Our study supports the previous evidence that additive psychosocial health problems—collectively known as a syndemic—exist among urban MSM. In our study the interconnection of these psychosocial problems magnified the effects of the HIV/AIDS epidemic in this population. In addition, we found that this effect is at least partially mediated by high-risk sexual behavior. We found that not only is sexual risk behavior an important mechanism underlying the relationship between the syndemic of psychosocial problems and HIV infection for gay and bisexual men, but it also partially mediates the syndemic effects of psychosocial problems on HIV infection.
Given the cross-sectional nature of the study design, some results should be interpreted with caution; for instance, depression and substance-use may be the result of HIV-infection in some men rather than a cause. All relationships identified in this study are restricted to bidirectional relationships, not causality.
In the mediation analysis, the size of the syndemic effect was not attenuated to one after adjusting for sexual risk behaviors, which may be explained by incomplete adjustment and residual confounding by unmeasured sexual behaviors. Another possible explanation is that the syndemic of psychosocial problems acts as a cofactor independent of its effect on sex behaviors. If any cofactor effect exists, it is not strong, given the weak and statistically insignificant associations we observed once the effect of sexual behavior was removed by restriction analysis.
If cofactor effects indeed exist, this may be due to differences in sexual networks. MSM have few places to meet with each other. Gay bars are one of the most visible and accessible places to meet, but the omnipresent alcohol and sexually charged environment may contribute to substance use and risky sexual behaviors 
. The biological effects of psychosocial problems on susceptibility to illness may also explain any cofactor effects. Research conducted among people who are already infected has revealed the biological effects of psychosocial problems on HIV infection. Depression, stressful life events and trauma are associated with a decrease in CD4 T-lymphocytes, an increase in viral load, faster clinical decline, and higher AIDS-related and all-cause mortality 
. To the best of our knowledge, few studies, if any, have been conducted on the effect of psychosocial problems on biological markers of HIV susceptibility in high-risk populations.
Mediation analysis requires careful consideration of potential confounders of the relationships between a potential risk factor and the proposed mediator and between the mediator and the outcome 
. We adjusted for all demonstrated sexual risk factors in this population as well as other sexual risk behaviors that confer a lower risk, in accordance with the epidemiological literature. We also used restriction analysis to provide corroborative evidence for the results derived from the adjustment approach. We performed an analysis restricted to men who engaged in the highest risk behavior, thereby creating a condition for (possible) exposure to HIV 
. We believe that the results of this restriction analysis provide the strongest evidence for the mediation hypothesis.
Multivariate analyses suggest that depression and the experience of trauma show the strongest relationship to HIV infection a finding that is partially mediated by risky sexual behaviors. Our results are partially consistent with those from Mustanski's study 
, which showed that violence (known as PTSD risk factors in our study) has the strongest relationship with HIV. However, substance abuse was not significantly associated with HIV in our study in contrast to Mustanski's study. This difference may be explained by the fact that substance use is rare among MSM in China.
This is the first study to investigate the mediating role of sexual risk behaviors on the syndemic relationship between psychosocial problems and HIV infection. The finding that the MSM population experiences multiple, concurrent psychosocial problems that interact to form a syndemic helps to explain why many patients in STD clinics find it difficult to change their sexual behaviors. It may be difficult for people to focus on sexual behavior changes when they are faced with numerous other daily problems and competing life priorities. Health care providers should be prepared to address MSM who experience a range of mental health, substance use and behavioral problems.
The data presented here should be interpreted with the following limitations in mind. First, the cross-sectional design cannot identify the temporal order between exposure and outcomes, which limits our ability to make causal inferences about the relationship between independent variables and HIV infection. Therefore, in this study we can only identify an association and not causality. Additionally, our sample is not a random sample, although it is believed to be representative of the local population of MSM. Generalizing the results to a more diverse MSM population should be conducted with caution. Third, because the duration of HIV infection of the men in our study is not known, we do not know if their sexual behavior in the last six months is relevant to their HIV infection. Finally, recall bias and social desirability bias are possible because the information on drug use and sexual behavior was collected by self-report. The majority of the information on sexual behaviors in the last 6 months was used in the data collection.
In sum, intimate partner violence, childhood sexual abuse, alcohol use, drug use, depression and high-risk sexual behavior frequently coexist among gay and bisexual men, and experiencing more of these psychosocial health problems was associated with greater odds of being infected with HIV. Thus, our data support the development of new sexual risk reduction interventions and the refinement of existing ones that MSM can incorporate using multi-faceted approaches, including mental health treatment and psychosocial intervention. Additional studies, such as case-control studies and cohort studies, should be performed to demonstrate the importance of syndemics in the acquisition of HIV among the MSM population.