The use of Respondent-Driven Sampling to Recruit Black Populations at Risk for HIV Infection for a Behavioral Survey in Tanzania and North Carolina, US: Effectiveness, Challenges and RecommendationsView Presentation
*Mario Chen, FHI360
Keywords: RDS, HIV, MARPS, Tanzania, Africa, North Carolina, Sampling
As part of site preparedness activities for further HIV prevention research, we conducted a cross-sectional study of women presumed to be at higher risk for HIV infection in Tabora, Tanzania using respondent-driven sampling (RDS). Women were eligible if they were between 18 and 49 years old and reported more than one sexual partner in the last month. Community mapping activities and preliminary formative research informed survey design issues (e.g., recruitment of seed participants for the survey, level of incentives). Starting with 25 seeds, but with only 10 productive seeds, a sample of 737 women was enrolled to answer questions about their HIV risk behaviors, contraceptive history and knowledge. Eligible women were recruited into the study in seven months. We present a RDS adjusted analysis of key characteristics of the sample, homophily estimates, and implementation issues. Homophily measures were low to moderate (maximum Hx = 0.42). The number of recruitment waves was adequate (mean = 10 and max = 19 per seed) surpassing the estimated number of waves needed to reach equilibrium for key sample characteristics. Our sample met all RDS requirements regarding network connectivity and equilibrium. We faced challenges in the estimation of network size, tracking recruitment links, and productivity of the initial seeds. However, our challenges were surmounted and could be better addressed in future applications with the recommendations we provide in this paper. Our recommendations address issues not normally discussed in the literature. We discuss the application of these recommendations in a study conducted among Black young adults in North Carolina with similar measurement objectives (N=513). In addition to the practical recommendations, we discuss RDS implementation issues in the U.S. and African settings.