HIV seroconversion and types of relationships among men who have sex with men: A cohort study in China

Abstract

Background:

The role of partner types in modifying HIV seroconversion risk among men who have sex with men (MSM) is complex. We sought to understand the role of regular sexual partners and risky sexual behavior in contributing to incident HIV infection among MSM.

Methods:

From July 2011 to August 2017, we recruited HIV-negative men who reported having anal or oral homosexual encounters in the past 6 months, following them every 3 months for two visits. We collected sociodemographic and behavioral characteristics by self-administrated questionnaire. HIV status was confirmed by sequential rapid test and confirmatory test. We used multivariable Cox regression to identify risk factors and interaction models to evaluate the relative excess risk between relevant variables.

Results:

Among 1218 participants, HIV seroconversion rate was 3.66/100 person-years. HIV seroconversion was associated with lower educational attainment (adjusted hazards ratio [aHR]=1.73, 95%CI: 1.06–2.81), having had penetrative sex with male before age of 18 (aHR=2.44, 95%CI: 1.20–4.99), not using condoms in the last sexual encounter (aHR=2.19, 95%CI: 1.29–3.71), having regular but not committed partners (aHR=3.33, 95%CI: 1.77–6.93). Among 890 (73%) of men reported having regular partners, HIV seroconversion was more frequent in men whose stable partners were not committed as boyfriends (aHR=3.31, 95%CI: 1.73–6.36) and in men having unprotected anal sex (aHR=2.61, 95%CI: 1.42–4.80). Interaction between these two factors was observed (relative excess risk of interaction=4.53).

Conclusions:

Incidence among MSM in China was high; unprotected sex with steady, but not committed partners was associated with increased seroconversion risk. It is imperative to expand safer sex education and training for MSM to reduce unsafe sexual behaviors, including awareness that casual partners are not the only source of infection.

Introduction

HIV prevalence among men who have sex with men (MSM) in China has been rising steadily from ≈1% in 2003 to 8% in 2015¹. MSM, as a sexual minority, experience social stigma that impedes efforts to mitigate HIV-related risk behavior. Unprotected anal intercourse (UAI) is the most prominent risk factor^2,3. While casual partners are risky, the role of partner types in modifying HIV seroconversion risk remains complex. MSM who have stable relationships may have fewer outside partners and, perceiving strong peer social support, may be less likely to engage in UAI⁴. Stable male-to-male sexual partnerships reduce likelihood of HIV acquisition⁵, likely due to elements of mutual trust and loyalty with regular male partners⁶. However, some studies suggest that regular partners could paradoxically increase possibility of UAI^7,8 if the relationship is not mutually monogamous^7,9. MSM in China can be socially ostracized and stigmatized, leading to shame and avoidance of open public engagement. MSM who are not in socially cohesive communities may be more likely to engage in high-risk behaviors including having UAI with casual sex partners¹⁰. A Hong Kong study reported that more than 40 percent of MSM had unprotected anal sex in last six months regardless of whether or not they had a regular sex partner¹¹. In China, new HIV infections among MSM with UAI often are attributed to regular male sex partners; a meta-analysis suggested that the prevalence of UAI with regular sex partners and casual male sex partners among MSM in China was 45% and 33% respectively¹². UAI occurring more frequently within the context of regular relationships has been reported in New Zealand¹³.

Most Chinese studies have been cross-sectional, limiting the causal relationships that can be inferred. When recently diagnosed, HIV seropositive MSM typically reduce their high risk sexual behaviors, at least for a number of months, diluting or otherwise modifying the association between risk behaviors and HIV infection¹⁴. To disaggregate complex dynamics, large longitudinal studies are needed. In a 6-year cohort study, we examined the relationship between sexual behavior with regular sex partners and HIV seroconversion among MSM. We tested a prior hypothesis that unsafe sexual behaviors within regular relationships are associated with HIV acquisition among MSM if the relationship is steady but not involving firm mutual commitment.

Methods

Study Design and Setting

This prospective cohort study was conducted among MSM in Beijing, China. From July 2011 to August 2017, recruitment and follow-up were carried out in the HIV Voluntary Counseling & Testing Clinic of Beijing Jingcheng Venereal Hospital. Each year from 2011 to 2016, a new cohort of participants was recruited. Participants were followed up every 3 months for two visits over at least six months. Once the participants had completed a recruitment and 6-month follow-up cycle, they could be enrolled again in a new round of recruitment with a new identity code. After obtaining informed consent from participants, we used cell phone or QQ (internet-based social media) numbers to link their identity codes with data from the baseline or follow-up surveys. We used birthdates to resolve rare inconsistencies.

Participants

Our study used non-probability sampling methods to recruit eligible participants with two strategies, recruitment from voluntary counseling and testing (VCT) clinic services of Beijing Jingcheng Venereal Disease Hospital or via peer recommendation. We screened for study eligibility, including: not known to be HIV-seropositive; biological male at birth; ≥ 18 years of age; self-reporting anal or oral sex with a male (past 6 months); providing ≥2 contacts (e.g., cell phone, QQ number); agreeing to HIV rapid testing and self-administrated questionnaires; ability to read and sign an informed consent; and sobriety (i.e., no acute influence of alcohol or drugs). Ineligibility men were referred to VCT. We excluded participants who were HIV positive at baseline (i.e., they reported seronegative or unknown HIV status at baseline, but were discovered to be HIV-infected upon testing), and those who reported having neither regular nor casual partners. Men were included in the analysis of this paper if they participated in at least 2 surveys in the 6 years period.

Each participant was paid 50 RMB (≈US$7.50) for travel compensation on their first visit and an additional 150 RMB (≈US$22.50) for the subsequent two visits, paid at the final 6-month follow-up visit. The participant was encouraged to take home a rapid HIV test for his regular partner, if any, and needed referrals for confirmation and care were facilitated.

In defining partner types, we characterized a regular partner as a male sexual partner in a stable relationship. Two types of regular partners were identified: boyfriends vs. stable partners but without a “boyfriend” commitment (i.e., “friends with benefits”, having stable sexual relationship but without emotional investment). Female spouses were not included in the “regular partner” definition. A casual partner was one with whom a transient sexual relationship was engaged; casual partners were often persons met by chance or hook-ups for sex, without relationship expectations and often anonymous. Follow-up visits collected the same data as at baseline without repeat collection of stable demographic data(Figure 1).

Figure 1:

Recruitment diagram of an HIV seronegative cohort of 1218 men who reported having had sex with men in Beijing, recruited from 2011–2016 and followed into 2017.

Data Collection

Trained health professionals facilitated a self-administrated structured questionnaire survey in a separate private room of the VCT clinic. We assigned each study participant a unique and confidential identification code for the questionnaire and blood samples. Baseline survey data included sociodemographic information (e.g., age, ethnicity, education, residence, income, and marital status), behavioral information (e.g., the number of male sex partners in the past 3 months, condom use during the last anal sex, ever unprotected anal intercourse over the past 3 months, and partner types), and HIV test result. We estimated seroconversion dates by using the midpoint between the first positive and last negative HIV test.

Laboratory tests

Samples were tested by ELISA-based rapid test (Intec Products, Inc., Xiamen, China), with positives confirmed by Western blot (HIV Blot 2.2 WB; Genelabs, Singapore). Testing methods and procedures were carried out in accordance with the test kit instructions by well-trained laboratory professionals.

Statistical Methods

We performed Cox regression using SAS® 9.4 for Windows software (SAS Institute Inc., Cary, NC, USA) to identify risk factors for HIV infection, with a focus on behavioral characteristics of MSM with regular partners and how they might be associated with HIV seroconversion. We included variables significantly associated with HIV seroconversion in univariate models (p<0.1, two-tailed) in the multivariable models. Using an additive effects model, we evaluated the qualitative and quantitative interactive effects of predictors for HIV seroconversion among MSM with regular partners. We reported interactions using the relative excess risk of interaction (RERI; $\text{RERI\hspace{0.17em}}=\text{\hspace{0.17em}HR\hspace{0.17em}}\left(\text{AB}\right)-\text{\hspace{0.17em}HR\hspace{0.17em}}\left(\text{A}\overline{\text{B}}\right)-\text{\hspace{0.17em}HR\hspace{0.17em}}\left(\overline{\text{A}}\text{B}\right)+1$), attributable proportions of interaction (API; $\text{API}\left(\text{AB}\right)=\frac{\text{RERI}}{\text{HR\hspace{0.17em}}\left(\text{AB}\right)}$), and synergy index (SI; $\text{SI\hspace{0.17em}}=\frac{\text{HR\hspace{0.17em}}\left(\text{AB}\right)-1}{\left|\text{HR\hspace{0.17em}}\left(\text{A}\overline{\text{B}}\right)-1|+|\text{\hspace{0.17em}HR\hspace{0.17em}}\left(\overline{\text{A}}\text{B}\right)-1\right|}$ HR(AB) is the hazard ratio of HIV seroconversion in the presence of both factors A and B, $\overline{\text{A}}$ and $\overline{\text{B}}$ represent the absence of two risk factors A or B, in the absence of interaction, both RERI and API(AB)=0 and SI=1¹⁵.

Ethical Considerations

The study protocol and informed consent were reviewed and approved by the ethics committee of the Chaoyang District Center for Disease Control and Prevention.

Results

The final cumulative participation rate in any baseline survey was 78.9% over six years (3153/3995). After we adjusted for those men who repeatedly participated in different rounds, there were 1218 unique HIV seronegative MSM in our cohort. Among them, 69 men seroconverted over an aggregate 1830 person-years of follow-up, for an HIV seroconversion rate of 5.7%. Given a 0.82 year median follow-up, there were 3.66 infections per 100 person-years. The median age was 30.3 years and 6.1% of MSM were from ethnic minorities, i.e., not of Han ethnicity. Fully 69.4% of MSM received ≥12 years of education while only 1.0% received ≤6 years of education. Three-quarters (74.1%) of MSM were unmarried and one-quarter (29.2%) were legal Beijing residents (i.e., with local hukou). Most MSM (85.1%) reported being employed with 64.3% reporting a salary of ≤5000 RMB (≈US$750) per month. Most participants (71.5%) self-identified as homosexual, while 24.6% self-identified as bisexual. (Table 1).

Table 1.

Sociodemographic characteristics of HIV seronegative men who have sex with men cohort in Beijing from 2011–2016 (N=1218^*)

Characteristics	n	%
Median age in years	30.3 (IQR=8)
Median follow-up in years	0.82 (IQR=1.6)
Ethnicity
Han	1144	93.9
Minority	74	6.1
Years of education
≤6	12	1.0
7–12	361	39.6
>12	845	69.4
Marital status
Unmarried	902	74.1
Married	188	15.4
Cohabiting with others	79	6.5
Divorced/widowed	49	4.0
Having residence permit (hukou) in
Beijing	355	29.2
Elsewhere	863	70.8
Years living in Beijing
<1	128	10.5
1–2	103	8.5
>2	987	81.0
Employed status
Employed	1037	85.1
Students	142	11.7
Unemployed	39	3.2
Monthly income (RMB)**
≤5000	783	64.3
>5000	435	35.7
Sexual orientation
Homosexual	871	71.5
Heterosexual	6	0.5
Bisexual	300	24.6
Unknown	41	3.4
Partners types
Boyfriend	527	43.3
Steady but not committed	363	29.8
Casual partners only	328	26.9

^*Cumulative number of men who were HIV seronegative in baseline. Numbers may not add to the total due to missing data. All data are from the latest survey.

^**In October 2011, 6.37 RMB per US$1.00 (5000 RMB=US$785), compared to the average closing price in 2017 of 6.74 RMB per US$1.00 and 5000 RMB=US$742).

Among 1218 MSM, 328 (27%) reported having only casual partners, and we observed a seroconversion rate of 4.14/100 person-years over 507.3 person-years of follow-up. Among 890 (73%) MSM who reported having either a steady boyfriend, we observed a seroconversion rate of 1.83/100 person-years over a total 764.8 person-years of follow-up, compared to 5.74/100 person-years over a total 557.9 person-years’ follow-up for MSM who had a steady partner, but without a strong commitment. The univariate Cox regression analysis suggested that MSM who seroconverted to HIV were more likely to have lower educational attainment (HR=2.01, 95%CI: 1.24–3.25), lower income (HR=1.78, 95%CI: 1.00–3.15), debut penetrative sex before the age of 18 (HR=2.06, 95%CI: 1.05–4.04), penetrative sex with a male before the age of 18 (HR=2.86, 95%CI: 1.46–5.61), no condom use during the last sexual encounter (HR=2.00, 95%CI: 1.19–3.37), and unprotected anal sex in last 3 months (HR=1.90, 95%CI: 1.17–3.09). The univariate Cox regression analysis also suggested that MSM with steady but not committed partners (HR=3.13, 95%CI:1.67–5.87) and MSM with casual partners only (HR=2.27, 95%CI:1.16–4.47) are more likely to HIV seroconvert (Table 2). In the multivariable model, HIV seroconversion was associated with lower educational attainment (aHR=1.73, 95%CI: 1.06–2.81), penetrative sex with a male before the age of 18 (aHR=2.44, 95%CI: 1.20–4.99), and no condom use in the last sexual encounter (aHR=2.19, 95%CI: 1.29–3.71). Considering partner type, having steady but not committed partners increased the risk of HIV seroconversion (aHR=3.33, 95%CI:1.77–6.93), but the difference between having boyfriend or having casual partners only was or marginal significance (p=0.052) (Table 2). Compared to MSM with boyfriends, MSM with steady but not committed partners (OR=3.53, 95%CI:3.15–5.72) and MSM with casual partners (OR=3.52, 95%CI:2.60–4.78) were more likely to have multiple partners.

Table 2.

Associations between HIV seroconversion and sociodemographic and behavioral characteristics (N=1218^*, n=67 seroconverters).

Factors	Persons who seroconverted	Follow-up years	Infection Rates (100 person-years)	Univariate analysis		Multivariable analysis
				Crude HR (95%CI)**	P value	Adjusted HR (95%CI)**	P value
Age
>25	45	1326.6	3.39	ref.
≤25	22	503.4	4.37	1.28 (0.77–2.14)	0.34
Ethnicity
Han	65	1729.7	3.76	ref.
Others	2	100.3	1.99	0.52 (0.13–2.15)	0.37
Years of education
≥12	35	1256.1	2.79	ref.		ref.
<12	32	573.9	5.58	2.01 (1.24–3.25)	0.004	1.73(1.06–2.81)	0.028
Marital status
Unmarried	49	1327.5	3.69	ref.
Others	18	502.5	3.58	0.98 (0.57–1.67)	0.93
Having residence permit in
Beijing	18	564.1	3.19	ref.
Others	49	1265.9	3.87	1.20 (0.70–2.06)	0.51
Years living in Beijing
<1	7	139.6	5.02	ref.
≥1	60	1690.4	3.55	0.74 (0.34–1.63)	0.45
Employed status
Employed	54	1594.9	3.39	ref.
Others***	13	235.1	5.53	1.59 (0.87–2.92)	0.13
Monthly income (RMB)
>5000	15	621.1	2.41	ref.
≤5000	52	1208.9	4.30	1.78(1.00–3.15)	0.051
Sexual orientation
Homosexual	46	1265.2	3.64	ref.
Others	21	564.8	3.72	1.03(0.62–1.73)	0.91
Main venue used to seek male sex partners
Others	15	415.5	3.61	ref.
Internet	52	1414.5	3.68	1.02(0.57–1.81)	0.95
Ever had HIV testing in last year
Yes	46	1302.9	3.53	ref.
No	21	527.1	3.98	1.11(0.66–1.86)	0.70
Knowing the result of HIV testing
Yes	46	1261.7	3.65	ref.
No	21	568.3	3.70	0.99(0.59–1.67)	0.98
Age in years of penetrative sexual debut
≥18	57	1688.6	3.38	ref.
<18	10	141.4	7.07	2.06 (1.05–4.04)	0.035
Sex of first sexual partner
Male	50	1333.5	3.75	ref.
Female	17	496.6	3.42	0.90(0.52–1.57)	0.71
Age of debut of penetrative sex with a male (years)
≥18	57	1725.7	3.30	ref.
<18	10	104.29	9.59	2.86 (1.46–5.61)	0.002	2.44(1.20–4.99)	0.014
Number of male partners over the past 3 months
1	33	928.68	3.55	ref.
≥2	34	774.89	4.39	1.24(0.77–2.00)	0.382
Condom use in last anal sex
Yes	45	1383.73	3.25	ref.		ref.
No	21	318.53	6.59	2.00 (1.19–3.37)	0.009	2.19 (1.29–3.71)	0.004
Ever unprotected anal intercourse over the past 3 months
No	28	983.29	2.85	ref.
Yes	39	719.49	5.42	1.90 (1.17–3.09)	0.010
Partners type
Boyfriend	14	764.79	1.99	ref.		ref.
Steady but not committed****	32	557.90	5.74	3.13(1.67–5.87)	<0.001	3.33(1.77–6.93)	<0.001
Casual partners only	21	507.29	4.14	2.27 (1.16–4.47)	0.017	1.97(0.99–3.92)	0.052

^*Cumulative number of men who reported having sexual partners and not lost to follow-up. Numbers may not add to the total due to missing data or skipped questions. All data are from the latest survey.

^**HR=hazard ratio, 95%CI=95% confidence interval

^***students or unemployed

^****”Friends with benefits” or “4N9” as it is known locally, mostly means having stable sexual relationship without emotionally invested.

Among 890 MSM reported having regular partners, MSM who had steady partners but without commitment were more likely to seroconvert than those who had boyfriends as regular partners (aHR=3.31, 95%CI: 1.73–6.36). Unprotected anal sex was associated with an increased risk of HIV seroconversion (aHR=2.61, 95%CI: 1.42–4.80) (Table 3).

Table 3.

Associations between HIV sero-conversion and regular sex partners’ behavioral characteristic among HIV seronegative MSM having regular sex partners over past 3 months. (N=890^*, n=46 seroconverters).

Factors	Persons who seroconverted	Follow-up years	Infection Rates (100 person-years)	Univariate analysis		Multivariable analysis
				Crude HR (95%CI)**	P value	Adjusted HR (95%CI)**	P value
Type of regular partners
Boyfriend	14	764.8	1.83	ref.		ref.
Steady but not committed***	32	557.9	5.74	3.16 (1.69–5.92)	0.001	3.31 (1.73–6.36)	<0.001
Number of regular partners over the past 3 months
1	31	1022.6	3.03	ref.
≥2	15	297.9	5.04	1.69 (0.91–3.13)	0.096
Having unprotected anal intercourse with regular partners over past 3 months
No	18	800.9	2.25	ref.		ref.
Yes	25	455.9	5.48	2.43 (1.33–4.46)	0.004	2.61 (1.42–4.80)	0.001

^*Cumulative number of men who reported having regular partners and not lost to follow-up. Numbers may not add to the total due to missing data or skipped questions. All data are from the latest survey.

^**HR=hazard ratio, 95%CI=95% confidence interval

^***”Friends with benefits” or “4N9” as it is known locally, mostly means having stable sexual relationship without emotionally invested.

In the additive interaction model, compared to those MSM who had a boyfriend as a regular sex partner and reported no unprotected anal sex, MSM who had other types of regular partners and unprotected anal sex were more likely to HIV seroconvert (HR=6.81, 95%CI: 2.86–16.20). The relative excess risk of interaction (RERI) was 4.53 (95%CI: 0.21–8.86), attributable proportions of interaction (API) was 0.67 (95%CI: 0.37–0.96), and synergy index (SI) was 4.55 (95%CI: 0.84–24.62) (Table 4).

Table 4.

Interaction between type of regular partner and unprotected anal sex on HIV seroconversion among MSM who reported having had regular sexual partners over the past 3 months. (N=890^*, n=46 seroconverters).

Factors		Persons who sero-converted	Follow-up years	Recent Infection rates (100 person-years)	Hazard ratio** (95% confidence interval)	P value
Type of Regular Partner	Unprotected Anal sex
Boyfriend	No	7	439.7	1.59	ref.
Boyfriend	Yes	6	279.2	2.15	1.34 (0.45–4.00)	0.597
no commitment***	No	11	361.2	3.05	1.94 (0.75–4.99)	0.172
no commitment***	Yes	19	176.7	10.76	6.81 (2.86–16.20)	<0.001

^*Cumulative number of men who reported having regular partners and not lost to follow-up. Numbers may not add to the total due to missing data or skipped questions. All data are from the latest survey.

^**HR=hazard ratio, 95%CI=95% confidence interval

^***”Friends with benefits” or “4N9” as it is known locally, mostly means having stable sexual relationship without emotionally invested

Discussion

In this cohort study among HIV seronegative MSM in Beijing over 6 years, we observed an overall HIV seroconversion rate of 3.66/100 person-years among all HIV seronegative participants, 1.83/100 person-year among MSM having a steady boyfriend, 5.74/100 person-year among men with other types of steady partners, and 4.14/100 person-year among MSM who had casual partners only. The first notable conclusion, therefore, is the high seroconversion rates among MSM in modern urban China. HIV seroconversion was most likely with steady but not committed partners and least likely with stable committed “boyfriend” relationships. Safer sex education should highlight that UAI and steady, but non-committed relationships (“friends with benefits” or “4N9”) represent high risks for HIV infection. Interaction between these two factors suggested a synergistically increased risk of HIV seroconversion, with 67% of the HIV seroconversions potentially attributed to UAI and non-committed relationships.

Our interactive model reinforces the potential role of HIV transmission of a regular partner, but who is not considered to be a boyfriend. A United States study estimated that most HIV transmissions (68%) of MSM came from principal partners⁹. Both a previous cohort study and a cross-sectional survey in China demonstrated that UAI with regular partners was a predictor of HIV seroconversion among MSM^7,16; neither prior study examined the role of a steady boyfriend vs. a steady, less committed partner. Given the US and Chinese studies, sexual concurrency may be catalyzing the transmission of HIV among MSM, even more than the number of partners¹⁷. It may be the case that early in an HIV epidemic (as in China), with relatively low background of HIV prevalence, that the number of sexual partners increases the probability of encountering an HIV positive one. In contrast, later when HIV is endemic at a high prevalence level among MSM, repetitive sex with the same man may increase risk of transmission, given that many more men will be HIV-infected.

Trust is a principal reason that MSM will not use condoms with regular partners¹⁸. The aforementioned Chinese cohort study found that about half of MSM had engaged in a sexual relationship with regular partners for <1 year, but nonetheless believed that their regular partners were loyal to them and cared about their wellbeing¹⁶. It is an immense challenge for health educators to find the correct educational messaging for MSM with steady relationships, namely that men should discuss openly that trust and condom and/or pre-exposure prophylaxis (PrEP) are compatible with steady relationships.

Along with feelings of loyalty and trust, desired intimacy and/or pleasure can lead MSM to eschew condom use with anal intercourse^19,20. In either context, PrEP is a valuable preventive tool. Though not protecting from non-HIV sexual transmitted infections (STIs), PrEP and post-exposure prophylaxis (PEP) both markedly reduce the risk of HIV infection when used correctly^21–23. It is complex to disentangle the relationship between the various types of regular sexual partners, casual partners, frequency of sexual encounters with each subgroup, and HIV seroconversion risk among MSM from a longitudinal and dynamic perspective. Even the best biomedical and behavior change interventions cannot succeed without ensuring that MSM can safely and confidently seek care and services, communicate openly about their sexual lives with partners and health providers, and be supported to adopt available preventive options²⁴.

Consistent with many previous studies, we found that lower educational attainment, penetrative sex with a male before the age of 18, and UAI were associated with higher risk of HIV seroconversion among MSM^25–29. Lower educational attainment typically accompanies low knowledge or awareness of the risk of HIV/AIDS and UAI^30,31; lower education is also a risk factor for other STIs such as syphilis³². Lower education is a socio-structural barrier to accessing HIV education, testing, and treatment services³³. Earlier homosexual debut was associated with risky sexual behavior, such as increased condomless sex and multiple sexual partners^34,35.

Knowledge does not guarantee changes in attitudes or adoption of safer sexual behaviors. Different segments of the MSM community may need diverse prevention options. Condoms are widely available in China, but their use is not promoted effectively on gay-oriented internet platforms (apps and dating websites) where persons meet sexual hook-ups or make “friends with benefits”. How to formulate a healthy and responsible culture among MSM community is a research imperative in China. Government, non-governmental organizations, and community-based organizations, and members of lesbian, gay, bisexual, and transgender communities share responsibility for promulgating safer sex education and availability of PrEP/PEP and condoms, particularly to people with lower educational attainment or who are more isolated from appropriate sexual education²⁴. The China Center for Disease Control and Prevention is expanding MSM-targeted education and availability of PrEP and PEP, combined with condom promotion. Couples voluntary counseling and testing (CVCT)³⁶ can play a vital role for MSM in China, providing prevention services that are agreed upon by steady MSM couples.

Strengths of our study include the multi-year, prospective study that represents a rolling cohort. Well trained study staff and state-of-the-art diagnostics enabled us to have substantial confidence in our data. At the same time, our study has limitations. First, with the exception of HIV seroconversion, all information was collected by self-report; social response bias may result in underreporting socially stigmatized behaviors. Second, we did not ask all of the salient questions. We only included behavioral characteristics of regular male partners in our model, while the behavioral characteristics of female or casual male partners can also impact the risk of HIV seroconversion. We would have improved the study by asking about serostatus of partners, and ascertaining whether seropositive partners were taking ART. We did not ascertain whether the “last occasion” of anal sex with or without condom was with a regular or casual partner. We failed to ask about the preferred/most frequent position of the participants in anal sex. Third, we studied MSM living in Beijing, a relatively prosperous city; findings may not generalize to MSM living elsewhere in China. The recruitment clinic was located in the north part of Beijing, not convenient for many men from other parts of the city, limiting generalizability to the rest of Beijing. Fourth, median follow-up was less than one year (Table 1).

Conclusions

In our observational MSM cohort, the HIV seroconversion rate was 3.66/100 person-years. Risk did not differ significantly between MSM with regular partners and MSM with casual partners only. An interaction observed between having non-committed regular partners and UAI suggested the etiology of increased HIV acquisition risk. Lower educational attainment, early homosexual debut, and UAI were associated with a higher risk of HIV seroconversion. We do not yet have robust strategies implemented to help reduce MSM sexual health risks with regular partners in China.