Li, Yan MD, PhD*†; Detels, Roger MD, MS†; Lin, Peng MD, MPH*; Fu, Xiaobing MD, MS*; Deng, Zhongming MD‡; Liu, Yongying MD*; Tan, Yihe MD§; Li, Jie MD*; Wu, Zunyou MD, PhD‖
China is experiencing a rapidly expanding HIV epidemic. The most common modes of HIV transmissions are through heterosexual contact and sharing contaminated needles among injection drug users (IDUs). Heterosexual transmission has been rising in recent years and has become the major driver of the HIV epidemic in China. According to the estimation conducted jointly by Ministry of Health of China, UNAIDS, and World Health Organization in 2007, there were about 700,000 people living with HIV/AIDS at the end of 2007, among whom heterosexual transmission accounted for 40.6%.1 The results from HIV sentinel surveillance also showed that HIV prevalence among the sex workers has increased in recent years.2 Guangdong Province is experiencing the same trend. The proportion of heterosexual transmission among the reported HIV/AIDS cases in Guangdong Province was 4.89% in 2001, 6.53% in 2002, 26.9% in 2006 and 37.9% from January to October 2007.3,4
The epidemic of STDs has expanded concurrently with the resurgence of the sex industry in China since the late 1970s. Results from the national STD surveillance system reveal that although the incidence of STDs was 12.32 per 100,000 population in 1989, by 1998, it had increased 4-fold to 50.68 per 100,000 population. The average annual increase in incidence for all STDs during the period from 1989 to 1998 was about 17.3%.5
Effective intervention strategies are urgently needed to control the transmission of HIV from IDUs to female sex workers (FSWs) and then to their clients. However, due to the illegal status of sex workers in China, little is known about the HIV/AIDS/STD epidemic situation and the related risk behaviors among FSWs. Surveillance for HIV and STDs and studies done among FSWs are mainly focused on the FSWs in the reeducation centers. Some research programs tried to recruit FSWs in the community using convenience sampling or chain referral sampling.6-8 A representative sample from the community was needed to understand the risk behaviors and HIV/AIDS and STD status and to better evaluate the potential roles of FSWs in the spread of HIV/AIDS.
Respondent-driven sampling (RDS) is a new method of chain-referral sampling for obtaining a more representative sample from a hidden population. It has been applied in many research programs to sample hard-to-reach and hidden populations such as IDUs and men who have sex with men9-13 Those studies demonstrated that RDS was a feasible strategy for surveying hard-to-reach populations. However, the RDS method has not been widely applied for recruiting FSWs. The Vietnamese Ministry of Health conducted a survey using RDS to assess HIV prevalence, risk factors, and service utilization among FSWs in 2004. The program demonstrated that RDS could reach the less visible FSWs and, thus, may provide better estimates for the population.14
The present study was conducted to estimate HIV and STD prevalence and to identify predictors of infection among FSWs in Guangdong Province, China. RDS was used to obtain a probabilistic sample and to get a less biased estimate for the population.
Participants and Procedures
The source population was the FSWs who lived or worked in the study county. FSWs were defined as those who had exchanged sex for money or goods in the past 6 months. It excluded those who exchanged sex for gifts or some other material rewards from their boyfriends or sweethearts. It also excluded those who were younger than 18 years.
RDS was used to recruit subjects into the study. Four seeds were recruited through local key informants or pimps to participate in the study. After the seeds completed the interview, each of them was given 3 coupons for recruiting their peers into the study. All new recruits were offered a dual incentive: one for completing the interview and one for successfully recruiting up to 3 peers into the study.
An office in the local Dermatology and STD Hospital was set up for the interviews. When the seeds or the participants with the referral coupon came to the office, screening was conducted to confirm their eligibility. An oral consent was obtained before the interview. The eligible participants underwent a face-to-face interview with a standardized questionnaire. The questionnaire was divided into 2 parts. The first part included questions dealing with knowledge and attitudes about HIV/AIDS and some basic information. The second part was about sexual-related and/or drug use-related behaviors. For the second part, the participants had the option to be interviewed face-to-face by the interviewer or to use a CD player to listen to prerecorded questions and put their answers on an answer sheet, which contained neither the questions nor identifying information.15 The interviews were anonymous, no identifying information was collected.
After each interview, a free gynecological examination was conducted by a female gynecologist at the outpatient clinic of the hospital. A cervical swab was collected for testing for gonorrhea and Chlamydia. A 5-mL blood sample was also collected for testing for HIV and syphilis.
Blood samples were collected for testing for HIV and syphilis. Serum samples were screened for antibodies to Treponema pallidum by the rapid plasma reagin test (Rongsheng Biotechnical Company, Shanghai, China). Positive specimens were confirmed by T. pallidum hemagglutination. Syphilis infection was defined as being positive on both rapid plasma reagin and T. pallidum hemagglutination tests. HIV antibody testing was done by enzyme-linked immunosorbent assay (Gibiai Biotechnical Company, Beijing, China). If a test was positive, 2 further assays (the original assay plus a different assay) were conducted in parallel. If both were negative, the result was recorded as HIV negative. If both were positive or there was discordance in the results, the specimen was sent to the confirmatory laboratory for Western blot testing.
Cervical swabs were collected for testing for gonorrhea and Chlamydia. The specimen was rolled on the culture plate and incubated in a CO2 jar at 37°C for 24 hours. Thayer-Martin selective culture media was used for culturing Neisseria gonorrhea. For confirmatory tests, oxidase and sugar fermentation tests were done. Typical colonies of oxidase-positive gram-negative diplococci on Thayer-Martin culture media with a positive glucose fermentation test were used as the criteria for the identification of N. gonorrhea. Chlamydia was diagnosed by the enzyme immunoassay tests.
All the tests were conducted in the laboratory of the Zhaoqing Dermatology and STD Hospital. The blood samples were also sent to the HIV Confirmatory Center in Guangdong Centers for Disease Control and Prevention for HIV antibody tests (enzyme-linked immunosorbent assay; Organon Teknika, Netherlands).
The data were double entered, using Epidata, by 2 researchers independently and were checked to assure accuracy and completeness.
Inconsistent condom use was defined as not using condoms 100% of the time for all sex acts during the last week, including any type of sex (oral, vaginal, or anal) with any kind of partner (clients, regular nonpaying, or nonregular nonpaying partners). STD infection was defined as any positive test for gonorrhea, Chlamydia, and/or syphilis.
The prevalence of STD and the potential predictors were described by measuring the frequency distribution. The frequency distribution was measured as the crude population proportion and the estimated population proportion (EPP) using RDS Analysis Tool, which was designed for analyzing RDS data adjusting for the long-chain referral recruitment design based on the linkage of the recruiters and recruitees and the participants' network sizes to produce population estimates.
Multivariate logistic regression was used to examine the associations of independent variables and the outcome (STD infection), simultaneously adjusting for potential confounders using statistical analysis system (SAS). The individualized weights generated for the dependant variable using RDS Analysis Tool were used to weight the data set for multivariate analysis.16 Variables were selected into the multivariate model based on the prior knowledge of the relationship between them and inconsistent condom use. If there was no prior knowledge, variables were selected depending on how much their presence/absence affected the confidence intervals of other variables in the model. The cut-off level of change of the confidence intervals of these variables of interest was predefined to be 10%.
Introduction of Recruitment Process (RDS)
A total of 320 participants, including the 4 seeds, were recruited for the study in 5 months. Each of the 4 seeds from street-based sex workers, massage parlor-based sex workers, hotel-based sex workers, and hair/beauty salon-based sex workers recruited 65, 209, 3, and 39 participants into the study, respectively. The longest recruitment tree had 16 recruitment waves. In total, 864 recruitment coupons were distributed, among which 316 (36.6%) were returned and 548 (63.4%) were not. Overall, the equilibriums with respect to the social and demographic characteristics were obtained after 2-10 waves.
Characteristics of Sample and EPP
Table 1 presents the characteristics of the FSWs. The median age of the 320 participants was 26 years. Most of the participants were Han (China's majority ethnic group) and had a junior middle school education (9 years of education). Over half of the participants (59.2%) had been married. Most of the participants were from Guizhou, Hunan, Sichuan, and Guangxi Provinces.
The FSWs were classified into 4 types according to the places where they are recruiting their clients or their working places: (1) massage parlor, karaoke bar, and drinking bar-based FSWs; (2) resort center and hotel-based FSWs; (3) hair/beauty salon-based FSWs; and (4) street, park, mall, construction site, rented apartment, and restaurant-based FSWs. FSWs from the above 4 places accounted for 25.3%, 22.6%, 25.6%, and 26.4%, respectively, of the total.
Age at first sex experience ranged from 10 to 27 years, the median age was 19 years (Table 2). The age for initiating commercial sex ranged from 16 to 47 years, and the median age was 23 years.
The charge for the last commercial sex ranged from 10 ($1.25) to 2000 RMB ($250) Chinese Yuan. The median was 150 Yuan (about $19). The number of paying clients during the last week ranged from 0 to 30 with a median of 5. The EPP who had regular nonpaying partners and nonregular nonpaying partners during the last week was 46.6% (40.1-53.2) and 11.0% (6.9-16.1), respectively.
Condom Use With Sex Partners
Table 3 presents condom use with paying clients and nonpaying partners among the FSWs. 93.5% (90.7% to 96.5%) of the participants used a condom the last time they had sex with a client although 6.5% (3.6% to 9.5%) of them did not. The proportion of consistent condom use with clients during the last week was 58.0% (50.4% to 65.5%).
Incorrect condom use including using condoms just before ejaculation, rupture or breaking of the condoms, or not using the condoms during the entire sex act was not unusual among the FSWs. The proportion using condoms incorrectly during the previous month was 39.8% (33.0% to 46.6%).
70.9% of the participants used a condom the last time they had sex with a nonregular nonpaying partner. The proportion of consistent condom use with nonregular nonpaying partners during the last week was 43.1% (19.8% to 64.4%).
Only 38.7% of the participants used a condom the last time they had sex with a regular nonpaying partner. The proportion of consistent condom use with regular nonpaying partners during the last week was 20.5% (13.6% to 28.7%).
None of the participants used heroin, cocaine, or valium. 9.75% (31 of 318) had used the drug “ecstasy”. Six participants had used drugs in the past 6 months. 1 of 28 (3.57%) of the participants had exchanged sex for drugs.
HIV and Sexually Transmitted Infection Prevalence
None of the participants was HIV positive. The adjusted prevalence of syphilis, gonorrhea, and Chlamyida was 8.0% (4.6% to 12.2%), 9.5% (5.6% to 14.3%), and 3.9% (1.7% to 6.2%), respectively. The adjusted prevalence of any STD was 19.7% (13.9% to 26.2%).
Predictors for STD Infection
The multivariate logistic regression model indicated that years of education and self-perception of the risk of HIV infection were protective factors independently associated with STD infection. Awareness of HIV/AIDS, regular sex partners deciding condom use when having sex, recruiting clients at hotels, hair/beauty salons, or streets instead of massage parlors, and higher charge for commercial sex last time were risk factors independently associated with STD infection (Table 4). Other factors such as the number of sex partners, condom use, and drug use were not found to be associated with STD infection.
This study found that none of the participants were HIV positive, suggesting that the HIV epidemic in Guangdong Province is still concentrated among IDUs and had not expanded to general FSWs in the study county. The expansion of the HIV epidemic from IDUs to the FSWs relies on the overlap of the 2 populations; that is, the FSWs who are also IDUs or who have drug using clients. Tran et al17 reported a prevalence of HIV of 12% among FSWs in Hanoi. The HIV prevalence among IDU-FSWs was 33%, 3.9% among noninjecting drug user-FSWs, and 1.6% among nondrug-using FSWs. Thus, sharing injecting equipment was the primary cause of the rapid increase of HIV infection among the FSW population in Hanoi. In his study population, there was a high proportion of drug use (that 38% used drugs, up to 83% of whom injected drugs). In our study, none of the participants were found to be IDUs, and only a few reported drug use by their sex partners. This may be the main reason for the low level of HIV observed among this population.
HIV sentinel surveillance among FSWs was conducted in the reeducation centers. It has been reported that incarcerated FSWs had a higher risk of HIV infection. In our study, we found, in the field assessment, that the FSWs who used drugs were more likely to be arrested and sent to a reeducation center. Only 2 of our participants had been arrested, and none of them had been sent to reeducation centers. The HIV prevalence in the sentinel surveillance which was conducted in the reeducation centers was higher than that in the comprehensive behavioral surveillance which was conducted in the community. The comprehensive behavioral surveillance in Guangdong Province found no HIV-positive cases in 2005 and 2007, whereas the HIV prevalence observed in the sentinel surveillance in 2005 and 2007 was 0.27% and 0.21%, respectively.3
The study found the prevalence of syphilis, gonorrhea, and Chlamydia to be 8.0% (4.6% to 12.2%), 9.5% (5.6% to 14.3%), and 3.9% (1.7% to 6.2%), respectively. The prevalence of syphilis and gonorrhea was much higher than among female market vendors (1.31% and 1.36%, respectively) in Fuzhou, China, as reported by Wu et al.18 However, Wu et al18 reported a higher prevalence of Chlamydia (11.2%) among the female market vendors in Fuzhou. Although the HIV prevalence among the FSWs in our study was 0%, the prevalence of STDs in our study population was relatively high, which is an indicator of high frequency of risk behaviors. Condom use was low among this population. Though the proportion of condom use for last sex with clients was high (93.5%), the proportion of consistent condom use with clients during the last week was only 58.0%. Condom use with nonpaying partners was even lower.
The study indicated that years of education and self-perception of the risk of HIV infection were protective factors independently associated with STD infection. Awareness of HIV/AIDS; regular sex partners deciding condom use; recruiting clients at hotels, hair/beauty salons, or streets instead of massage parlors; and higher charge for commercial sex last time were risk factors independently associated with STD infection.
Baltazar et al19 reported that illiteracy was associated with acquiring a sexually transmitted infection and/or reproductive system infection among FSWs. Tran et al17 reported that being a low social economic FSW, having higher income compared with peers, and perception of self being at low risk for HIV were associated with HIV infection among noninstitutionalized FSWs. These reported risk factors for HIV were quite similar to those we found for STD infections, which suggested that HIV infection and STD infection share some common risk factors.
Regular nonpaying partners deciding condom use was independently associated with having an STD infection after controlling for confounders in our study. This suggested that lack of power in condom use decision making was a risk factor for STD infection, which suggested that future interventions not only need to target FSWs but also need to target their clients.
RDS was used to recruit FSWs in this study. It is a new method of chain-referral sampling utilizing an incentive for being interviewed and another incentive for recruiting peers to be interviewed. Heckathorn20 asserted “RDS produces samples that are independent of the initial subjects from which sampling begins; and RDS reduces the biases resulting from voluntarism and masking and provides means for controlling the biases resulting from differences in the sizes of personal networks.”20 An underlying assumption of RDS is that the long-chain recruitment represents a first-order Markov process that reaches a dynamic equilibrium between the tendencies of persons with similar characteristics to associate with each other, relative network sizes, and underlying makeup of the target population. By tracking the linkage of the recruiters and recruitees and by recording the relative network sizes of participants, the biases inherent in the referral recruitment can be quantified and adjusted for in the analysis. Data for the adjusted analyses are sufficient when the sample reaches “equilibrium”. In this program, the overall equilibriums with respect to the social and demographic characteristics were obtained after 2-10 waves. This study is the first using RDS to recruit subjects among FSWs in China. It showed that RDS was feasible among FSWs in China and provided a probabilistic sample for understanding the risk behaviors and HIV/STDs among FSWs. However, the number of the participants was very different from the 4 seeds ranging from 3 to 209. The results are dominated by the disproportionately large seed recruiting 209 participants. Another limitation in this study was that only 36.6% of the potential subjects returned with the distributed coupons. The subjects who did not return with the coupons may have different risk behaviors.
This study found a high prevalence of STD in spite of finding no HIV-positive in FSWs. It also revealed a high proportion of inconsistent condom use especially when having sex with nonpaying partners. This indicates the high potential for HIV transmission if the epidemic spread to FSWs and their clients. We found little overlap of FSWs and IDUs. This indicates that now is an excellent time to introduce intervention programs to reduce HIV transmission from IDUs to STDs general population through FSWs. Therefore, effective intervention programs for FSWs in this area should be implemented immediately.
The authors thank Feng Deng, Ruiheng Xu, and Yurun Zhang for their valuable support. We are grateful to the staff of the Institute of HIV/AIDS Control and Prevention, Guangdong Provincial Center for Disease Control and Prevention, the staff of the Zhaoqing Dermatology and STD Hospital and Center for Disease Control and Prevention of Zhaoqing prefecture for their contributions to the conduct of the study and to Wendy Aft for assisting in the preparation of this article.
1. Ministry of Health, People's Republic of China, Joint United Nations Programme on HIV/AIDS, World Health Organization. 2007 Update on the HIV/AIDS Epidemic in China
. December 2008.
2. Ministry of Health China, UNAIDS and WHO. Update on HIV/AIDS Epidemic in China
. November 2006.
3. Center for Disease Control and Prevention of Guangdong province. HIV/AIDS Epidemiology Annual Report in 2006 in Guangdong Province, China
. December 2007.
4. Department of Health, Guangdong. Update on HIV/AIDS Epidemic in Guangdong
. November 2007.
5. Zhang D, Lv F, Wang L, et al. Estimating the population of FSWs in two Chinese cities on the basis of the HIV/AIDS behavioural surveillance approach combined with a multiplier method. Sex Transm Infect
6. Chen X, Yin Y, Liang G, et al. Sexually transmitted infections among FSWs in Yunnan, China. AIDS Patient Care STDs
7. Van HA, Fu Y, Dukers N, et al. High prevalence of syphilis and other STD among sex workers in China: potential of fast spread of HIV. AIDS
8. Ruan Y, Cao X, Qian H, et al. Syphilis among FSWs in Southwestern China: potential for HIV transmission. Sex Transm Dis
9. Stormer A, Tun W, Guli L, et al. An analysis of respondent driven sampling with injection drug users (IDU) in Albania and the Russian Federation. J Urban Health
10. McKnight C, Jarlais DD, Bramson H, et al. Respondent-driven sampling in a study of drug users in New York City: notes from the field. J Urban Health
11. Ma X, Zhang Q, He X, et al. Trends in prevalence of HIV, syphilis, hepatitis C, hepatitis B, and sexual risk behavior among men who have sex with men. J Acquir Immune Defic Syndr
12. Millett GA, Ding H, Lauby J, et al. Circumcision status and HIV infection among black and Latino men who have sex with men in 3 US cities. J Acquir Immune Defic Syndr
13. Ma X, Zhang Q, He X, et al. Trends in prevalence of HIV, syphilis, hepatitis C, hepatitis B, and sexual risk behavior among men who have sex with men: results of 3 consecutive respondent-driven sampling surveys in Beijing, 2004-2006. J Acquir Immune Defic Syndr
14. Johnston LG, Sabin K, Hien MT, et al. Assessment of respondent driven sampling for recruiting FSWs in two Vietnamese cities: reaching the unseen sex worker. J Urban Health
15. Liu HJ, Detels R. An approach to improve validity of response in a sexual behavior study in a rural area of China. AIDS Behav
16. RDS Incorporated. RDS Analysis Tool. v5.6. User Manual
; December 2006.
17. Tran TN, Detels R, Long HT, et al. HIV infection and risk characteristics among Female Sex Workers in Hanoi, Vietnam. J Acquir Immune Defic Syndr
18. Wu Z, Rotheram-Borus MJ, Li L, et al. Sexually transmitted diseases and risk behaviors among market vendors in China. Sex Transm Dis
19. Baltazar RMC, Rivera RL, Cruz VA, et al. Prevalence of sexually transmitted infections and associated risk factors among female commercial sex workers in Cuautla, Morelos. Ginecol Obstet Mex
20. Heckathorn DD. Respondent driven sampling: a new approach to the study of hidden populations. Soc Probl
© 2010 Lippincott Williams & Wilkins, Inc.