², Ron Brookmeyer, PhD¹, and David Vlahov, PhD³. (1) Department of Biostatistics, School of Public Health, Johns Hopkins University, 615 N. Wolfe Street, Baltimore, MD 21205, (410)955-3505, cfrangak@jhsph.edu, (2) Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, 615 N. Wolfe Street, Baltimore, MD 21205, (3) Center for Urban Epidemiologic Studies, New York Academy of Medicine, 1216 Fifth Avenue, New York, NY 10029

Background: Evaluating the efficacy of needle exchange programs (NEPs) for preventing HIV transmission is difficult because attenders and non-attenders often differ with respect to HIV risk factors. Standard statistical adjustments (e.g., regression) are problematic, since they cannot account for unmeasured confounders.

Method: We evaluated the impact of the Baltimore NEP on HIV seroincidence in a cohort of injection drug users (IDUs) from 1994-1998, using a new statistical approach called “principal stratification.” The central idea is to use the distance that attenders lived from the NEP site to define the comparison groups, since the study only controlled the placement of NEP sites. Results from this method are more reliable than standard adjustments because the former does not require unmeasured confounders to be balanced between attenders and non-attenders.

Results: Of 1170 initially HIV-seronegative IDUs, 77% were male, 94% were African-American and the median age was 34. HIV incidence was 9.9 per 1000 person-years. Using the data on proximity of NEPs from IDUs, needle exchange was found to yield an 88% reduction in HIV seroincidence (95% confidence interval: 0.00-2.29).

Conclusion: Principal stratification provides an important means of evaluating NEPs and may also help in the planning of NEP locations. The magnitude of our findings, as well as the considerable uncertainty associated with the estimate, call for independent studies to evaluate NEPs in other cities using data on proximity to exchange sites.

Learning Objectives:

• Recognize difficulties involved in evaluating the efficacy of intervention in population-based studies.
• Identify problems in commonly used statistical methods.
• Evaluate intervention efficacy using a more rigorous approach.

Keywords: HIV/AIDS, Treatment Efficacy

Related Web page: biosun01.biostat.jhsph.edu/~cfrangak/papers/ps.html

Presenting author's disclosure statement: