The Dynamic Trends of HIV Prevalence, Risks, and Prevention among Men Who Have Sex with Men in Chongqing, China

Objective. This study was to characterize the continuously changing trends of HIV prevalence, risks, sexual behaviors, and testing behaviors among men who have sex with men (MSM) in Chongqing, China. Methods. Five consecutive cross-sectional surveys were conducted among MSM in 2006, 2008, 2010, 2012, and 2013. Testing for HIV and syphilis was performed, and HIV risks, sexual behavior, prevention, and HIV testing behavior were collected using the same questionnaire. Results. HIV prevalence increased from 13.0% to 19.7% from 2006 to 2013 (P = 0.004), with an increase of 1.0% per year. Syphilis prevalence peaked in 2008 with a positive rate of 11.6% and then experienced a sharp drop to 2.8% in 2012 and 2.9% in 2013. Percentage of those who ever received HIV testing in the last year increased from 17.0% to 43.3% (P < 0.001); condom use at the last anal intercourse and reported consistent condom use in the last 6 months increased from 51.8% to 71.0% (P < 0.001) and from 24.7% to 47.9% (P < 0.001), respectively. Conclusions. HIV continued to spread among MSM in Chongqing even when a decline in prevalence of syphilis and increase in awareness rate, condom use, and HIV testing seeking behaviors seemed to occur.


Introduction
Unprotected sex among men who have sex with men (MSM) is the main route of HIV transmission in many countries and regions that were first affected by the HIV/AIDS epidemic, such as North America, Western Europe, and Australia [1]. Over the past years, HIV infection rate among MSM is on the decline in some cities in the United States [2]. On the contrary, HIV infection rate is increasing obviously among MSM in China [3,4]. According to the national estimations of the HIV/AIDS epidemic in China, the proportion of HIV/AIDS cases infected via homosexual contact increased from 7.3% in 2005 to 13.0% in 2011 [5,6]. It was estimated that 29.4% of new HIV infections in 2011 were transmitted via homosexual contact [6]. In 2008, a survey was conducted among 18,101 MSM in 61 cities in China and indicated an overall HIV prevalence of 5.1% among MSM. In some cities, HIV infection rate exceeded 10% among MSM [7,8].
Chongqing is a city affected most heavily by homosexual transmission. The first HIV infection via homosexual contact was reported in Chongqing in 2005. Since then, the proportion of new HIV/AIDS cases via homosexual contact has shown a rapid upward trend [9], reaching 22.2% from January to June 2013, which is 0.8% higher for the same period in 2012. Understanding the trends of HIV/AIDS epidemics is crucial to identify priorities, revise strategies, and allocate resources in the response to HIV. In Chongqing, serological and behavioral surveillance has been conducted among MSM since 2006, and sound historical data have been accumulated [10,11]. This study aims to capture the changes in the trends of the HIV/AIDS epidemic and risk behaviors of the MSM population based on the surveillance data from 2006 to 2013 in Chongqing and provide scientific evidence for the development of strategies and measures in the city's response to HIV.

Study Population.
Men who were at least 18 years of age and had sexual intercourse with at least 1 male during the  23.7% of subjects were migrants. In 2008, 38.0% of subjects (the highest figure across the years covered by the study) were migrants. 62.3% of subjects have received education for at least 13 years. The education level was obviously higher among subjects surveyed in 2010 than those surveyed before 2010. There are significant differences in demographic characteristics of subjects across the survey years ( < 0.001).

Dynamic Trends of HIV and Syphilis.
HIV prevalence among the study participants showed an upward trend and increased from 13.0% to 19.7% from 2006 to 2013 ( = 0.004), with an increase of 1.0% per year. After age stratification, HIV prevalence significantly increased from 6.3% to 19.6% among MSM aged 18-24 ( < 0.001) and had a modest increase

Sexual Behavior and Trends of Reported Condom Use.
Percentage of receiving HIV testing in the last year increased from 16.8% to 43.1% ( < 0.001). Reported condom use at the last anal intercourse and reported consistent condom use in the last 6 months increased from 51.9% to 71.0% ( < 0.001) and from 24.7% to 47.9% ( < 0.001), respectively. HIV/AIDS awareness rate rose from 87.0% to 98.0% ( < 0.001). The proportion of subjects ever having had sex with women decreased from 33.0% to 8.8%. Just like the rate of condom use with men, the rate of condom use with women was also on the increase. The rate of condom use at last sex increased from 30.8% to 51.5%. The rate of consistent condom use in the past six months increased from 21.6% to 45.5%. See Table 3.

Discussion
Despite huge amounts of effort in HIV prevention and control, this analysis has shown that there has been a significant increase in HIV prevalence among MSM population in Chongqing from 2006 to 2013, coinciding with the general trend in China [4,12]. This rise in prevalence is mainly due to the rapid increase in HIV infection among MSM individuals aged 18-24 years, which resulted in the fact that all age groups had similar HIV prevalence in 2013 although there was a huge gap (6.3% for 18-24, 13.7% for 25-34, and 17.7% for over 35) in 2006. Previous study indicated that younger MSM are more likely inclined to have multiple sex partners and engage in more complicated sexual networks and older MSM may prefer to have young MSM partners [13]. The sexual intercourse between older MSM with higher HIV prevalence and young MSM could be a contributing factor to the increase of HIV prevalence among young MSM. HIV prevalence among older MSM has stabilized in recent years, as a result of more people having access to ART treatment and decreasing number of MSM dying from AIDS. However, it is also likely that the HIV prevalence may have been overestimated. Firstly, in all rounds of surveys, HIV positive cases were defined as two ELISA positives, which may include some false positives. Secondly, the participants were conveniently sampled. One meta-analysis indicated that the unprotected anal intercourse prevalence was significantly lower in studies that used random or systematic sampling methods compared with studies that used convenience sampling, as more highrisk MSM could have been recruited, resulting in higher estimate of HIV prevalence [14].
Our results suggest that there is a significant increase in condom usage and more MSM are willing to seek HIV testing. This could be partly a result of the numerous MSM intervention projects implemented by the local government, as well as the work of all the MSM nongovernment organizations in Chongqing these years [10]. The positive trend could also be seen in other studies in China [15,16]. However, the current level of condom usage is still very low and could be a reason for the elevated prevalence. Many countries across the world are struggling to reduce unprotected sexual behavior among MSM [17,18] and to reduce HIV infection among MSM, additional effort is needed to expand HIV prevention programs, taking into consideration of the biomedical, behavioral, and structural aspects of the epidemic, which could lead to possible improvements in safe sexual behavior, HIV testing seeking and access to care, and subsequently a decline of HIV prevalence in the long run.
There are a few reasons for the discrepancy between HIV prevalence increase and condom use rate. Firstly, prevalence reflects joint action of incidence and mortality. Prevalence might go up when both incidence and mortality go down. In fact, with expansion of ART, the mortality does decrease significantly in recent years in China [19]. In general, population incidence can reflect epidemic trends most directly. The surveillance of incidence, however, has not been implemented widely due to constraints of laboratory capacity, staffing, timing, and resources. Many studies adopted prevalence from continuous cross-sectional surveys to estimate the trends of HIV/AIDS epidemic [20,21]. Secondly, Sentinel surveillance is grounded on anonymous unlinked survey. The duplication of the recruitment over the years might have overestimated condom use rate and HIV testing rate.
The analysis indicated a significant decline in syphilis infection rate in Chongqing, which could also be seen in other Chinese cities [22]. Since TRUST reagents were not used for syphilis screening in 2006 and 2008, the use of single ELISA might produce a high proportion of false positive cases. Excluding data from these two years, an obvious decline was still observed for syphilis infection rate after 2010. Additionally, the proportion of subjects self-reporting sexually transmitted diseases in the previous year was also on the decline, consistent with syphilis testing results and coinciding with the self-reported increasing rate of condom use. This indicated that safer sex could be effective in preventing STDs but needed to be confirmed by more data. Results showed that the rate of syphilis infection and selfreported STD was higher among HIV-positive subjects than among HIV-negative subjects, consistent with other studies [23]. The high rate of coinfection justified the necessity of screening in both HIV positive cases and STD patients.
There are several limitations in this analysis. Firstly, the sentinel surveillance relied on convenience-based samples and the participants were recruited through several different BioMed Research International 5 methods in every round. Since the characteristics of samples in each survey showed statistically different results, stratified analysis was performed to minimize the effect of the change in population structure.
Secondly, changes in HIV tests may also influence trends. HIV antibody tests have improved over the years. New generation assays are more accurate and the rate of falsepositive results that may skew prevalence becomes lower. Thirdly, Reporting bias due to desirability is more likely to affect the assessment of trend data. Since incidence is the best indicator of HIV trend in a short period, trend analysis based on incidence estimates was strongly recommended for future study.