Background
China now faces a serious human HIV epidemic among men who have sex with men (MSM) [
1‐
4], and a large proportion of new HIV infections among this population are attributed to regular male sex partners (RP) [
5]. HIV prevalence among MSM in China grew from 1.4% to 7.7% from 2003 to 2014 [
4,
6,
7], and new HIV infection among MSM transmitted through RP increased from 34% to 40% from 2002 to 2010 [
5]. These statistics are similar to those of other countries. In the United States, 68% of HIV transmissions come from main partners, whereas only 32% come from casual male sex partners [
8]. In New Zealand, 40% of new infections are transmitted by RP, with 37% of these by casual male sex partners and 23% by commercial male sex partners [
9]. These statistics indicate that HIV risk behaviors occur more frequently within the context of regular or intimate relationships [
10,
11]. A meta-analysis showed that the prevalence of unprotected anal intercourse (UAI) with RP among MSM in mainland China was 53%, whereas it was 45% with casual male sex partners [
12]. In Hong Kong, the prevalence of UAI with RP among MSM is 60.2%, while it is 45.8% with casual male sex partners [
13]. In the United States, the frequency of condom use with boyfriends among MSM is the lowest (38.6%), then almost doubling with regular sex partners other than boyfriend (61%), and is the highest with casual male sex partners (74.9%) [
10]. An MSM couple-based study in the United States showed an 84% prevalence of UAI with RP, and a 66% prevalence of UAI with NRP [
14]. Without the possibility of legal marriage and lacking an accepting social environment, the regular or intimate relationship among MSM is vulnerable and fluid, and thus MSM who have RP will usually have other sexual relationships outside their regular relationships and will engage in concurrent UAI [
15‐
17]. These instances of concurrent UAI facilitate the spread of HIV among this population [
18‐
20]. While HIV prevention programs targeting MSM made considerable efforts to promote the use of condoms with sex partners in general, MSM with regular or intimate relationships are substantially understudied [
21‐
23].
The frequency of UAI within and outside intimate relationships among MSM is associated with the quality of the relationship [
11,
24,
25]. A U.S. cross-sectional study revealed that a sexual agreement that did not allow sex with casual male sex partners (OR = 0.05) and which participants valued (OR = 0.20) could reduce UAI with both RP and NRP [
26]. Another U.S. study showed that intimacy between RP could increase UAI within concordant negative relationships (OR = 1.03), but the length of the relationship (OR = 0.92) and the dependability of a partner (OR = 0.88) could reduce UAI within concordant negative relationships. This study also showed that greater attachment could increase UAI with RP both among concordant positive (OR = 1.09) and discordant (OR = 1.07) couples and open agreements that allow UAI with NRP could increase engaging in UAI with outside partners among men in concordant (OR = 9.08) and discordant (OR = 5.87) relationships [
25]. In Chinese Confucian culture, MSM couples are less accepted by society and family [
27]. Compared with other countries, then, MSM in China have fewer supports to maintain their intimate relationships, and MSM intimate relationships are therefore predictably more vulnerable and dynamic [
28]. Consequently, MSM frequently change their RP or engage in sex with NRP [
29]. Research investigating intimate relationships among this population is rare, particularly investigation into the association between intimate relationships and HIV-related risk behaviors which is now urgently warranted.
Interdependence theory (IDT) is a classic theory that describes how individuals interact. IDT can predict outcomes and is frequently applied to couple relationships [
30]. In this respect, the theory emphasizes the interaction within a dyad, and it has been applied in many social science researches [
24,
31,
32]. According to IDT, the behaviors of dyad can be infected by the two members’ dependence model, conflicts of interest, expectations and communications [
30]. The dependence model includes the level of dependence (the degree an individual relies on his or her interaction partner), mutuality of dependence (the degree to which two members are equally dependent on one another), and the basis of dependence (the way the dependence of two persons derives from partner control or joint control) [
30]. Conflicts of interest describe a situation in which the outcome can benefit member A while it may or may not benefit member B [
30]. Expectations in IDT mean the expectations of members in the relationship (whether the relationship will continue for a long time and the length of the relationship). Communications relates to members’ communicating with each other during the interaction, that is, partners communicating their relevant needs, goals, and motives to each other, which is also called information-seeking [
30]. These variables reflect the closeness of intimate relationships such as that of regular partners. Recently, interdependence theory has also been applied in a few MSM couples studies to investigate the risk factors of HIV-related risk behaviors [
24,
25]. For example, one study showed that both in seroconcordant and serodiscordant MSM couples, those with a higher level of relationship satisfaction and commitment were less likely to engage in UAI with outside partners [
24]. However, IDT-based research among MSM intimate relationships is limited, and more research is warranted.
The purpose of this study is to describe the intimate relationships of MSM based on interdependence theory, and to investigate the association between intimate relationship characteristics and HIV-related behaviors among MSM with RP in Guangzhou, China. Our hypothesis is that intimate relationship characteristics can influence UAI within and outside the intimate relationships among MSM.
Methods
Participants and recruitment
MSM participants were recruited during May and August of 2014 in an MSM peer-friendly HIV testing service center in Guangzhou, China. The center is a well-known LGBT community-based organization (Lingnan Fellow Health Support Center) and is co-operated by the Guangzhou Center for Disease Control and Prevention (Guangzhou CDC) [
33]. Almost 80% of HIV testing among MSM in Guangzhou was completed in this service center. Eligibility criteria included men who self-reported having anal intercourse with at least one RP in the past six months, and were age 18 years or older. Those who self-reported as HIV positive were excluded since, if MSM know their HIV positive or negative status, they would change their sexual behaviors and a positive HIV status would also influence their intimate relationships. Per our study objective, MSM without a known HIV positive status comprised the target population. By examining valid associations between intimate relationships and risk sexual behaviors in this population, we could develop appropriate interventions for MSM couples with HIV unknown or negative status. The exclusion of self-reported HIV positive MSM was shared by other studies [
26,
34,
35].
The convenience-sampling method was used in this study. All MSM who sought HIV testing services in this center were asked to enroll. Eligible participants were invited to a private room and informed that the questionnaire was anonymous. The questionnaire took 10 min to complete on average. Written informed consent was obtained by the experienced MSM peer staff in the center. Questionnaires were then self-administered, but participants could consult the staff if there was any confusion. The staff reviewed the questionnaire when it was completed. If there was any problem, such as missing answers or inconsistency, etc., the staff would confirm with the participant and ensure the quality of data collection.
Ethical considerations
Verbal and written consent were provided by all participants before commencement of questionnaire completion, and to keep absolute anonymity, written consent could be signed with a nickname. Participants could quit at any time before finishing the questionnaire. The study protocol was approved by the ethics review committee of the Guangzhou Center for Disease Control and Prevention.
Measures
Background characteristics
Socio-demographic characteristics (age, marital status, residence, duration of stay in Guangzhou, education, and income) and MSM-related information (sexual orientation, duration of being MSM, and recruitment venue) (Table
1) were collected in this study. Sex partnership information and UAI in the past six months were also obtained (Table
2).
Table 1
Demographics and MSM-related information among MSM who had regular male sex partners
Socio-demographic characteristics |
Age |
< 25 | 37.7 (153) |
≥ 25 | 62.3 (253) |
Currently married |
No | 83.3 (338) |
Yes | 16.7 (68) |
Guangzhou permanent resident |
No | 59.8 (243) |
Yes | 40.2 (163) |
Stayed in Guangzhou more than two years |
No | 21.2 (86) |
Yes | 78.8 (320) |
Higher than post-secondary education level |
No | 23.6 (96) |
Yes | 76.4 (310) |
Currently a student |
No | 83.0 (337) |
Yes | 17.0 (69) |
Monthly personal income (1000 RMB = 150 USD) |
< 4000 RMB (600 USD) | 47.5 (193) |
≥ 4000 RMB (600 USD) | 52.5 (213) |
MSM-related information |
Sexual orientation |
Bisexual/uncertain | 22.7 (92) |
Homosexual | 77.3 (314) |
Duration being MSM |
< 5 years | 45.1 (183) |
≥ 5 years | 54.9 (223) |
Male sex partners mainly recruited via |
Bar/dance hall/teahouse/club/bath/park/toilet/grassland/others | 11.8 (48) |
Internet/Dating apps | 88.2 (358) |
Table 2
HIV risk behaviors in the past six months among MSM who had regular male sex partners
Sex partnership information |
Had anal sex with multiple male sex partners |
No | 43.8 (178) |
Yes | 56.2 (228) |
Had casual male sex partner(s) |
No | 50.5 (205) |
Yes | 49.5 (201) |
Had commercial male sex partner(s) |
No | 96.3 (391) |
Yes | 3.7 (15) |
Had non-regular male sex partners (casual or commercial male sex partners) |
No | 50.2 (204) |
Yes | 49.8 (202) |
Condom use information in the past six months |
Had UAI with regular male sex partners |
No | 46.1 (187) |
Yes | 53.9 (219) |
Had UAI with NRP (casual or commercial male sex partners) |
No | 76.4 (310) |
Yes | 23.6 (96) |
Had concurrent UAI with both regular male sex partners and non-regular male sex partners |
No | 79.3 (322) |
Yes | 20.7 (84) |
Two types of sex partner in this study were defined: regular partner (RP) and non-regular partner (NRP). Regular partners were those the participant had sex with four times or more in the past six months, including “boyfriend,” “lover,” or “regular sex partner other than boyfriend” [
10,
18,
36]. Non-regular male sex partners included casual male sex partners and commercial male sex partners. Casual male sex partners were those the participant had sex with no more than three times and had no cash or kind payment, but were not considered regular partners [
10,
18,
36]. Commercial male sex partners were those who had sex with the participant by the payment of cash or kind [
36]. In this study, three kinds of UAI situations in the past six months were applied as the main outcomes, including UAI with RP, UAI with NRP, and concurrent UAI. Concurrent UAI in this study was defined as having UAI with both RP and NRP in the past six months, which is usually defined as risk behaviors among MSM in relationships [
18,
24,
37].
Intimate relationship characteristics
In accordance with different constructs of the interdependence theory [
30], we asked all participants about their relationship with their RP (Table
3), which included 1) Expectations for this relationship: whether they believed this was a serious relationship (with responses of “Defining as boyfriend by each other in this relationship,” “Having not yet defined as boyfriend by each other,” “Defining as regular sex partner other than boyfriend,” and “Defining as other kind of relationship.” Those who responded as “Defining as boyfriend by each other in this relationship” were classified as “Yes”; others were classified as “No.”); Length of the relationship (with responses of “less than three months,” “three to six months,” “six or more than six months and less than 12 months,” “one to three years,” “three years or more than three years.” Those who responded as “less than three months” and “three to six months” were classified as “< 6 months”; others were classified as “≥ 6 months”) and whether they perceived this relationship would continue for a long time (with responses of “Yes” or “No”); 2) Conflicts of interest: whether they exchanged money or materials to maintain this sexual relationship (with responses of “Yes” or “No”); 3) the dependence model: whether the participant invested more emotion in this relationship (with responses of “I invested more,” “He invested more,” “We invested the same”, “None of us invested emotion in this relationship.” Those who responded as “I invested more” were classified as “Yes”; others were classified as “No.”), and whether their RP was dominant in this relationship (with responses of “I was dominant,” “He was dominant,” “We have equal power in this relationship.” Those who responded as “He was dominant” was classified as “Yes”; others were classified as “No.”); 4) Open communication in sexual relationship (openly discussed being monogamous in this relationship (with responses of “Yes” or “No”), and openly discussed a sexual agreement or requirement to use a condom when having sex with other partners (with responses of “Yes” or “No”)). In addition, we collected information about the number of RP (those who had two or more RP in the past six months were classified as “has multiple RP”).
Table 3
The characteristics of the relationship with regular male sex partners
Had multiple RP in the past six months |
No | 67.5 (274) |
Yes | 32.5 (132) |
Expectation for this relationship |
Length of the relationship |
< 6 months | 52.2 (212) |
≥ 6 months | 47.8 (194) |
Believed this was a serious relationship |
No | 44.8 (182) |
Yes | 55.2 (224) |
Perceived that this relationship wound continue for a long time |
No | 20.7 (84) |
Yes | 79.3 (322) |
Conflict of interest |
Exchanged money or materials to maintain this sexual relationship |
No | 98.0 (398) |
Yes | 2.0 (8) |
The dependence model |
Invested more emotion in this regular relationship |
No | 65.3 (265) |
Yes | 34.7 (141) |
Your RP is dominant in this relationship |
No | 76.4 (310) |
Yes | 23.6 (96) |
Open communication on sexual relationship |
Openly discussed being monogamous in this relationship |
No | 60.8 (247) |
Yes | 39.2 (159) |
Openly discussed a sexual agreement or requirement to use condoms when having sex with other partners |
No | 47.5 (193) |
Yes | 52.5 (213) |
Statistical analysis
Except for description of data, we constructed univariate and multivariate logistic regression models to investigate the association between intimate relationship characteristics and UAI (UAI with RP, UAI with NRP and concurrent UAI) in the past six months, and we built multiple forward stepwise logistic regression models to confirm the association between intimate relationship characteristics and UAI in the past six months. A three-stage strategy was used to investigate our final objective.
First, univariate odds ratios (and 95% confidence intervals) were derived for associations between intimate relationship characteristics variables and the dependent variables (UAI with RP, UAI with NRP, and concurrent UAI in past six months).
In the second stage, after adjusting for the confounders, adjusted odds ratios (and 95% confidence interval) were applied to describe the association between intimate relationship characteristics and the dependent variables. Confounders adjusted in this stage and the final multiple forward stepwise logistic regression models were detected by universal directed acyclic graphs (DAGs). DAGs use causal diagrams to visualize the causal effects between exposures and outcomes and their use is gaining popularity in epidemiology and biostatistics [
38]. Based on the aforementioned selection process, age (as a continuous variable), marital status, education level, Guangzhou permanent residency, sexual orientation, duration being MSM, and recruitment venue were treated as confounders in all multivariate and multiple forward stepwise models. Similar data analysis methods can be seen in other studies [
34,
39,
40].
Last, after adjusting for the previously mentioned confounders, the independent intimate relationship characteristics variables in the previous stage were included in the final multiple forward stepwise logistic regression models. The multiple forward stepwise logistic regression model has also been used in another study [
11]. In all models, statistical significance was defined by
p value < 0.05. Data from this study was analyzed with SAS (SAS 9.1 for windows; SAS Institute Inc., NC).
Discussion
This is one of the first studies investigating the association between intimate relationship characteristics and HIV-related risk behaviors among MSM who have RP in China. This study showed that sexual risk behaviors happened frequently outside intimate relationships among MSM. Expectations for the relationship, conflicts of interest, and open communication on sexual behaviors with RP were associated with UAI within and outside the intimate relationship, whereas the dependence model was not associated with any high risk sexual behaviors.
First, our findings identified a high concurrent UAI prevalence within and outside intimate relationships among MSM. The data showed that 56.2% of MSM with RP had multiple male sex partners, which indicates that half of MSM had concurrent partnerships. The result was similar to that of other published research in China [
16], and the prevalence of concurrency is also similar to the data reported in the United States (45% to 63.2%) [
41‐
43]. Especially in this study, a quarter (23.6%) of MSM had UAI outside their regular relationship, and one fifth (20.7%) of MSM practiced the six-month period concurrent UAI. Concurrent UAI prevalence among MSM in China is relatively higher than that in the United States which is around 16% [
41,
44]. However, it should be noted that concurrent UAI in this study indicated a 6-month window, and that this study did not confirm the overlapping periods of UAI with more than one partner, which could include a few participants who practiced serial monogamy in the past six months such as having UAI with a casual partner/a new regular partner after breaking up with a boyfriend within the six month period. For this reason most likely, the rate was relatively higher than that in the United States. It is well known that concurrent UAI will contribute to HIV transmission. Our results suggested that MSM in China have a high possibility of becoming infected by their RP. A previous cohort study also showed that UAI with RP, but not UAI with casual male sex partners, was the predictor of HIV seroconversion among MSM [
45]. Trust is often mentioned as a reason for not using condoms with RP [
35,
46‐
48]. It has been estimated that in the United States 68% of HIV transmissions come from main partners, whereas only 32% come from casual male sex partners [
8]. Many MSM in China still think that UAI with a regular partner is safe, and they have not recognized the risk of UAI within regular relationships, even though, as we mentioned before, more HIV infection is transmitted by RP. Thus a great deal needs to be done to educate MSM on HIV-related risk behaviors related to their main partners and to make them aware of potential risks. Sexual intercourse involves at least two people, so the interpersonal process of an intimate relationship and a sexual-related discussion on behaviors among MSM couples warranted investigation to develop MSM couple-based interventions as a first and necessary step.
Second, the factors related to expectations for the intimate relationship were positively associated with risk-taking with RP. In contrast, they were negatively associated with risk-taking with NRP. These factors comprise the aspect of the perception of relationship quality and commitment, and they remained in the final models. These results were similar to findings in the United States [
24,
35], but this is the first such report in China. A mixed method study among male couples showed that these couples expressed love, trust, and commitment as the most frequent reasons for not using a condom with a partner [
49]. Stark et al. found that commitment was positively associated with UAI compared with not engaging in UAI with regular partners [
50]. Another cohort study among MSM couples revealed that with higher levels of positive relationship dynamics (e.g., commitment, satisfaction), RP were less likely to engage in UAI with NRP [
24]. It is reasonable to assume that MSM having a higher quality relationship and who value the commitment of the relationship would have less UAI outside the relationship and thus fewer risk behaviors. In a similar way, it is difficult for heterosexual couples generally to maintain a high quality relationship. However, compared with heterosexual couples, the relationship of male couples is particularly vulnerable and fluid in China due to the unacceptable social environment of a thousand-year-old traditional culture. Changing the social environment for MSM couples will take time. From the perspective of HIV prevention, coaching male couples on how to maintain a quality relationship would be a novel approach to prevent intra-dyadic and extra-dyadic HIV transmission among MSM.
Although open communication in a sexual relationship, including open discussion about being monogamous and a sexual agreement, did not remain in the final models, the results from the adjusted OR revealed these factors were positively associated with UAI within relationships and negatively associated with UAI outside relationships. Mitchell’s study [
26], which was conducted among HIV-negative gay couples in the United States, also found that MSM were less likely to have UAI outside their relationship if having a sexual agreement in place that did not allow sex outside the relationship. According to these findings, open communication about monogamy and a sexual agreement between partners may improve condom use with outside partners among MSM. Couple-based HIV counseling and testing could be considered as the appropriate intervention, providing the opportunity to discuss the sexual agreement under guidance of experienced counselors. This has been proven to be effective in the potential decrease of HIV incidence among heterosexual couples in Africa [
51,
52], and it has also proven to be acceptable and safe among MSM in America [
21,
23]. Since few studies in China have focused on this aspect of HIV prevention and because there is an apparent lack of awareness existing among MSM, the development of interventions to promote open communication skills among MSM couples, such as couple-based HIV testing and counseling, is needed in China.
Our findings suggested that compared with MSM whose intimate relationships had no monetary or materials exchanged with RP, MSM who had monetary or materials exchanged with RP were almost six times more likely to have UAI with NRP or have concurrent UAI. According to our literature review, few studies reported an association between HIV risk behaviors and tangibles exchanged within an intimate relationship. It is possible that two individuals are not in equal power or control within a relationship if there is an exchange of monetary or materials, and this exchange may undermine the quality of their relationship. Compared with subjective variables such as believing this was a serious relationship and the perception that the relationship would continue for a long time, etc., monetary or materials exchange might function as a sensitive indicator of the quality of relationships.
Another finding was that the dependence model was not associated with HIV-risk behaviors among MSM who had RP. Very few studies reported on the dependence model among this population. We found that only one study conducted among MSM couples reported that couples with a higher degree of emotional attachment were more likely to have UAI with their RP [
24]. The association between dependence models and risk behaviors among MSM who are in intimate relationships should be investigated. More research is required.
Limitations
The results from this study should be viewed in light of several limitations, many of which are shared by other studies in this area of research. Only one individual within the intimate relationship was recruited [
32,
34,
53]. Also, participants were convenience-sample recruited from an HIV testing service clinic, which is a recruitment method for a hidden population [
35,
54,
55]. In addition, all measures of sexual behaviors were based on self-reporting but efforts were made to minimize bias (e.g., anonymity, training). Our study also excluded those who self-reported being HIV positive MSM, and the potential confounding effect of this variable was not able be analyzed [
26,
34,
35]. Additionally, we used categorical variables to describe intimate relationships [
25,
35]; these variables were not as effective as scales, and IDT-related scales [
56,
57] will be considered in our future study. Another limitation is that, as mentioned above, the concurrent UAI in this study means participants had UAI with both RP and NRP in the past six months, and the overlapping periods of UAI with more than one partner was not confirmed. The rate of concurrent UAI might include those in serial monogamy within a 6-month window. Future studies should differentiate these two situations, even though the risk of transmitting HIV is similar. Finally, this is a cross-sectional study, and causal relationships cannot be inferred.