Comparison Of Respondent-driven Sampling And A Venue-based Sampling Method Among Female Sex Workers

Backgroud:Globally, HIV/STIs epidemic is a very important public health problem. Many hidden populations with high risk behaviors, such as female sex workers (FSWs), men who have sex with men (MSM) and injecting drug users (IDU), have been disproportionately affected by the HIV/STIs epidemics in many parts of the world including China. Accurate and precise data on the behaviors in these at-risk populations is critical for understanding and tracking the direction of the epidemic, planning and evaluating effective intervention responses. Methods commonly used for sampling these populations include venue-based sampling and respondent-driven sampling (RDS). RDS was modified from chain-referral sampling and first introduced by Heckthorn in1994. It has been increasingly applied in recent years. But few empirical research have evaluated RDS in concurrent comparison with other methods.Objectives:To evaluate RDS through concurrent implementation and comparison of two sampling method, RDS and a venue-based sampling method, among FSWs. And to provide information for local HIV/STIs control efforts.Method:From October2009through January2010, two cross-sectional studies using RDS and a venue-based sampling method called "PLACE"(Priorities for Local AIDS Control Efforts, PLACE) respectively were simultaneously conducted among FSWs in Liuzhou city, Guangxi Zhuang Autonomous Region. RDS sampling process started with seven seeds and participants were referred by peers through FSWs’ social networks. For PLACE, a venue sampling frame was constructed through interviews with community and venue informants. A subset of venues was sampled from the sampling frame and participants were recruited at selected venues. Information on socio-demographic characteristics, sexual behaviors, drug use behaviors and sex work were collected through face-to-face interviews. Syphilis rapid test was employed to test Treponemal pallidum specific antibodies using finger-prick whole blood specimens. RDS data was analyzed using Respondent-driven sampling analysis tool to produce adjusted population propotions and confidence intervals. PLACE data was weighted incorporating sampling strategies and participation rate at venue and individual level to generate population estimates. Overlaps in confidence intervals around each of the two samples’population estimates were compared to examine significant differences at a significance level of5%.Results:A total of583and161eligible FSWs were recuited in RDS and PLACE respectively. The two methods found significant different population characteristics. RDS only recruited2.5%(95%CI1.4,3.8) FSWs who resided in rural counties, while49.7%(95%CI49.3,50.1) PLACE FSWs resided in rural counties. RDS found a group of FSWs who only solicited though telephone in the past six months and a greater proportion of FSWs who solicited at public outdoor venues. RDS estimated that telephone-based FSWs accounted for7.6%(95%CI3.3,12.5) of the total FSWs population. In contrast, PLACE did not find any FSWs who only solicited through telephone. Compared with FSWs based in establishments, FSWs who only solicited through telephone or at public outdoor venues had higher rate of a positive syphilis rapid test results, relatively lower rate of condom usage, and less exposure to HIV prevention activites at sex work venues.79.4%(95%CI68.8,90.1) PLACE FSWs had experienced HIV prevention activities at sex work venues, which was considerably higher than that among RDS FSWs (48.2%,95%CI42.7,53.2). The overall prevalence rate of a positive syphilis rapid test was24.0%(95%CI13.2,34.8) among PLACE FSWs, which was significantly higher than that among RDS FSWs (8.6%,95%CI5.3,12.8). After excluding PLACE FSWs recruited at rural counties, there was no significant difference in the overall prevalence rate of a positive syphilis rapid test between RDS FSWs and PLACE urban FSWs. The prevalence rate of a positive syphilis rapid test was the highest among rural FSWs (34.4%,95%CI17.0,51.8). Conclusions:In the absence of a gold standard, it is difficult to judge the representativeness of the samples produced by different methods. In the surveillance and investigation of hidden populations, it is especially important to reach all segments of the target population. In urban Liuzhou, RDS found a more hidden group of FSWs who only solicited through telephone which was not found by PLACE and more FSWs who were in less contact with HIV prevention activites and services. Thus, we recommend that in areas where venue-based method is the routine practice for surveillance and intervention activities, efforts should be taken to learn the dynamics of local sex work industry. If a substantial proportion of FSWs only solicit clients through more hidden ways such as telephone or internet, findings from a hybrid method combining venue-based method and RDS could collect more accurate data for surveillance and provide more information upon which to make program and policy decisions. However, more research is needed to test the applicability of RDS in the context of rural sex work industry. More serious epidemic of syphilis in rural counties warrants more enhanced interventions. FSWs who solicited through telephone and public outdoor venues are even less accessible and at higher risk of HIV/STI. Focued interventions are urgently needed.