Montealegre, J.R., Johnston, L.G., Murrill, C., et al. AIDS and Behavior (September 2013), Vol. 17 No. 7, pp. 2313-2340.
In this review of respondent-driven sampling (RDS) recruitment for HIV biological and behavioral surveillance surveys (BBSS) among key populations in Latin America and Caribbean (LAC) countries, the authors described RDS approaches, challenges, and considerations for future studies using RDS. Respondent-driven sampling has commonly been used in HIV BBSS to access high-risk, hard-to-reach populations in LAC. However, many studies encountered operational challenges. The authors identified 87 studies (conducted between 2005 and 2011) in 15 countries, mainly in South America and primarily on men who have sex with men. All studies conducted formative research before applying RDS. The studies mainly identified social network size using face-to-face interviews, and nearly all studies that reported information about analysis, adjusted for biases in chain-referral sampling. All but three studies collected biological specimens, usually conducting non-invasive HIV screens, with few reported challenges. There were challenges regarding appropriate incentives for participants, but most studies did not alter the incentives during data collection. Most studies adequately defined eligibility criteria, but 11 percent lacked a geographic parameter. The authors recommended addressing challenges by conducting formative research to understand social networks and define operational issues; using strategies to estimate social network size, monitor recruitment, and account for sub-populations; and analyzing data to counter biases in chain-referral sampling. Lastly, the authors suggested using the same sampling methods in the same populations over time as a best practice for monitoring HIV trends or evaluating program effectiveness.