This application addresses PA-10-069. The development of methods to produce less-biased, more generalizable samples of men who have sex with men (MSM) is vital to HIV prevention science. The proposed study will assess the feasibility of improving representativeness of samples of MSM recruited online, a population at acute risk for sexually transmitted infections. While the existing network of Internet-using MSM (IUMSM) is wide ranging, densely connected and diverse, studies of this population have been limited to convenience and snowball samples that over-represent heavy Internet users. This project would employ respondent-driven sampling (RDS) a chain-referral method that starts by selecting of initial respondents (seeds) who recruit peers (other members of the target population) as additional respondents. Those new participants in turn recruit others, forming recruitment chains. RDS is an adaptive sampling technique designed to overcome a major limitation of generalizability, over-sampling population members with greater numbers of peer ties and hence greater likelihood of recruitment. It accomplishes this by weighting observations based on each participant's reported number of acquaintances in the target population that he himself could recruit. To date, RDS among MSM has been conducted using face-to-face interviews. While RDS has demonstrated effectiveness in recruiting MSM and allowing collection of behavioral data and diagnostic specimens for sexually transmitted infection (STI) testing, restricted chain lengths and low recruitment productivity have limited sample representativeness. By permitting participants to select peers for referral from online contacts, and refer those peers to the study online and to collect data via a website, on-line respondent driven sampling (ORS) would considerably ease both referral and participation relative to reliance on presentation at a physical site. ORS has the potential to produce longer referral-chains that penetrate more broadly across the population of interest, thus increasing sample representativeness. Recruitment and data collection can plausibly be accomplished entirely online, but collection of the biological specimens often obtained using RDS necessitates innovation. Therefore the proposed study will determine the feasibility of obtaining though postal delivery home-collected biological specimens for STI testing from a subset of the IUMSM recruited via ORS. The project will pilot ORS among MSM, a population characterized by a high degree of online interconnectedness, to assess the acceptability to the target population of (a) referring peers online, (b) collection of behavioral data from peer-referred participants via a website, (c) the feasibility of obtaining biological specimens. Information systems will feature stringent security measures put into place by Yale Information and Technology Services in accordance with the Yale Human Investigation Committee.
Improved sampling methods for IUMSM who use the Internet will serve three purposes: 1) better representativeness of IUMSM in epidemiological studies, 2) more generalizable inference in studies conducted among IUMSM, and 3) more informed behavioral data and potentially behavioral surveillance for prevention interventions targeting IUMSM. The proposed project will a) assess the ability of ORS to recruit IUMSM, b) determine the degree to which ORS meets criteria for representativeness and c) investigate the feasibility of obtaining biological specimens from ORS. It will lay the groundwork for a number of much-needed measures the prevalence of risk behavior and infection, and more valid and generalizable correlates of behaviors that can transmit STIs, among IUMSM.
|White, Edward; Goldbaum, Gary; Goodreau, Steven et al. (2010) Interpopulation variation in HIV testing promptness may introduce bias in HIV incidence estimates using the serologic testing algorithm for recent HIV seroconversion. Sex Transm Infect 86:254-7|