Abstract
Background: Men who have sex with men (MSM) are at high risk of HIV infection in all regions globally. Surveys among MSM in Hanoi and Ho Chi Minh City in Vietnam have shown prevalence rates of 15% and 17% respectively. Representative data about MSM and other hidden populations are extremely difficult to collect, due to stigma. Much available data may thus not be representative of the MSM population as a whole, preventing an evidence-based response to the HIV epidemic. Respondent-driven sampling (RDS) was developed to enable representative sampling of hidden groups but relies on untested assumptions. In addition, an Internet-based version of RDS (webRDS) can potentially improve several key limitations of the RDS method.
Aims: To develop a webRDS system for representative sampling of internet-using MSM in Vietnam and to increase knowledge about sexual risk behavior for HIV among MSM in Vietnam.
Methods: RDS studies were simulated on anonymized data from a large webcommunity for lesbian, gay, bisexual and transgender persons. The assumptions under which RDS is known to be unbiased were systematically violated and the effects were studied. In-depth interviews (IDIs) were performed with 17 purposively sampled MSM in Hanoi, Vietnam and latent content analysis was used. A webRDS system, designed for MSM in Vietnam, was developed and used in two crosssectional surveys in Vietnam.
Results: Simulations indicate that bias using RDS is large if respondents choose to invite people based on characteristics that are correlated with study outcomes. Bias and variance increased if participants preferentially invited frequent contacts. Diverse seed selection was highly beneficial. IDIs showed that male-male relationships were perceived to be short and unstable. Faithfulness was highly valued but largely seen as unobtainable. The webRDS system enabled sustained recruitment of 676 and 983 submissions respectively in the two surveys. Maximum wave length was 29 waves and equilibrium was reached for all but one variable. The sample was younger and of higher education than the Vietnamese average. Thirty-six percent (CI: 32-42%) of MSM in the second webRDS survey had unprotected anal sex (UAS) at least once during the last three months. Thirty-six percent (CI: 32-41%) had ever taken an HIV test and received the result. Ninety-two percent of all UAS acts were reported to have taken place with non-casual partners. The numbers of UAS partners and UAS acts were positively correlated. Modeling showed that this sexual behavior pattern is likely to lead to explosive HIV transmission.
Conclusions: Simulations showed that RDS is relatively robust but empirical studies of non-random recruitment and a better understanding of real-life variance is needed. WebRDS is a highly promising method. Further evaluation and development is warranted. HIV prevention services niched for MSM needs to be aggressively scaled up in Vietnam.
