Since 1981, when the first case with acquired immune deficiency syndrome (AIDS) was reported in the USA, human immunodeficiency virus (HIV) infection has become a worldwide epidemic.1 China’s AIDS epidemic began in the early 1990s among needle-sharing, injecting drug users (IDUs) in some border communities in Yunnan and Xinjiang provinces.2 In the middle 1990s, while the HIV epidemic in border regions slowly spread to nearby regions and inland China, plasma-collection activities in central China such as Henan and surrounding provinces caused large numbers of commercial plasma donors to become infected with HIV through infusion of pooled contaminated blood cells. At the same time, HIV was also spreading through sexual transmission.3 4 By 1998, HIV had reached all 31 provinces and was in a phase of exponential growth, which, by 2007, had culminated in an estimated 700 000 infections.5

Sexual transmission is now the main mode of HIV transmission in China.5 Among HIV positives, 40.6% were infected through heterosexual transmission and 11.0% through homosexual transmission. The trend of HIV spread from high-risk groups to the general populations also occurred mainly through sexual transmission.4 At the present time, men who have sex with men (MSM) have become one of the high-risk groups for HIV infection and other sexually transmitted diseases (STDs) because of their active sexual activities without protection.2 It is estimated that 70% of MSM have had sex with more than one partner in the past 6 months, and only 30% use condoms for anal sex, while 50% use condoms when they have commercial homosexual encounters.5 Among different transmission modes, homosexual transmission was 0.4% in 2005 and showed a dramatic increase to 3.3% in 2007.5

There are between 2 and 8 million MSM in China, and the rapidly rising prevalence of HIV/AIDS among this high-risk population draws worldwide concern.6 7 Multiple epidemiological studies have addressed HIV prevalence among MSM in China in recent years.8^–33 However, the results were heterogeneous, ranging from 0% to 20%, due to the sample size limitation and varied sampling methods. In addition, the poor HIV testing uptake rate (less than 20%) among MSM made epidemic estimates more difficult and may have contributed to potential bias in the previous studies.9 12 The objective of this study is to summarize HIV prevalence among MSM in China using meta-analysis based on systematic review of published articles. We also assessed the prevalence of syphilis in the included studies.

METHODS

Literature search

Studies investigating HIV infection among MSM in China were identified by searching for articles in the MEDLINE database and Chinese Technology Periodical Database until 15 September 2008. Various combinations of the terms “HIV,” “AIDS,” “gay,” “MSM,” “sex between men,” “homosexual,” “bisexual” and “China” were used to screen for potentially relevant studies. Additional studies were also identified using cross-referencing. Cross-sectional or cohort studies presenting original data of HIV infection prevalence in MSM population in China were included. If the study was reported in duplicate, the article published in English or published earlier was included in our analyses.

Data extraction and statistical analysis

For all studies, we extracted the following data from original publications: first author and year of publication; study site and period; sampling methods, sample size and age of the participants; methods and results of HIV infection and syphilis detection (study period and age of the participants are presented in supplementary table).

In this review, events of HIV infection and syphilis were extracted from the studies where available. For some studies, numbers had to be calculated from reported positivity. Meta-analysis on the prevalence of HIV infection was carried out by Comprehensive Meta-Analysis (V2.0, Biostat, Englewood, New Jersey). Stratified analyses were performed by study locations (for the most widely studied sites with at least five studies), sampling size (⩽200 or >200) and sampling methods (MSM venue; MSM network and voluntary confidential testing (VCT); respondent-driven sampling (RDS) and snowballing sampling). The χ² test was used to assess the differences between the subgroups. Random effects models were used for meta-analysis, taking into account the possibility of heterogeneity between studies which was tested with the Q test (p<0.10 was considered indicative of statistically significant heterogeneity) and the I2 statistic (values of 25%, 50% and 75% are considered to represent low, medium and high heterogeneity respectively). The Begg rank correlation method12 was used to assess the potential for publication bias (p<0.05 was considered indicative of statistically significant publication bias). Similar analyses were performed for syphilis.

RESULTS

Eighty-eight articles were identified by the search terms and 56 were excluded by screening abstracts (30 were not cross-sectional or cohort studies, 21 were reviews, and five were duplicated in the two databases), leaving 32 articles8^–39 for full-text review. Of these, six were excluded34^–39 from this review because of the duplicated results from the same study population (for a flow diagram of study identification, see supplementary figure).

Characteristics of the 26 included studies, which represent 12 different locations, are presented in table 1.8^–33 Less than half of them were published in English (n = 10). In most studies, HIV infection was defined by antibody positivity in serum, and only one study used urine samples.24 Twenty studies presented serological detection results of syphilis, but stage-specific data were not provided by most of these studies. The sampling methods were varied, 12 studies sampled from MSM venues, seven studies recruited participants from MSM network or VCT, and nine studies used snowball sampling or RDS methods. The sample size of the studies ranged from 11 to 1292. In summary, there were 9288 and 6248 subjects included in the meta-analyses on prevalence of HIV infection and syphilis, respectively.

Figure 1 shows the summarised estimate of HIV prevalence among MSM during 2001–2008 countrywide, which is 2.5% with 95% CI of 0.9% to 3.3% ranging from 0.2% (0% to 3.8%) to 20.0% (9.3% to 37.9%). No obvious publication bias was observed (p = 0.11). However, there is a substantial heterogeneity between studies (p for Q test <0.01; I2 = 71.33). As shown in table 2, stratified analyses were conducted according to sampling methods, sampling size and study locations. Heterogeneity between studies was reduced, at least in part, in subgroup analyses. A lower summarised prevalence (2.1%) was observed for RDS or snowballing sampling method-based studies as compared with the prevalence (3.8%) of the studies sampling by VCT or MSM network (p<0.05). The subgroup of studies with larger sample size (>200) showed a lower prevalence of 2.1% (p<0.05). All of the included studies were conducted in urban areas. For the most widely studied two locations, Beijing and Guangdong, the summarised prevalence was 2.6% (1.7% to 3.9%) and 2.3% (1.4% to 3.9%), respectively. The difference between the two sites was non-significant (p = 0.90).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Meta-analysis of the HIV prevalence among men who have sex with men in China.

As shown in fig 2, a high prevalence of syphilis seropositivity was observed in the MSM population among included studies with a summarised estimate of 9.1% (7.6% to 10.8%), which ranged from 1.3% (0.3% to 4.9%) to 19.3% (13.8% to 26.4%). There was a significant heterogeneity between studies (p<0.01; I2 = 77.35). Marginal publication bias was observed as assessed by the Begg rank correlation analysis (p = 0.03).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Meta-analysis of syphilis prevalence among men who have sex with men in China.

DISCUSSION

This review addressed the prevalence of HIV infection among MSM in China. Twenty-six eligible studies, published during 2001–2008, were included. The results of these studies were heterogeneous. The summarised estimates of prevalence of HIV infection and syphilis, assessed by meta-analyses, was 2.5% (0.9% to 3.3%) and 9.1% (7.6% to 10.8%), respectively. Differences between sampling methods, sampling size and study locations need to be taken into account and may explain some of the inconsistencies observed between studies.

MSM are at high risk for HIV infection in worldwide because of their high partner numbers without protection and high migration rates.40^–42 Among 30 956 AIDS cases reported among men in the USA in 2005, 53% were MSM. HIV prevalence among black MSM (46%) was more than twice that among white MSM (21%).41 The epidemic is severe in the developing countries as well. A systematic review in low- and middle-income countries found a weighted average HIV prevalence of 12.8% among MSM.42 However, MSM data from China are comparatively sparse. Chinese scientists began to study HIV-related risk behaviours among MSM early in 1993, but epidemiological studies assessing HIV prevalence were not conducted until 2000. Recently, several reviews profiled current status of spread and control of HIV among this special population.2 6 However, only descriptive statistics were used in these studies. For the first time, to our knowledge, HIV prevalence among MSM in China was systematically reviewed and summarised by meta-analyses.

Of the 26 included studies, 23 studies were performed in the past 5 years (2004–2008), sampling methods were varied, and sample size ranged from 11 to 1292. HIV antibody positivity among MSM was heterogeneous and ranged from 0% to 20%. The results of meta-analyses showed that the summarised estimate of HIV prevalence was 2.5% (0.9% to 3.3%), which is still much lower than the data (12.8%) from the other developing countries.42 However, compared with a systematic review on heterosexual transmission of HIV in China,43 our results showed a higher prevalence among MSM than among other risk populations, such as female sex workers (with an average infection rate of 0.88%) and STD patients (0.24%).

In the stratified analyses, heterogeneity between studies was reduced, suggesting that the varied sampling method, sampling size and study locations may partly explain the inconsistency observed among the studies. A higher prevalence of HIV infection was found in the studies sampling though VCT or MSM network compared with studies using RDS or the snowballing sampling method. A possible explanation could be that individuals with high-risk behaviours are more likely to take up HIV testing through the VCT clinic or MSM network.44 Such selection bias should also be considered for sampling from venues where unprotected sexual contact behaviour and transmission of STDs are more common. Therefore, more scientific and standard sampling methods are essential for more precise estimation of HIV epidemic among MSM. In addition, sample size could also be a probable explanation for the potential bias. As our results suggested, the prevalence may be overestimated by the studies with smaller sample sizes. Locations of the studies should be taken into account as well, even though no significant difference was found between the two most wildly studied regions, Beijing and Guangdong. However, studies reporting HIV testing among MSM were mostly conducted in urban areas in China. Very few studies were performed to address the MSM population in rural areas.45 Further studies of MSM in rural China are needed because of their high migration rate. This population may be an important potential bridge to spread STDs from cities to the general populations in the rural areas.

In the included studies, the high prevalence (9.1%) of syphilis among MSM may indicate a high prevalence of unprotected sex behaviours and suggests a potential risk of rapid HIV spread among MSM, although syphilis is more infectious than HIV. Recently, a systematic review on China’s syphilis epidemic also reported high levels among MSM with a median prevalence of 14.6%.46 In China, most homosexual men hide their sexual orientation, and about one-third of them are married.2 Therefore, MSM may play a bridging role in the spread of HIV and other STDs from the high-risk population to the general population. However, most included studies did not analyse HIV infection and syphilis-related risk factors in detail, which makes it difficult to ascertain the correlation between syphilis and HIV infection. Additional epidemiological surveys among MSM are required to better understand their population size, demographic characteristics and the epidemic situations, thereby assisting the design of effective comprehensive intervention models for MSM. Meanwhile, the assessment of the feasibility and efficiency of sampling methods should be addressed to develop an effective surveillance strategy and to facilitate year-to-year and location-to-location comparisons.

The limitations of this study should be kept in mind. First, as pronounced results are more likely to be published, publication bias cannot be excluded, although no indication of major publication bias was evident in the meta-analyses. Second, some of the included studies were sampled in MSM venues or through VCT rather than population-based which makes the results more prone to potential selection bias. Third, one study detected HIV antibodies using urine samples.24 It cannot be claimed to be perfect to summarise the results of different methods due to their different accuracy. Fourth, because not all necessary information could be obtained from all included studies, relevant detailed analyses could not be performed to determine the characteristics of the infected MSM and related risk factors. Furthermore, small sample sizes also restricted subgroup analyses. Finally, our literature search was not designed for syphilis, which may introduce bias into the meta-analysis of syphilis prevalence.

In conclusion, our analyses suggest that MSM form a high-risk population for HIV infection and syphilis in China, and an effective strategy for prevention and control is expected for this specific population. However, the results of the included studies varied greatly. Differences between sampling methods, sample size and study locations need to be taken into account and may explain some of the inconsistencies. To ensure a precise estimate of the epidemic status of HIV infection and other STDs among MSM in China, further large-scale studies, with adequate power and standard sampling methods, are essential in the future.