HIV-Related Behavioral Studies of Men Who Have Sex with Men in China: A Systematic Review and Recommendations for Future Research

Abstract

HIV prevalence among men who have sex with men (MSM) has increased significantly in China since early 2000, with MSM accounting for 32.5% of the newly infected HIV cases in China in 2009. This study reviews a total of 33 studies published in English literature on MSM in China from 2000 to 2009. Existing studies indicated that frequently MSM in China were married (especially older MSM); a majority were migrants and self-identified as gay. A large and increasing proportion of MSM sought sexual partners online. Sexual risk behaviors among MSM were prevalent, including high rates of unprotected sex, concurrent sexual relationships with both men and women, and commercial sex. Illicit drug use, however, was relatively low among general Chinese MSM, but higher among those engaged in commercial sex with men. The existing literature suggests a lack of socio-behavioral studies, lack of intervention studies specifically targeting MSM, and lack of studies in smaller cities, towns, and rural areas. Future studies need to employ more rigorous research methodology, including longitudinal study design, multiple sampling methods, and common measurements of HIV-related risk behaviors.

Introduction

Since the year 2000, the annual number of deaths due to HIV/AIDS has declined worldwide, in part as a result of the substantial increase in access to HIV treatment [1, 2]. However, in China this trend is reversed, with a rise in the HIV/AIDS epidemic in the past decade [3, 4].

China’s HIV epidemic has been described as the “rapid expansion phase” [4, 5]. Between 1995 and 2000, the reported cases of HIV infection in China increased 30% per year; it increased 58% in 2001, 122% in 2003, and 45% in 2007 [6, 7]. The China Ministry of Health (CMOH) estimated that 740,000 people were infected with HIV by the end of 2009, among whom 48,000 were newly infected in the year 2009 [8]. In addition, China’s HIV epidemic has shown new patterns in infection routes [8, 9]. Sexual transmission, including both heterosexual and homosexual transmission, has surpassed intravenous drug use and commercial blood/plasma donation and become the dominant route of HIV infections in China, accounting for more than 70% of the estimated new infections in 2009 [8, 9].

Globally, men who have sex with men (MSM) are considered to be among the most-at-risk populations for HIV infections in virtually all regions other than sub-Saharan Africa [1]. In some developed countries, unprotected sex between men is the major route of HIV transmission. In the United States, more than half (53%) of the new seropositive cases were among MSM in 2006 [10]. In China, the percentages of newly reported HIV cases that were attributable to MSM were 0.2% in 2001, 7.3% in 2005, 12.2% in 2007, and 32.5% in 2009 [8, 11–13]. A national survey among MSM from 61 cities in 2008 revealed that 4.8% of MSM were infected with HIV [8]. Thus, the contribution of MSM to the HIV/AIDS epidemic in China is increasing.

The Chinese government has been making efforts to combat the HIV/AIDS epidemic in the past two decades, but efforts targeting MSM are limited. The dominant public discourse regarding homosexuality in current Chinese society is still largely negative, often associated with words as “abnormal”, “deviant”, “promiscuous”, and “immoral” [14, 15]. Those who disclose their homosexual identity are likely to face severe social discrimination including loss of employment [16]. Due to the predominant homophobia in the society, combined with the strong influence of Confucianism which stresses the continuation of family name, MSM frequently hide and/or deny their homosexual preference by getting married and having children [17–19].

Given the increasing importance of MSM in the HIV/AIDS epidemic in China and the estimated size of the MSM population, ranging from 10 to 20 million [11], addressing the HIV epidemic among this socially marginalized group is both important in and of itself and represents an important strategy to reduce sexual transmission and avert larger HIV epidemic in the general population.

With more attention being paid to MSM, a growing number of empirical studies were conducted in recent years in China. Two review articles and one meta-analysis have been published as efforts to summarize and synthesize these studies. Liu and colleagues wrote the first review on MSM in China in 2006 that covered 16 articles, both published and unpublished, between 1990 and 2004 [20]. Due to the increasing new HIV infection cases among MSM and a growing number of studies focusing on this group, an updated knowledge base on behavioral studies among MSM in China is needed. A recent review by Wong and colleagues included 13 Chinese articles published from 1989 to 2008 and 18 English articles from 1981 to January 2008 [21]. The review examined HIV/sexually transmitted infection (STI) prevention and policy efforts targeting Chinese MSM and provided important recommendations in combating the HIV/AIDS epidemic from structural and socio-cultural perspectives. The review, however, was not designed to comprehensively address methodological issues. A meta-analysis was also published in 2009 that examined 26 studies (16 in Chinese and 10 in English) from 2001 to 2008, and aimed to synthesize the estimates of HIV and syphilis prevalence among MSM in China [22]. The meta-analysis provided analyses of estimates of HIV and syphilis prevalence and their time trend in recent years among MSM in China. However, the meta-analysis was not designed to review the existing behavioral research on MSM in terms of research methodology, individual characteristics of MSM, and their HIV-related sexual risk behaviors such as unprotected sex with both men and women, commercial sex, online partner seeking, and substance use, which are the foci of this literature review.

There are four main objectives of the current review: (1) summarize and evaluate the methodologies employed in the existing studies on MSM in China; (2) examine demographic characteristics of Chinese MSM; (3) examine HIV-related sexual risk behaviors among Chinese MSM; and (4) identify gaps in the existing literature and make recommendations for future research.

Methods

A systematic literature search was conducted to identify studies that met the following inclusion criteria: (1) peer-reviewed studies published in English language journals prior to December 2009, (2) studies that focus on MSM in China, and (3) quantitative studies that report rates of condom use or HIV infection. Articles published in Chinese were not included in this review because readers outside of China would not have access to the literature. We did not include qualitative studies in this review, as we felt that the scope and depth of the qualitative studies merit a separate literature review.

Publications were retrieved from the following electronic databases: Medline, PubMed, PsycInfo, and Sociological Abstract. Each database was searched using the following key words, alone or in combination: China, Chinese, MSM, men who have sex with men, money boys (MBs, men who sell sex to men), male sex workers, homosexual, gay, HIV, AIDS, condom use, and sexual risk. Initially 127 articles were retrieved; 85 articles (including six qualitative studies) were excluded because they did not meet one or more of the inclusion criteria. We hand-searched the reference sections of the previous review articles on MSM in China [20, 21], but found no additional articles that met the inclusion criteria. Among the 42 articles that met the selection criteria, nine reported information published in other articles by the same leading authors based on the same datasets. For the purpose of this review, we combined the information from different articles that were based on the same data and only retained one article from the same dataset to avoid duplication, which resulted in a total of 33 quantitative studies in this review. For nine studies that did not provide basic information on study design (e.g., year of data collection, sampling method) or demographic information (e.g., mean or median age, age range), authors were contacted using the available email addresses from the articles; seven responded with information or clarification.

Results

Data Collection and Sampling Methods

As shown in Table 1, the sample sizes of the MSM participants in the reported studies ranged from 41 to 2059 (with a median of 423). Most of the studies (29 out of 33) on MSM in China were published after 2005 and all but two studies collected data after 2000. Specifically, among the 31 studies that reported the year(s) of data collection, four were conducted in 2004, six in 2005, seven in 2006, and five in 2007 and beyond. Most of the studies (25 out of 33) were based on a one-time cross-sectional survey. Among the eight studies that collected serial cross-sectional data, five studies collected data at two time points [23–27] and three studies collected data at three time points [28–30]. All the studies solely focused on MSM populations except two, one of which compared MSM with non-MSM college students [31] and the other compared MBs with general male migrants [32].

Table 1.

Summary of research design and sample characteristics among MSM studies in China (in the order of year of data collection)

Reference No.	Author, year	Year of data collection	Location	Sampling method	Sample size	Age distribution (years)	% Married	% Migrant	% Gay-identified MSM
32	Zhang et al. 2000	1997–1998	N/A	N/A	N = 426	Mean: 28	Nearly one-third (no figure)	N/A	71.6
34	Wei et al. 2009	1999–2000	Nationwide	A subsample from a national survey (N = 1861)	N = 41	20–64	68.3	N/A	N/A
50	Lau et al. 2004	2001	Hong Kong	Telephone survey	N = 283	18–60	39.2	N/A	N/A
33	Wang et al. 2002	2001	Nationwide	Internet	N = 353	Mean: 24.5	13.3	N/A	69.6
17	Choi et al. 2004	2001–2002	Beijing	Venue-based and snowball	N = 482	Mean: 27; Range: 18–69	9	65	59
28	Zhang et al. 2007	2002	Harbin	Venue-based	N = 215	Median: 29; Range: 18–67	N/A	N/A	58.2
		2004	Harbin	Venue-based	N = 397	Median: 26; Range: 18–75	N/A	N/A	74.9
		2006	Harbin	Venue-based	N = 647	Median: 27; Range: 18–69	N/A	N/A	79.4
31	Cong et al. 2008	2003	Two universities in Zhejiang Province	All the students were asked to participate in the survey	Students: N = 1824; MSM: N = 68	N/A	N/A	N/A	N/A
48	Jiang et al. 2006	2003	Nanjing, Yangzhou, Changzhou, Wuxi, and Suzhou	Venue-based	N = 144	Range: 18–70	28.5	N/A	65.1
52	Lau et al. 2008	2003–2004; 2005–2006	Kunming	Venue-based, snowball, and internet	N = 896	Range: 15–75	18.2	N/A	N/A
39	Choi et al. 2007	2004–2005	Shanghai	Community outreach and snowball	N = 477	Mean: 28; Range: 18–56	13	78	57
51	Lau et al. 2008	2004–2005	Hong Kong	Internet and venue-based	N = 280	Mean: 24.5; Range: 13–53	2.85	N/A	46.4
23	He et al. 2008	2004	Guangzhou	Mass media, internet, and snowball	N = 201	(Mean: 29.5; Range: 16–66) 18–25: 10.5%;	25.9	62.2	59.7
		2006	Guangzhou	RDS	N = 423	18–25: 33.6%	N/A	86.1	63
29	Ma et al. 2007	2004	Beijing	RDS	N = 325	Mean: 29.6; Range: 16–61	9.3	62	56.6
		2005	Beijing	RDS	N = 427	Mean: 28.9; Range: 16–71	17.9	73.4	39.5
		2006	Beijing	RDS	N = 540	Mean: 28.4; Range: 18–60	29.2	82.7	43.2
45	Ruan et al. 2007	2005	Beijing	Internet, venue-based, community outreach, and snowball	N = 526	Mean: 26.2; Range: 17–54	6.5	64.3	Exclusive gay: 69.4
									Predominantly gay: 24.3
37	Xing et al. 2008	2005	Changsha	Venue-based and internet	Community: N = 270; Internet: N = 102	Mean: Community: 31; Internet: 25; Range: 14–64	Community: 31.7; Internet: 15.7	N/A	Community: 55.4; Internet: 57
24	Lau et al. 2009	2005	Kunming	Venue-based, snowball, and internet	N = 387	Range: 15–75	20.4	N/A	N/A
		2006	Kunming	Venue-based, snowball, and internet	N = 316	Range: 15–75	11.4	N/A	N/A
25	Li et al. 2008	2005–2006	Beijing	Internet, snowball, and venue-based	N = 1067	Mean: 31; Range: 19–54	N/A	N/A	69.4
46	Zhang et al. 2007	2005–2006	Beijing	Clinic	N = 753	Mean: 26.1; Median: 25; Range: 18–55	4.2	74.3	N/A
30	Feng et al. 2008	2005	Shenzhen	Venue-based	N = 242	Mean: 28.5; Range: 18–68	24	N/A	54
		2006	Shenzhen	Venue-based	N = 458
		2007	Shenzhen	Venue-based	N = 676
40	He et al. 2009	2006	Guangzhou	RDS	N = 423	Range: 16–62	17.1	83.7	63.0
44	He et al. 2007	2006	Shanghai	Snowball and RDS	Money boys (MBs): N = 239; General male migrants (GMM): N = 100	Mean: MBs: 25.2; GMM: 29.9	MBs: 10.5	100	MBs: 80.7
35	Zhang et al. 2007	2006	Nationwide (all 31 provinces)	Internet	N = 2059	Median: 25; Range: 18–64	16.3	N/A	63.8
36	Zhang et al. 2008	2006	Harbin	Internet and community outreach	Internet: N = 585; community: N = 642	Mean Internet: 25.6; community: 30.4	Internet: 15.8 Community: 23.5	N/A	Internet: 62.3 Community: 79.6
42	Ruan et al. 2009	2006–2007	Beijing	Snowball, community outreach, and internet	N = 541	Mean: 27; Range: 18–62	14.4	65.7	53.6
47	Guo et al. 2009	2006 and 2007	Nanjing and Yangzhou	Venue-based	Nanjing: N = 296; Yangzhou: N = 173	Nanjing Median: 36.64; Yangzhou: Median: 30.36	N/A	N/A	N/A
26	Feng et al. 2009	2006	Chongqing	Venue-based, community outreach, peer-recruitment, and internet	N = 1000	Median: 26.0; Range: 16–69	18.2	19.7	58.1
		2007	Chongqing	Venue-based, community outreach, peer-recruitment, and internet	N = 1044	Median: 24.0; Range: 15–65	22.7	15.7	62.2
41	Ruan et al. 2008	2007	Jinan	RDS	N = 428	Mean: 29.8; Median: 28.0; Range: 18–63	17.5	57.7	65.4
18	Liu et al. 2009	2007	Shenzhen	RDS	MBs = 58; Non-MBs = 293; Total N = 351	Range: 18–45	MBs: 7 non-MB: 25	N/A	N/A
38	Hong et al. 2009	2007	Shenzhen	Venue-based and clinic	N = 1146	Mean: 28.6	22.4	N/A	N/A
27	Ruan et al. 2009	2007	Jinan	RDS	N = 428	Median: 28.0; Range: 18–63	17.5	57.7	65.4
		2008	Jinan	RDS	N = 500	Median: 26.0; Range: 18–78	28.5	62	46.4
43	Zhong et al. 2009	2008	Guangzhou	RDS	N = 379	Range: 18–51	21.5	78.7	52.5
53	Gao & Wong. 2007	N/A	Chengdu	Venue-based, internet, clinics, and snowball	N = 160	Mean: 25.88; Range: 17–50	10	N/A	63.8
49	Mi et al. 2007	N/A	Beijing and Qingdao	Snowball and venue-based	MBs: N = 85	Median: 22; Range: 18–40	1	N/A	44

N/A data were not available at the study; RDS respondent-driven sampling; for RDS method, we reported adjusted rates

With respect to the geographic coverage of the studies, one study did not specify the location of data collection [33], three studies used samples (or subsamples) from nation-wide surveys [34–36], the others were based on regional samples that were recruited from 17 cities. Among these cities, there were three of the four municipalities that were directly under the jurisdiction of the central government, including Beijing (seven studies), Shanghai (two studies), and Chongqing (one study). The study areas also included Hong Kong Special Administrative Region (two studies), Shenzhen Special Economic Zone (three studies), and several provincial capitals such as Harbin (two studies), Jinan (two studies), Nanjing (two studies), Changsha (one study), Guangzhou (three studies), Kunming (two studies), and Chengdu (one study). Other study sites included Qingdao in Shandong province (one study) and four cities (Yangzhou, Changzhou, Wuxi, and Suzhou) in Jiangsu province (one study).

One study did not provide information on sampling method [33]; 16 studies used one sampling method; seven used two sampling methods; seven used three methods, and two used at least four methods for recruitment. The most frequently used sampling methods included venue-based sampling (15 studies), Internet sampling (14 studies), social-network-based sampling including snowball sampling or long-chain referral (14 studies), and respondent-driven sampling (RDS, eight studies). Other sampling methods include community outreach (five studies), voluntary counseling and testing (VCT) clinic- or STD clinic-based sampling (three studies), newspaper and television advertisement (one study), and telephone survey (one study). Among the studies that employed a single sampling method, the most commonly used sampling method was RDS (six studies), followed by venue-based sampling (four studies).

Demographic Profile of MSM in China

Age

The age of sampled MSM in the studies ranged from 13 to 78 years. Of the reviewed studies, 18 reported mean age, eight reported median age, two reported both mean and median age, and nine reported neither. The mean age of all the sampled MSM in the studies, based on available data, was 28.1 years, and the median age was 26.5 years. Significant differences in age were found between online and community samples. MSM recruited from communities were usually older than those from the Internet, 30.4 versus 25.6 years in one study [37], and 31 versus 25 in another [38]. In addition, age differences were also found among MSM from different venues. Those recruited from saunas were usually older than from bars with an age difference of 10 years, 32.2 versus 22.5 years, in one study [39].

Marital Status

There was a large variation in the proportion of MSM currently being married in the reviewed studies, ranging from 1 to 68.3%, with a median of 17.5%. The variation was largely associated with the age range of the samples, with older MSM being more likely to be married. Studies found that MSM recruited from the Internet were less likely to be married than those recruited from communities. For example, one study reported that 15.7% of MSM from the Internet sample were married compared to 31.7% from the community sample [38]. In another study, 15.8% of the Internet sample was married compared to 23.5% of the community sample [37]. Among different venues, those who were recruited from saunas were much more likely to be married than those from bars, 39.6 versus 4.1% [39]. The patterns of variation in marital status among samples were therefore consistent with those of variation in age by sampling methods and across venues. Likewise, when comparing MBs with non-MB MSM, the former were less likely to be married than the latter (7 vs. 25%), as MBs tended to be younger than non-MB MSM [18].

Migratory Status

About one-third of the studies reported the proportion of migrants (i.e., those without a permanent local residence) in their samples [17, 23, 26, 29, 32, 40–44], including one study that solely focused on migrant MBs [32]. The reported proportions of migrant MSM in those none-migrant-exclusive samples ranged from 15.7 to 86.1%, with a median of 69.6%. The proportions of migrant MSM varied by the location of the studies. As big metropolitan cities attract more migrants than other cities, the proportions of migrants among MSM samples in these cities were higher than those in other smaller cities. The proportions of migrants in the MSM samples were, for example, 78% in Shanghai [40], 64.3–74.3% in Beijing [17, 43, 45, 46], and 62.2–86.1% in Guangzhou [23, 41, 44]. By contrast, migrant MSM accounted for 57.7% of all MSM in the survey in Jinan [27], and 15.7–19.6% in Chongqing [26]. An increasing proportion of migrant MSM over time was reported in some major metropolitan areas by authors who collected serial cross-sectional data over multiple years. One study in Guangzhou showed that 83.7% of the sampled MSM were migrants in 2006, compared to 62.2% in 2004 [23].

Sexual Orientation

Most of the studies reported sexual orientation of the sampled MSM and the percentages of self-identified gay ranged from 39.5 to 80.3%, with a median of 62.2%. There was no clear trend in proportions of gay-identified MSM by years of data collection, geographic locations of the studies, or sampling methods. The highest percentage of gay identification was reported among MBs [32]. Higher proportion of bisexual identification among MSM was reported from gay bars (70.2%) and STD clinics (68.8%) than from saunas (44.4%) [39].

HIV Prevalence

General Pattern

As shown in Table 2, more than half of the studies reported the HIV prevalence among MSM (19 out of 33 studies). Among these studies, the HIV infection rates ranged from 0 to 12.5%, with a median of 3.0%. All the studies that reported HIV prevalence used a biomarker (e.g., laboratory testing) for HIV infection, except one that employed self-report from the respondents [32]. Most MSM had never been tested for HIV and did not know their HIV status. For example, one study in Beijing revealed that 82% of MSM had never been tested for HIV before the study, and of the 15 participants (3.1%) who tested positive for HIV, 14 did not know their status before the study [17]. In another study on migrant MBs in Shanghai, a majority (58.2%) reported not knowing their HIV status [32].

Table 2.

Summary of HIV prevalence and HIV-related risk behaviors among MSM studies in China (in the order of year of data collection)

Reference No.	Author, year	Year of data collection	HIV prevalence	Unprotected anal sex with men in the last 6 months unless indicated otherwise	Unprotected sex with women in the last 6 months unless indicated otherwise	Commercial sex with men in the last 6 months unless indicated otherwise	Substance use (illicit drugs unless indicated otherwise)	Seek sexual partners online
32	Zhang et al. 2000	1997–1998	N/A	Nearly two-thirds never used condoms, only 2 consistently used condoms	N/A	N/A	N/A	N/A
34	Wei et al. 2009	1999–2000	N/A	Condom use w. primary partner: Never: 53.3%; Sometimes: 40.0%	93%	Sold sex in the last year: 5.0%; Bought sex in the last year: 17.1%	N/A	N/A
50	Lau et al. 2004	2001	N/A	60.0%	N/A	20.6%	N/A	Successful: 17.7%
33	Wang et al. 2002	2001	N/A	W. casual sex partner: Receptive UAI1: 51.9%; Insertive UAI: 48.9%	N/A	Bought sex: 14.2%	N/A	N/A
17	Choi et al. 2004	2001–2002	3.1%	49%	22%	Sold sex: 24%; Bought sex: 10%	N/A	N/A
28	Zhang et al. 2007	2002	1.3%	89.7%	Sex w. women: 41.9%	Sold sex: 21.3%; Bought sex: 22.6%	6.1%	N/A
		2004	0.9%	81.4%	Sex w. women: 24.7%	Sold sex: 7.5%; Bought sex: 11.2%	2.1%	N/A
		2006	2.2%	71.5%	Sex w. women: 25.0%	Sold sex: 5.6%; Bought sex: 11.4%	0.2%	N/A
31	Cong et al. 2008	2003	N/A	Rarely/never used condom in lifetime: 61.8%;	N/A	N/A	N/A	N/A
48	Jiang et al. 2006	2003	0%	In the past 3 months: 46.2%	Sex with men and women in the past 3 months: 41.3%	N/A	Alcohol use: Nondrinker: 22.6 %; Light drinker: 59.1%; Heavy drinker: 18.2%	N/A
52	Lau et al. 2008	2003–2004; 2005–2006	N/A	UAI: 54.7%;	71.9%	30.8%	N/A	N/A
39	Choi et al. 2007	2004–2005	1.47%	57%	19% (W. both men and women: 13%)	N/A	N/A	59%
51	Lau et al. 2008	2004–2005	N/A	W. regular partner: 66.0%; W. casual partner: 44.0%	N/A	4.3%	N/A	46.4%
23	He et al. 2008	2004	0%	(Total: 54.7%; W. regular partner: 61.5%; W. casual partner: 46.6%)	(UI2w. both men and women: 12.9%)	Sold sex: 1.0%; Bought sex: 4.5%	3.0%	38.8%
		2006	1.3%	N/A	N/A	Sold sex: 9.5%; Bought sex: 8.9%	5.4%	N/A
29	Ma et al. 2007	2004	0.4%	Insertive UAI: 51.8%; Receptive UAI: 44.8%	N/A	N/A	N/A	Successful: 33.4%
		2005	4.6%	Insertive UAI: 55.3%; Receptive UAI: 55.1%	N/A	N/A	N/A	Successful: 33.2%
		2006	5.8%	Insertive UAI: 49.8%; Receptive UAI: 42.1%	N/A	N/A	N/A	Successful: 32.8%
45	Ruan et al. 2007	2005	3.2%	77.6%; (W. regular partner: 78.6%; W. casual partner: 63.2%)	(W. regular partner : 5.5%; W. casual partner: 2.8%)	Sold sex: 8.6%; Bought sex: 4.2%	In the past 6 months: 2.5%	72.6%
37	Xing et al. 2008	2005	N/A	Last anal sex: community: 34.8%; Internet: 58.1%	N/A	N/A	N/A	N/A
24	Lau et al. 2009	2005	N/A	55.1%	Last sex episode: 66.0%	12.4%	N/A	N/A
		2006	N/A	48.9%	51.6%	20.6%	N/A	N/A
25	Li et al. 2008	2005–2006	4.0%	Insertive UAI: w. steady partners: 36.0%; w. casual partners: 26.4%; Receptive UAI: w. steady partners: 36.2%; w. casual partners: 33.3%	N/A	10.7%	N/A	N/A
46	Zhang et al. 2007	2005–2006	2.1%	Insertive UAI: 40.2%; Receptive UAI: 40.1%	N/A	N/A	N/A	N/A
30	Feng et al. 2008	2005	1.7%	N/A	N/A	N/A	N/A	N/A
		2006	2.6%	N/A	N/A	N/A	N/A	N/A
		2007	3.8%	N/A	N/A	N/A	N/A	N/A
40	He et al. 2009	2006	1.3%	48.0%	14.3%	1.9%	5.4%	Successful: 28.3%
44	He et al. 2007	2006	MBs: 3%	UI w. steady sex partner in the past year: MBs: 66.5%; GMM: 75%	N/A	MBs3: Sold sex in the past 30 days: 84.9%; in the past year: 98.7%; Sold UI in the past 30 days: 35.6%; in the past year: 50.2% GMM: Bought sex: 17%; UI: 53.8%	UI under the influence of alcohol: MBs: 34%; GMM: 34%; (UI under the influence of drugs, MBs: 4.60%)4	N/A
35	Zhang et al. 2009	2006	N/A	56.2%	N/A	Sold sex: 5.8%; Bought sex: 7.8%	1.9%	86.1%
36	Zhang et al. 2008	2006	N/A	Insertive UAI: Internet: 78.3%; Community: 64.1% Receptive UAI: Internet: 79%; Community: 71.7%	Internet: 74%; community: 100%	Sold sex Internet: 8.1%; Community: 10.7%; Bought sex: Internet: 6.0%; Community: 5.6%	Internet: 1.6%; Community: 1.7%	Internet: 88.4%; community: 60.4%
42	Ruan et al. 2009	2006–2007	4.8%	In the past month, UAI with regular male sex partner: 21.3%; With casual male sex partner: 14.6%	7.0%;	6.8%	Drugs: 0.7%; Alcohol ≥ Once per week: 20.7%; Three-month drinking rate: 42.1%	72.6%
47	Guo et al. 2009	2006 and 2007	Nanjing: 4.7%; Yangzhou: 8.0%	N/A	N/A	N/A	N/A	N/A
26	Feng et al. 2009	2006	10.4%	68.2%	76.4%	Sold sex: 5.5%; Bought sex: 5.1%;	6.5%	N/A
		2007	12.5%	63.8%	77.5%	Sold sex: 8.2%; Bought sex: 5.7%	2.9%	N/A
41	Ruan et al. 2008	2007	0.5%	61.4%	N/A	Sold sex: 21.0%; Bought sex: 10.4%;	2.1%	47.9%
18	Liu et al. 2009	2007	N/A	MBs: 45%; Non-MBs: 43%	MB: 64%; Non-MB: 69%	Bought sex MBs:16%; Non-MBs: 9%	MBs: 21%; Non-MBs:12%	N/A
38	Hong et al. 2009	2007	Sauna: 3.6%; Bar: 0.8%; Clinic: 4.2%	N/A	N/A	N/A	N/A	N/A
27	Ruan et al. 2009	2007	0.05%	Unmarried men: 66.1%; married men: 40.8%	N/A	N/A	Unmarried: 2.4%; Married: 1.0%	N/A
		2008	3.1%	Unmarried men: 36.8%; married men: 26.5%	N/A	N/A	Unmarried: 0.1%; Married: 0.3%	N/A
43	Zhong et al. 2009	2008	5.2%	UAI: 60.3%	19.0%	N/A	N/A	68.0%
53	Gao & Wang. 2007	N/A	N/A	Always used a condom for anal sex with regular sexual partner: 3.2%	Always used a condom with regular sexual partner: 5.6%	N/A	N/A	N/A
49	Mi et al. 2007	N/A	N/A	N/A	N/A	UI in commercial sex in the past 3 months: Oral sex: 66%; Receptive anal sex: 16%; Insertive anal sex: 12%; Vaginal sex: 42%	44%	N/A

¹UAI unprotected anal intercourse;

²UI unprotected intercourse;

³MBs money boys;

⁴Information in parentheses was extracted from other article(s) based on the same dataset

Geographic Variation

The rates of HIV infection varied significantly across geographic locations. In 2007, for example, the reported HIV prevalence ranged from 0.05 to 12.5% [26, 27]. The highest HIV infection rate (12.5%) among MSM was reported in Chongqing in 2007 [26]. The most recently reported HIV infection rates were 8.0% in Yangzhou during 2006–2007 [47], 4.8–5.8% in Beijing during 2006–2007 [25, 29, 45], 5.2% in Guangzhou in 2008 [44], and 4.7% in Nanjing during 2006–2007 [47].

Time Trend

The studies reviewed in this article clearly showed an increasing trend of HIV prevalence over time among MSM in various cities in China, especially in recent years. One study in Beijing, based on three RDS samples, reported an increase of HIV prevalence from 0.4% in 2004 to 4.6% in 2005, and then to 5.8% in 2006 [29]. Similarly, another study in Beijing reported an increasing HIV prevalence from 3.2% in 2005 to 4.8% in 2006 [25]. In Guangzhou, there was no reported case of HIV infection among MSM in 2004 [23], but the rate was 1.3% in 2006 [23], and rose to 5.2% in 2008 [44]. In Shenzhen, three venue-based sampling surveys revealed that HIV prevalence was 1.7% in 2005, 2.6% in 2006, and 3.8% in 2007 [30]. In Jinan, the capital city of Shandong province, HIV prevalence was 0.05% in 2007 and 3.1% in 2008 based on two RDS surveys [27]. In Chongqing, a study based on a combination of several sampling methods reported HIV prevalence from 10.4% in 2006 to 12.5% in 2007 [26].

Variations Across Venues and by Migratory Status

Noticeable variations in HIV prevalence were found across venues and by migratory status (with or without a local residency). Studies reported that MSM recruited from gay bars were much less likely to be infected with HIV than those recruited from saunas and STD clinics [26, 39]. A study in Shenzhen revealed that the HIV infection rate was 0.8% for MSM recruited from gay bars, 3.6% from saunas and 4.2% from clinics [39]. A study in Chongqing in 2007 revealed a similar pattern but with a higher HIV prevalence; the HIV infection rate was 26.5% among MSM recruited from bathhouses and saunas compared to 10.3% among those from gay bars [26]. Regarding the migratory status of the respondents, migrant MSM usually had relatively higher rates of HIV infection than non-migrant MSM. A recent study in Jinan reported a 4.3% HIV infection rate among migrant MSM compared to 1.1% among non-migrant MSM in 2008 [27].

Unprotected Sex

General Pattern

About 90% (29 out of 33 studies) of the studies reported unprotected sex (e.g., having sex without using a condom) among MSM. Among these studies, 14.6–99.5% (with a median of 55.1%) of MSM had unprotected anal sex with men. Fourteen studies reported unprotected sex with women and such rates ranged from 2.8 to 100% (with a median of 64%).

The proportion of MSM having unprotected sex has gradually declined over time in China, even though the rates of unprotected sex among MSM are still high. Only two out of 426 (0.5%) respondents consistently used condoms in one study conducted between 1997 and 1998 [33], and 6.7% reported consistent condom use in another study between 1999 and 2000 [35]. A study in Harbin documented a decrease in the rate of unprotected anal sex in the past 6 months among MSM over time: 89.7% in 2002, 81.4% in 2004, and 71.5% in 2006 [28]. High rates of unprotected anal sex were also reported in other places, including 68.2% in Chongqing in 2006 and 63.8% in 2007 [26], 61.4% in Jinan in 2007 [42], 60.3% in Guangzhou in 2008 [44], 55.1% in 2005 and 48.9% in 2006 in Kunming [24], 57% in Shanghai during 2004–2005 [40], and 40.1% in Beijing during 2005–2006 [46].

Regular and Casual Partners

A number of studies documented differences in unprotected sex between regular partners and casual partners among MSM. Consistent with findings from other high-risk populations such as female sex workers [5], MSM usually had higher rates of unprotected sex with regular partners than with casual partners. A study in Beijing reported that the rate of unprotected sex with regular partners in the past 6 months was 78.6% compared to 63.2% with casual partners [45]. Another study in Beijing found that the rate of unprotected anal sex in the past month was 21.3% with regular partners and 14.6% with casual partners [42].

Insertive and Receptive Anal Sex

The existing literature revealed mixed findings regarding rates of condom use in receptive or insertive anal sex. One study reported higher rates of unprotected insertive anal sex than receptive anal sex in 2004 (51.8 vs. 44.8%) and in 2006 (49.8 vs. 42.1%), but similar rates in 2005 (55.3 vs. 55.1%) [29]. Another study reported similar rates of unprotected insertive and receptive anal sex (40.2 vs. 40.1%) [46]. When further stratified with types of sexual partners, one study reported similar rates of unprotected insertive and receptive anal sex (36.0 vs. 36.2%) with steady partners, but higher rates of unprotected receptive anal sex than insertive anal sex with casual partners (33.3 vs. 26.4%) [25]. One study reported similar rates of unprotected insertive and receptive anal sex among Internet-based sample (78.3 vs. 79%), but higher rate of unprotected receptive anal sex than that of insertive anal sex among community-based sample (71.7 vs. 64.1%) [26].

Other HIV-Related Risk Behaviors

Commercial Sex

Existing studies reported that 1.0–30.8% (with a median of 10%) of non-MB MSM had commercial sex with men, either selling or buying sex from men or doing both. A survey of MBs in Shanghai reported that most MBs (84.9%) had sold sex in the past 30 days, and almost all (98.7%) had sold sex in the past year [32]. In addition, a relatively high proportion of these MBs engaged in unprotected sex when they provided commercial sexual service to other men, 35.6% in the past 30 days and 50.2% in the past year [32]. Two studies reported commercial sex with women among MSM. Based on a survey of 423 MSM in Guangzhou in 2006, a study reported that 5.2% of the respondents had unprotected intercourse with female sex workers [41]. One study in Shenzhen found that MBs were much more likely than non-MB MSM to provide sexual service to women (14 vs. 4%) [18].

Concurrent Relationships with Both Men and Women

The reviewed studies suggested that MSM in China had a high rate of concurrent sexual relationships with both men and women. A study in five cities in Jiangsu province in 2003 reported that 41.3% of the MSM had sex with both men and women [48]. Another study in Harbin reported that the rates of concurrent sexual relationships with both men and women in the past 3 months were 41.9% in 2002, 24.7% in 2004, and 25.0% in 2006 [28]. In a 2004 study in Guangzhou, 12.9% of MSM had unprotected sex with both men and women [23]. In addition, 5.5–100% (with a median of 64%) of the MSM reported unprotected sex with women in the last 6 months. One study in Harbin reported 100% unprotected sex with women among a community-based MSM sample in 2006 [37].

Substance Use

With respect to substance use, 0.1–44% (with a median of 2.4%) of MSM reported having used illicit drugs. Drug use is more common among MBs than among non-MB MSM. The study in Shenzhen reported that 21% of MBs had used drugs, compared to 12% among non-MB MSM [18]. Another study in Beijing and Qingdao among 85 MBs reported that 44% of the respondents had used illicit drugs [49]. Alcohol was also frequently used among MSM. A study in five cities in Jiangsu province in 2003 reported that 59.1% were light drinkers, and 18.2% of MSM were heavy drinkers, defined as those who frequently consumed five or more drinks during a single occasion in the past 3 months [48].

Seeking Sexual Partners Online

With increasing access to the Internet in China, the Internet has become a popular means of seeking sexual partners among MSM. One-third of the studies (11 studies) reported MSM seeking sexual partners online. Specifically, 28.3–88.4% (with a median of 53.5%) of the sampled MSM used the Internet to seek sexual partners. MSM were more likely to seek sexual partners online in big metropolitan cities than in smaller cities. For example, one study in Beijing during 2006 and 2007 found that 72.6% of MSM used the Internet to seek sexual partners whereas another study in Jinan in 2007 reported a rate of 47.9% [42, 43]. Two RDS-based studies reported that the proportion of MSM using the Internet to seek sexual partners rose from 28.3% in 2006 [41] to 68% in 2008 [44] in Guangzhou. In addition, a study showed that MSM recruited from the Internet were more likely to seek sexual partners online than those recruited from communities (88.4 vs. 60.4%) [26]. High rate of Internet use (86.1%) to seek sexual partners among MSM was also reported from a national online survey in 2006 [36]. One study in Hong Kong reported that 17.7% of MSM had been successful in seeking a sexual partner via the Internet in the past 6 months [50].

Discussion

The existing literature has revealed “a profile of MSM” in the studies among MSM in China. The samples of MSM in the reviewed studies were mostly young (with a median age of 26 years), many of them were currently married (especially among older MSM), and a considerable proportion were migrants. A majority of MSM were self-identified as gay, but still a large proportion of MSM (nearly 40%) considered otherwise (bisexual or uncertain). A large and increasing proportion of MSM sought sexual partners online. The median rate of HIV infection among MSM found in the current review was 3.0%, slightly higher than the average rate (2.5%) reported by a recent meta-analysis [22]. Studies have shown an increasing trend of HIV prevalence among MSM in various cities in China over time. High-risk sexual behaviors were prevalent among MSM, including high rates of unprotected sex, concurrent relationships with both men and women, and commercial sex. The rates of illicit drug use among MSM were relatively low, but MBs were more likely to use illicit drugs than non-MB MSM.

Among the results, several findings on MSM in China need to be highlighted. First, the increasing prevalence of HIV infection among MSM poses a great public health concern in China. Increased surveillance and intervention efforts (both primary and secondary interventions) are needed to curtail the increase of HIV infection among this most-at-risk population. Second, MSM’s high rates of concurrent relationship with both men and women, coupled with their high rates of unprotected sex, indicate a great potential of spreading HIV/AIDS from this high-risk group to the general population and thus fueling up the scale of HIV infection in China. Similar concerns stem from the relatively higher marriage rates among Chinese MSM than their counterparts in the western countries and the social pressure in this collective culture for young MSM to get married in the future [18, 23, 51]. The potential bridging effect of MSM in connecting high and low risk populations deserves more attention in future research [23, 52]. Third, with increasing Internet access in China, there has been a growing proportion of MSM seeking sexual partners online, especially in big metropolitan or urban areas. This increasing use of the Internet to seek sexual partners among Chinese MSM provides both challenges and opportunities in HIV prevention intervention targeting MSM.

While the existing studies have provided valuable information on MSM in China regarding their risks of HIV infection and transmission, there are some limitations in the existing literature. First, most quantitative studies in the literature are epidemiological studies, and more socio-behavioral studies are needed in the future. For example, research is needed on sexual risk behaviors such as concurrent relationships with both men and women, inconsistent condom use, commercial sex, and seeking sexual partners online, as well as the underlying psychological (e.g., self-identified sexual orientation, perceived pressure, mental health), cultural (e.g., desired gender role), social (e.g., family and social network, discrimination), economic (e.g., income, migratory status), and structural (e.g., available healthcare services) factors related to these HIV risk behaviors.

Second, this review found that some existing studies did not report basic information on study design or key demographic characteristics of the study samples. The lack of these basic data made it difficult for readers to assess the appropriateness of the design and representativeness of the sample. We urge future studies to report basic information on research design including year of data collection, study area, and sampling method as well as basic demographic information such as age distribution, marital status, migratory status, and occupation of the sample. Third, few studies identified subgroups of MSM population (e.g., MBs, those who had sex with men only, those who also had sex with women). As those subgroups may differ substantially in their demographic characteristics and HIV risk behaviors, identifying and studying these subgroups may provide important data to inform the development and implementation of culturally effective interventions among MSM in China.

Fourth, we were able to identify only two intervention studies on Chinese MSM; each had a relatively short duration of follow-up (5 or 6 months). One of the studies was a venue-based intervention that employed a socially and culturally appropriate HIV/AIDS intervention (participatory communication) and had a positive effect on condom use with casual sexual partners [53]. The other study was an Internet-based intervention in Hong Kong that did not demonstrate significant reduction in respondents’ risk behaviors (e.g., rate of unprotected sex) at post-intervention follow-up [51]. More research is urgently needed to develop feasible and effective biomedical or behavioral interventions that specifically target MSM population. Future intervention studies should be guided by culturally and socially appropriate theoretical framework. In addition, future studies on Chinese MSM can either adapt the existing programs that have been shown to be effective in developed countries into the Chinese setting or develop new theory-based programs that are socially and culturally appropriate for the Chinese MSM. Although the online intervention program in Hong Kong did not show significant effect in reducing respondents’ risk behaviors [51], it should not preclude the potential of Internet-based intervention in reducing HIV-related sexual risk among MSM in China, given the increasing use of the Internet among this population. The feasibility and efficacy of both online and venue-based interventions need to be further evaluated with rigorous research methodology.

Fifth, the current review indicated a geographic concentration of existing research in major cities in China, with more than half of the studies (17 out of 33 studies) being conducted in five big metropolitan cities or special administrative or economic zones, including Beijing, Shanghai, Guangzhou, Hong Kong, and Shenzhen, few in smaller cities, and none in towns and rural areas. Because of the considerable gaps in the economic development and public exposure to issues such as HIV/AIDS and homosexuality among big metropolitan cities, smaller cities, towns, and rural areas, it is likely that the societal environment (e.g., public attitude toward MSM, health facilities and services) and behaviors of MSM in big metropolitan areas may be different from those in other areas. Thus, studies in other areas, including rural areas, are needed to understand the particular challenges and risks of MSM in these contexts compared to those in major urban centers.

Sixth, since MSM are still a largely hidden population in China, the representativeness of the MSM samples in the studies is a concern for the internal validity of the research. As shown in this review, MSM sampled using different methods differed significantly in many aspects of their demographic characteristics and HIV-related risk behaviors. In addition, no data were available regarding some segments of the MSM population in China such as those in military service and in prison. Given the hidden nature of the MSM population in China and the lack of census or other framework for random or probability sampling, future research needs to use a combination of multiple sampling methods to reach MSM from diverse backgrounds and in different social segments to improve the representativeness of the study samples.

Seventh, special attention should be paid to the measurements of sexual risk behaviors. The variations in measurements might have contributed to the large variations in rates of sexual risk behaviors across studies. For example, in measuring the rates of unprotected sex with women among MSM, studies used different denominators (e.g., a subgroup of MSM who had sex with women or the entire sample). This incongruent calculation might have accounted for a great variation in the range (2.8–100%) of the rates of unprotected sex with women among MSM. Another example of measurement variations was the different time-periods of the measurements (e.g., last year, the past 6 months, the past month, the last three episodes, or the last episode), which made the comparison across studies difficult, if not impossible. Future studies need to develop and use a set of common measurements that best describe the typical risk behaviors among MSM and thus enable meaningful comparisons across studies.

The strengths of this review include a comprehensive and synthesized review with a large body of the existing literature and a focus on research methodology, demographic characteristics of MSM, their HIV prevalence, and HIV-related sexual behaviors. However, this literature review is subject to several limitations. First, we did not include unpublished studies and reports either in Chinese or English, which may constitute a large part of the literature related to the topic of this review. Second, we did not perform a meta-analysis of HIV infection because of the potential overlap with the efforts of a recent meta-analysis [22].

In summary, the continuously increasing trend of HIV infections among MSM in China has captured the attention from academia, international organizations and private donors, as well as the Chinese government. However, the documented sexual risk behaviors (e.g., inconsistent condom use with both men and women, commercial sex) are still prevalent among MSM. A critical concern about this most-at-risk group is not only the high HIV prevalence in themselves, but also the potential bridging effect of this group that facilitates the spread of HIV infections from high-risk populations to the general population. Future studies also need to target diverse MSM populations, including those with different behavioral patterns (e.g., having sex with men only, having sex with both men and women), those from various backgrounds and in different social segments, those in rural areas, township, or smaller cities, and those in military service and in prison. Finally, more risk-reduction intervention efforts targeting MSM are in urgent need and these intervention efforts should be guided by culturally and socially appropriate theoretical framework and be responsive to unique demographic characteristics and risk profiles of different subgroups of MSM in China (e.g., those who have sex with men only, those who have sex with both men and women, MBs, migrant MSM).