Respondent Driven Sampling: Optimizing peer recruitment by selecting seeds with recruitment potential, NYC (2006-2007)

Respondent Driven Sampling (RDS) has proven effective in unbiased recruitment of hidden populations. By providing incentives for each peer recruit, RDS eliminates biases that are common with other chain referral sampling techniques and produces a representative sample that is independent of the seed composition. Since the sample population distribution reaches equilibrium only when recruitment is productive and referral chains reach 4-6 waves, seeds should be selected to optimize recruiting potential. The purpose of this analysis is to identify characteristics associated with successful peer recruitment. Outreach Workers initially recruited 48 seeds from disadvantaged, ethnographically mapped neighborhoods aged 18-40 who 1) injected heroin/crack/cocaine ≤ 3 years; or 2) never injected and used non-injection heroin/crack/cocaine ≥1 year. Eligible participants were given three RDS coupons to recruit peers, received individual recruitment trainings, and were invited to attend group-facilitated RDS trainings (RDST). Characteristics associated with recruiting at least one eligible peer network were determined using chi-square tests and multiple logistic regression. In 2007, 228 participants were enrolled with a median age of 34 years; 71% were male, 55% Black, 37% Hispanic, 89% were non-injectors, and 68% attended RDST. In the multivariate analysis, lower income (OR=2.6; 95%CI:1.1-5.9), RDST attendance (OR=5.8; 95%CI:2.7-12.6), and ease of travel to the research site (OR=2.1; 95%CI:1.0-4.3) were associated with successful peer recruitment. These findings suggest that when sampling a highly selective drug-using population, use of group-facilitated trainings on peer recruitment strategies, and travel time and proximity to the research site should be considered when selecting seeds.