HIV prevalence and high-risk sexual behaviours among MSM repeat and first-time testers in China: implications for HIV prevention

Introduction Little is known about HIV testing, HIV infection and sexual behaviour among bathhouse patrons in China. This study aims to assess differences in HIV prevalence and high-risk sexual behaviours between repeat and first-time testers among men who have sex with men (MSM) attending bathhouse in Tianjin, China. Methods Between March 2011 and September 2012, a HIV voluntary counselling and testing station was established in a gay bathhouse, which provided HIV testing and conducted a survey among participants recruited through snowball sampling. Differences in demographic and high-risk sexual behaviours between repeat and first-time testers were assessed using the chi-square test. Univariate and multivariate logistic regression analyses were conducted to identify predictors for HIV infection. Results Of the 1642 respondents, 699 (42.6%) were repeat testers and 943 (57.4%) were first-time testers. Among repeat testers, a higher proportion were men aged 18 to 25, single, better educated, had a history of STIs and worked as male sex workers or “money boys” (MBs). Repeat testers were less likely to report having unprotected anal intercourse in the past six months. The overall HIV prevalence was 12.4% (203/1642). There was no difference in HIV prevalence between repeat (11.2%, 78/699) and first-time (13.3%, 125/943) testers. The HIV prevalence increased with age among first-time testers (χtrend2=9.816, p=0.002). First-time MB testers had the highest HIV prevalence of 34.5%. Conclusions MSM attending bathhouse had an alarmingly high HIV infection rate, particularly in MB. Targeted interventions are urgently needed especially focusing on older MSM and MBs.


Introduction
Sexual transmission has become the primary route of HIV transmission in China, and men who have sex with men (MSM) constitute almost one-third of all cases of new HIV infections [1]. The proportion of newly reported HIV cases attributable to homosexual transmission rose from 2.5% in 2006 to 13.7% in 2011 [1]. Among MSM, HIV prevalence is estimated to be between 2.5 and 7.4% and increasing at an annual rate of 1.1% [2,3].
Voluntary counselling and testing (VCT) is an important strategy for the prevention and control of HIV and AIDS by engaging clients in risk reduction. A recent model of the epidemic in China found that a four-fold increase in VCT would avert 42,000 cases of HIV and 11,000 HIV-related deaths over the next five years [4]. Since 2007, significant resources have been directed towards scaling up MSM testing services under China's National Plan for HIV/AIDS Prevention and Control [1]. In the period after the adoption of the plan, there were significant increases in both lifetime testing rates (from 24 to 47%) and rates of testing in the past 12 months (from 21 to 38%) among MSM [5]. In 2009, the lifetime rates of testing reached 72% in Beijing among younger MSM, suggesting that repeat testing is increasing among this group [6].
Previous research has found that repeat testers and first-time testers differ in several ways. Among MSM, repeat testers are often younger, better educated and unmarried compared to first-time testers [7Á9]. However, repeat testing is associated with several risk factors for HIV: a higher number of sexual partners [10], having ever had an STI [11], unprotected anal intercourse (UAI) [12,13] and working as a male sex worker [14]. Additionally, in a study among MSM in seven US cities, over 75% of repeat testers who seroconverted did so within a year after their last test [13]. Continued engagement in risk behaviour by this specific group of repeat testers may undermine the effectiveness of VCT among MSM [15]. There has not yet been a study examining differences in HIV risk between MSM repeat and first-time testers in China, despite the distinct demographic and risk behaviour characteristics of these two groups [6Á9,14]. Therefore, in the context of increasing rates of testing, it is necessary to understand whether repeat testing may reduce the risk of HIV among MSM. The objective of this study was to assess the prevalence of sexual risk behaviours and HIV infection of repeat and first-time MSM testers attending a bathhouse VCT station.

Study participants and setting
This study was conducted in the city of Tianjin, which is a major Chinese city located 137 km east of Beijing. The overall HIV prevalence among MSM in Tianjin was 6.4% in 2008 [16].
In March 2011, an HIV VCT station was established in a gay bathhouse by the National Centre for AIDS/STD Control and Prevention (NCAIDS) of the Chinese Center for Disease Control and Prevention (Chinese CDC), Tianjin CDC, Tianjin Hongqiao District CDC and the Tianjin Dark Blue Working Group, a non-governmental organization (NGO) for MSM. The cost of entry to the bathhouse was 20 CNY (approximately US$3.28). Between March 2011 and September 2012, 1642 men attending this bathhouse were recruited through both snowball sampling and convenient sampling. Volunteers from the Tianjin Dark Blue Working Group encouraged the patrons to seek testing, and the patrons who accepted voluntary testing then referred their friends to receive HIV testing. Eligibility criteria included being 18 years of age or older, having had oral intercourse or AI with another male in the last 12 months, and being willing and able to provide informed consent. Attendees who self-reported being diagnosed with HIV were excluded. After obtaining informed consent, participants completed rapid HIV testing and an anonymous face-to-face questionnaire administered by trained staff from the Tianjin Dark Blue Working Group.

Study design
The questionnaire collected information on socio-demographics, AIDS knowledge, sexual risk behaviours, HIV risk awareness and HIV testing history. Socio-demographics included date of birth, marital status, residence, ethnicity, length of stay in Tianjin, educational level, STI history and money boy (MB) status. Participants reported whether or not they had been diagnosed with STIs in the past year. MBs who sell sex to men were defined as those who self-disclosed their status during the questionnaire survey. Additionally, participants were assessed on their knowledge of HIV and AIDS prevention and transmission through eight questions adapted from the Chinese national surveillance survey. The eight questions included the following: (1) whether people can look healthy when infected with HIV; (2) whether mosquito bites can transmit HIV; (3) whether HIV can be transmitted through sharing meals; (4) whether HIV can transmit through blood transfusion and using blood products; (5) whether the use of non-sterile needles can transmit HIV; (6) whether HIV can be spread from mother to child; (7) whether condom use can reduce HIV transmission; and (8) whether having sex with only one person will prevent HIV infection. Participants were considered knowledgeable about HIV and AIDS if six or more questions for each area were answered correctly. HIV risk awareness was defined as the perception of the likelihood of being infected with HIV, and was divided into five levels of very high, high, moderate, low and none. For HIV testing history, participants were considered first-time testers if they had never received HIV testing before and repeat testers if they had ever received HIV testing and were not diagnosed with HIV.
HIV testing and counselling All participants were presented to pre-testing HIV counselling and then testing. It took half an hour to receive the results, during which the informed consent and the questionnaire were completed. HIV testing was performed initially using a rapid saliva test (AWARE, Beijing, China) for screening. If the initial result was negative, the subject was considered HIV negative and was given a post-test counselling following national guideline. If the initial screening test was positive, then an immediate repeated rapid saliva HIV test with the same brand of test kit and a rapid HIV blood test (ACON, Hangzhou, Zhejiang province, China) would be performed. If the results from either or both tests were positive or indeterminate, the participant was considered HIV positive and was referred to treatment services. NGO staff who received extensive VCT training provided pre-test counselling and post-test counselling in a routine manner as a risk reduction strategy.

Statistical analysis
Data were double entered into a database constructed using EpiData 3.0 (EpiData Association, Odense, Denmark). Age was recoded categorically for analysis. Analyses of differences in the demographics and behavioural characteristics between first-time and repeat testers were conducted using the chi-square test. To identify predictors for HIV infection, we conducted univariate and multivariate logistic regression analyses among the overall MSM group, first-time tester and repeat tester groups, respectively. All variables with p B0.1 in the univariate logistic regression model were then entered into a multivariate regression model using forward stepwise selection with the significance level set at p00.05. Statistical analysis was performed using SPSS version 13.0 software (IBM Corporation, Chicago, IL).

Ethic statement
The study protocol has been reviewed and approved by the Institutional Review Board of NCAIDS of Chinese CDC (IRB approval number: X130205267). A written informed consent was obtained from all study participants. Incentives of 50 CNY in cash (equivalent to US$8. 19) and a pack of three condoms as a gift were provided for subjects who participated in the study.

Study participants
All participants who were approached and received testing between March 2011 and September 2012 consented to participating in the study. The mean age of the 1642 MSM who completed the survey was 36 years (with a range of 18 to 76 years). Most were aged 35 to 44 years (33.7%), followed by men aged 25 to 34 years (33.6%), older than 45 years (19.3%) and younger than 25 years (13.4%). Most were married (57.4%), ethnically Han (97.4%) and did not have a Tianjin household registration (56.2%). However, the majority had resided in Tianjin for more than two years (62.3%). Roughly three quarters of the participants had completed middle school education. In the past year, 75 participants (4.6%) were diagnosed with STIs and 88 (5.4%) worked as MBs.

Demographic characteristics of repeat and first-time testers
Of the 1642 respondents, 699 men (42.6%) had received HIV test once or more in the past and 943 men (57.4%) were first-time testers. Among first-time testers, a higher proportion consisted of men older than 35 years (58.2%) compared to repeat testers (45.9%) (x 2 024.428, pB0.001). A higher percentage of first-time testers were married or cohabitating with a female partner (62.3%) than repeat testers (50.8%) (x 2 021.971, pB0.001). Though the majority of first-time testers had attended at least some high school (71.5%), repeat testers had a slightly higher proportion of those with this level of education (77.4%) (x 2 07.317, p 00.007).
Among repeat testers, the proportion of participants who had STIs in the past year were 6.4%, which was significantly higher than that of first-time testers (3.2%) (x 2 09.766, p 00.002). The percentage of respondents who worked as sex workers was 8.4% and 3.1% in repeat testers and first-time testers, respectively, which differed significantly (x 2 022.784, pB0.001) ( Table 1).
Differences in HIV and AIDS knowledge and high-risk sexual behaviours between repeat and first-time testers Compared with first-time testers, repeat testers reported a higher awareness rate of HIV and AIDS knowledge (95.1% vs. 88.2%; x 2 023.787, p B0.001), and were significantly more likely to practice AI (91.1% vs. 84.7%; x 2 014.989, p B0.001) and commercial male sex (13.0% vs. 7.6%; x 2 013.011, p B0.001). A majority of both repeat testers (64.2%) and first-time testers (88.0%) reported having engaged in UAI with men and this difference was statistically significant (x 2 033.216, p B0.001). First-time testers were also more likely to ever practice UAI during group sex (73.0% vs. 57.0%; x 2 05.260, p00.022), have UAI with their last male commercial sex partner (36.1% vs. 19.8%; x 2 05.440, p 00.02) and unprotected vaginal sex with a female partner  Table 2).
HIV prevalence and associated risk factors among overall MSM population, repeat and first-time testers A total of 203 MSM were tested HIV positive with an overall prevalence of 12.4% (203/1642). Seventy-eight men among the 699 repeat testers (11.2%) and 125 of the 943 first-time testers (13.3%) were found to be positive for HIV. There was no significant difference in HIV prevalence between first-time testers and repeat testers (x 2 01.629, p00.202) ( Table 3).
Among repeat testers, in the univariate analysis (Table 4), non-local residents were two times more likely to be HIV positive (OR 02.0, 95% CI 01.2Á3.2, p 00.009). Having lived in Tianjin for more than two years were less likely to be infected with HIV (OR 00.5, 95% CI00.3Á0.9, p 00.012).
For first-time testers, MSM being over 45 years of age (OR 02.4, 95% CI01.1Á5.2, p00.025) and MB (OR 03.7, 95% CI01.7Á8.1, p 00.001) remained at an increased likelihood of being HIV infected. Not having UAI during group sex was a protective factor (OR 00.5, 95% CI 00.3Á 0.9, p00.010) ( Table 4). In the multivariate model, being aged 35 to 45 years (aOR 02.9, 95% CI01.3Á6.4, p00.010) and over 45 years (aOR 03.6, 95% CI01.6Á8.3, p00.003), being a male sex worker (aOR 06.3, 95% CI02.6Á14.9, p B0.001) and not having UAI during group sex (aOR 00.5, 95% CI00.3Á0.9, p 00.018) remained significant (Table 4). There was a significant positive trend in HIV prevalence with age among first-time testers (x 2 trend 09.816, p00.002), but no trend was detected across age groups among repeat testers (x 2 trend 00.076, p00.783).  There was no difference in HIV prevalence between repeat testers and first-time testers. However, the context of HIV risk differed for repeat testers and first-time testers. Unknown serostatus persons were more likely than HIV-negative persons to engage in UAI with male partners, including during group sex and commercial sex. A higher proportion of first-time testers also reported practicing unsafe sex with their female partners. For first-time testers, VCT should be tailored to include basic HIV and AIDS knowledge as well as safer sex practices with both men and women. Given the high rate of both unprotected sex with female partners and marriage among first-time testers, comprehensive knowledge of safer sex practices is important to limit the transmission of HIV to otherwise low-risk female partners [20]. Our findings suggest that previous findings among Chinese MSM where married men were less likely to practice risk behaviours in order to protect their partners may not be applicable to this population [7].
In comparison, repeat testers were more likely than first-time testers to be not only knowledgeable about AIDS but also to have had an STI in the past year. Because risk behaviours for STIs are similar to those of HIV, it is likely that a substantial minority of MSM continue to engage in high-risk sexual behaviours [10]. As the risk of acquiring and transmitting HIV is amplified in the presences   of STIs, it is necessary that post-test counselling addresses risk reduction for those chronically engaging in sexual risk behaviours [21]. This study also confirmed that male sex workers bore a disproportionate burden of HIV infection among both repeat and first-time testers. In this study population, MBs were three times more likely to be infected with HIV compared with non-sex workers. Among first-time testers, over onethird of all male sex workers were infected with HIV, and the risk of HIV infection for male sex workers was six times higher. These disparities are comparable to those reported in a study among MSM in Chengdu, China [22]. In our study, first-time MB testers were more likely to practice UAI during the past six months (82.8% vs. 54.2%; x 2 06.835, p 00.009). This result further supported the need to increase interventions targeting this group, including introducing VCT to MBs and increasing condom use not only with male clients but also with male casual sex partners and boyfriends [23,24].
Over half of the patrons were non-local residents, among which almost half of the non-local residents had lived in Tianjin less than three years. Non-local residents were 1.6 times more likely to be HIV positive. Risk behaviours were more common among floating population than local residents because they were more likely to transmit HIV to people from/to other areas [25]. Health interventions which targeted the floating population should be carried out, such as rapid HIV screening.
There are limitations to the present study. First, this study was conducted in one MSM bathhouse, which may hinder generalizing our findings to the larger MSM population. Ethnographic data on MSM in China have found that sexual practices and risks vary across venues, so findings such as HIV risk from this bathhouse population may be substantially higher than lower risk venues of bars and teahouses [7,26,27]. Additionally, because our study was conducted at a venue rather than in the larger community, we were not able to reach MSM who did not frequent venues. However, given that widespread stigma against homosexuality is a significant barrier to participation in HIV prevention activities, it would have been difficult to conduct this study outside of a dedicated MSM venue [28]. Second, lack of response rate further undermines the generalizability of the study results. It is difficult to estimate how many and what proportion of the total number of customers has taken up the testing because hundreds of customers come in to the bathhouse every day and are constantly coming and going. Also, the fact that consent was collected might have influenced the composition of the sample. Men who did not want to be identified or documented as HIV positive may have been unwilling to take part in the study. However, we believe that the impact was minimal because all participants who were approached consented to participating in the study. Third, this study was cross-sectional, so causation cannot be established. Future longitudinal studies are needed to establish causation between repeat testing and changes in HIV risk. Finally, our study took a convenience sample. Snowball sampling might have influenced the ratio of previously tested because participants referred those they knew, who were likely to be quite similar

Conclusions
HIV prevalence differs significantly between repeat and first-time MSM testers. For MSM who have never been tested, VCT may be an important strategy to detect more HIV cases, especially among older MSM. For both repeat and first-time testers, VCT should be tailored to meet the specific needs of these groups. Given the higher proportion of repeat testers who are sex workers compared to first-time testers, VCT may be a viable channel to provide multiple opportunities for risk reduction consultations for this hidden population. Future studies should examine differences in HIV risk behaviours and social networks for older versus younger MSM, and how VCT should be implemented to reach subgroups such as older MSM to mitigate disparities within this population.