Prevalence of HIV and Syphilis Infection among Men Who Have Sex with Men in China: A Meta-Analysis

Objectives. To figure out the most current prevalence of HIV and syphilis in MSM in China. Methods. A meta-analysis was conducted on the studies searched through PubMed, CNKI, and Wanfang published between 1 January 2009 and 11 April 2013. Results. Eighty-four eligible studies, either in Chinese or in English, were included in this review. The pooled prevalence of HIV and syphilis infection in MSM in China was 6.5% and 11.2%, respectively. The subgroup analyses indicated that the prevalence of HIV infection was higher in the economically less developed cities than that in the developed cities (7.5% versus 6.1%, P < 0.05). In contrast, the prevalence of syphilis infection was lower in less developed cities than in developed cities (8.6% versus 15.1%). Studies with a sample size smaller than 500 had a lower prevalence of HIV and syphilis infection than those with a sample size greater than 500 (5.9% versus 7.2% for HIV; 11.0% versus 11.5% for syphilis, respectively). Conclusions. HIV and syphilis infection are prevalent in MSM in China. The different prevalence of HIV and syphilis infection between developing and developed cities underscores the need to target prevention strategies based on economic conditions.


Introduction
The human immunodeficiency virus (HIV) and syphilis infection epidemics present an intractable problem to the world. This is also true in China, where there is a marked increase of people living with HIV/AIDS, from 740,000 in 2009 to 780,000 in 2011. Approximately 17.4% of new HIV cases are attributed to homosexual transmission [1]. It was said that the epidemic of HIV infection among men who have sex with men (MSM) has been the gravest of the six newborn HIV/AIDS challenges in China [2]. From 2009 to 2013, the prevalence of HIV among MSM ranged from 1.7% to 21.0% in China [3,4]. A recent meta-analysis reported that almost 5.3% (95% CI: 4.8%-5.8%) of MSM in China were living with HIV, which was more than 90 times higher than the general public [5]. The prevalence of syphilis in MSM increased from 6.9% to 11.8% between 2003 and 2009 [6,7].
The fast growth of the HIV epidemic has been observed in some reports [5,8]. Although there have been several metaanalyses on prevalence of HIV and syphilis in MSM [5,8,9], the most recent meta-analysis only included studies until 2008. Thus, it is important for us to have a better and more thorough understanding of the current prevalence of HIV and syphilis. Therefore, we performed a meta-analysis of all available literature published from 2009 to 2013.
Given the poor awareness of AIDS-related knowledge in economically developing cities, which may lead to higher prevalence of marriage and bisexuality and lower condom use than economically developed cities, we also carried out a subgroup analysis to explore the differences in HIV and syphilis prevalence between developing and developed cities in China. Results from the analyses may guide future policies for more rational allocation of prevention and treatment resources.

Data Extraction.
The first author extracted the following data from the included studies, which was later verified by the second author: the first author and published year, study period, study site, sampling methods, average age of MSM participants, sample size, and prevalence of HIV infection or syphilis infection or both. The study sites were divided into developing and developed cities (as assessed on the basis of based on per capita gross domestic product (GDP) for each city). Developing cities were defined as those with a per capita GDP lower than 15,000 RMB and developed cities were defined as cities with per capita GDP higher than 15,000 RMB. GDP of each city per personyear was obtained from http://tieba.baidu.com/p/2312163828. Disagreements were resolved by discussion. If no consensus could be reached, a third author would be referred to.

Validity Assessment.
Validity assessments were accomplished by two workers. Studies were considered to be of high validity if they met the following criteria: (1) reported both HIV and syphilis infection; (2) used two or more recruitment methods; (3) sample size larger than 500; (4) published in English.

Statistical Analysis.
Meta-analyses of the prevalence of HIV and syphilis infection among MSM were carried out by Meta-Analyst software (Beta 3.13). -test (significance level < 0.10) and 2 (significance level 2 > 0.25) statistics were calculated to measure homogeneity of studies. Studies that had either ⟨0.10 2 ⟩ 0.25 were defined as heterogeneous. Potential publication bias was assessed by Begg's test. Pooled prevalence of HIV and syphilis infection, 95% confidence intervals, and the relative weight for each study were calculated by using random effects models. Due to the heterogeneity which is commonly observed in meta-analysis concerning sensitive groups such as MSM, we conducted a meta-regression with 10,000 permutations in a Monte Carlo simulation to explore the sources of betweenstudy heterogeneity with the following covariates: economy status, sampling methods, sample size, and published year. Subgroup analyses were conducted to obtain the prevalences of HIV and syphilis from different stratifications.

Study Selection for Meta-Analyses
. 1222 articles were retrieved after the initial search from databases and another 6 articles were included from reference lists. A total of 1096 articles were removed after screening of titles and abstracts because they were not topically relevant ( = 1077), were reviews ( = 7), and used the same data ( = 12). 151 articles were eligible for full-text screening and 67 were further excluded because they did not report the sampling methods or study site ( = 19), did not mention the study period ( = 5), had a sample size smaller than 150 ( = 4), included MB or young MSM ( = 6), presented neither HIV nor syphilis prevalence ( = 12), just used one method to diagnosis syphilis ( = 20), and tested HIV infection by urine ( = 1). The detailed process of data selection is illustrated by Figure 1.

Study Characteristic.
Eighty-four studies with estimates of HIV prevalence (English 18, Chinese 66) and seventysix studies with estimates of syphilis prevalence (English 17, Chinese 59) were finally identified in this review, which covers 59 cities from 29 provinces and municipalities in China (no studies from Hainan, Tibet, Hong Kong, Macao, and Taiwan). The prevalence of HIV and syphilis ranged from 1.7% to 21.0% and 0.9% to 31.2%, respectively [3,4,. In the meta-analysis of pooled prevalence of HIV, the sample size ranged from 150 to 6101 (total 71,845), and in the estimating of syphilis prevalence, the sample size oscillated between 157 and 6101 (total 65,162). About half of them were located in developing cities (50.7% and 51.4% in analyses of HIV and syphilis prevalence, resp.,). The following sampling methods were used to recruit participants: snow ball ( = 27), respondent driven sampling (RDS) ( = 15), multiple sampling methods ( = 23), and others ( = 19). The basic information of the 84 included studies is represented in Table 1.

Meta-Regression and Subgroup Analyses.
Results of metaregression indicated that economic status and sample size contributed to the heterogeneity between studies ( Table 2).
Four and two studies were further excluded because of the absence of specific study cities in the subgroup analyses of HIV and syphilis prevalence by different economic status, respectively. The prevalence of HIV infection was slightly higher in developing cities (7.5%) than in developed cities (6.1%), whereas the prevalence of syphilis was significantly lower in developing cities (8.6%) than in developed cities   (15.1%). Studies with a sample size smaller than 500 had a lower prevalence of HIV and syphilis infection than those with a sample size greater than 500 (5.9% versus 7.2%, 11.0% versus 11.5%, resp.) ( Table 2). Although the published year and sampling methods did not contribute to between-study variance, we also conducted subgroup analyses based on them. Although we only divided the published year into two groups, an uptrend of HIV infection and a decrease in syphilis infection were still observed (Table 2).

Discussion
This is an updated meta-analysis that presents the most current prevalence of HIV and syphilis infection in MSM in China. In order to obtain the prevalence from developing and developed cities, we also carried out a subgroup analysis by different economic status according to meta-regression. To our knowledge, this is the first study to compare the prevalence of HIV and syphilis infection between developing and developed cities in China. A national study survey of 61 cities by Wu reported the prevalence of HIV in MSM from the southwest, east, south, and northeast of China but did not account for the different economic conditions of each city [6].
In  [8]. Results from our study indicate that the prevalence of HIV among MSM is still increasing. The expanding epidemic may be associated with the special role of MSM in China. Homosexuality is still not widely accepted by the general population. Therefore, marriage between MSM and women remains common, nn which case MSM might act as a bridge for HIV transmission to other MSM and the general population. It was reported that the prevalence of bisexual behavior among MSM in China was as high as 31.2% [124].
The subgroup analyses showed that the prevalence of HIV infection was 7.5% from developing cities and 6.1% from developed cities, whereas the prevalence of syphilis was 8.6% from developing cities and 15.1% from developed cities. The exact reasons for this finding are unclear, but one possible reason is that syphilis is a curable disease and several large scale public health programs were conducted in several developing cities in the past few years [74].
Several limitations of our analyses must be kept in mind. First, substantial heterogeneity was observed in our analysis. We found that study site, sample size, and sampling methods contributed to the heterogeneities. These factors may also have contributed to the increasing trend of HIV and syphilis infection in MSM in China. A meta-analysis of Chow reported that articles published in English were more likely to present high prevalence than Chinese articles, which may have resulted in between-study variance. The presence of publication bias was also a considerable limitation in our analysis (Figure 4). With the strict inclusion criteria, we excluded studies with a sample size smaller than 150 and those with only one testing method to diagnose syphilis. There are also a large number of unpublished articles (e.g., government documents) and studies that reported relatively low prevalence. Third, there were few studies from rural areas, which may have led to an overestimation of HIV and syphilis prevalence. More studies of rural MSM studies are needed in the future. Fourth, several studies that did not specify the city location were excluded from the subgroup analysis, which may have increased the possibility of publication bias. Fifth, the economic classification of the cities was only based on the per capita GDP; results may only partially reflect the true economic status of the city. Despite the limitations listed above, our meta-analyses employed a strict inclusion criteria and valid search strategy in order to provide an objective, authentic, and current estimate of HIV and syphilis prevalence among MSM in China, with a large aggregate sample size of 75036. We also conducted a subgroup analysis to present the different prevalence rates of HIV and syphilis infection in developing and developed cities.
In conclusion, HIV and syphilis are prevalent among MSM in China; HIV prevalence is higher in developing cities than in developed cities, while the situation of syphilis is just the opposite. These results urgently indicate the need for strategies aimed at prevention, surveillance, and treatment. Moreover, corresponding policies should be drawn up by the local government on the basis of local economic status.

Conflict of Interests
There are no conflict of interests.