Background In 2009, the CDC funded the Transgender HIV Behavioural Surveillance pilot to assess the feasibility of using respondent driven sampling (RDS) to recruit ethnic minority male-to-female transgender (MTFTG) participants. RDS is a chain-referral strategy that restricts the number of people each participant can recruit and offers incentives for both participation and recruitment. With proper adjustment for sampling, data collected through RDS can be used to develop population prevalence estimates. RDS assumes that respondents know one another as members of the target population.

Methods Because of funding constraints, formative work and data collection lasted only 4 months. Our goal was to recruit 100 MTFTG participants, at least 15 years old, who were assigned male sex at birth and who currently identified or presented as female. We established a field office in the Houston Transgender Center and identified a charismatic MTFTG interviewer. Office hours were set for early afternoon; incentives were $20 for the interview and $10 for each recruit (up to $50). Recruiter instructions were to give the RDS coupon to known Latina or Black MTFTG.

Results Our University prohibited the hiring of our first-choice interviewer because she had a previous moral turpitude conviction. Recruitment was delayed until we hired another. We planted 11 seeds; 5 failed, 3 recruited 7 individuals in 2 waves; and 3 recruited 40 individuals with the longest chain consisting of 6 waves. We distributed 201 coupons and interviewed 48 eligible and consenting MTFTG.

Conclusions Formative work is essential to identify the size and features of the study population social networks, time and place for recruitment, and appropriate incentive. RDS requires respondents to be linked by the target attribute, in this case ethnic minority MTFTG. By restricting recruitment to transgender individuals of colour, we violated that assumption. Our sample networks were not structured around race/ethnicity; rather they were networked through work affiliations (sex worker, performer, unemployed). Recruitment also failed because the incentive was not enough to encourage performers and sex workers to participate in a health survey. Evening hours would have been better for our group's work schedules so they could participate on their way out for the evening. Our failure to get RDS initiated reinforces the importance of formative work to determine if RDS will succeed in your target population.