Study design and sampling methods
Our study team held a series of stakeholder meetings in China with gay community leaders, HIV prevention intervention staff at municipal Centers for Disease Control and Prevention (CDC), and antiretroviral and STI clinic providers to discuss potential models for sexual health services for Chinese MSM. Importantly, stakeholders posited that MSM who conceal their same-sex behavior may differ in the qualities they prioritize within sexual health services that they would like to access, compared to counterparts who have already disclosed their same-sex behavior within a health care setting.
To test this hypothesis, between February and May 2018, we conducted a cross-sectional study in four cities (Shanghai, Beijing, Changsha, and Guangzhou) in a convenience sample of Chinese MSM. Staff at gay-oriented CBOs recruited participants through four methods: voluntary counselling and testing (VCT) clinics with posters, peer networks, outreach at gay venues with fliers, and online through CBOs’ WeChat posts. Four trained research coordinators (RF, LX, JH, and SM) assessed eligibility criteria and screened men who identified as: 18 years or older, assigned male at birth, had ever engaged in sex with another man, and HIV-negative by self-report.
Computer-based surveys were administered in person by either CBO staff or research coordinators or self-administered by participants themselves. Participants gave written informed consent prior to survey administration and received compensation of $7.40 USD or equivalent condom and lubricant as compensation for their time. Recruitment halted after reaching the pre-specified sample of 700, of whom 362 were recruited through VCT clinic. We restricted the sample to VCT-recruited participants for this research question. Protocol and all study procedures were approved by the ethics review committee of Fudan University (Shanghai, China).
The survey was anonymous and collected information on sociodemographic characteristics and sexual health related behavior, including testing behaviors, sexual risk and risk perception, and preferences for dimensions of a sexual health service model. Most survey items were adapted from an existing assessment instrument for Chinese MSM .
The outcome measure was lifetime disclosure of same-sex behavior to HCPs when seeking advice for sexual health issues. This advice-seeking could relate to bleeding, pain, itching, infections in genital or anal areas, or sexual dysfunction.
Sociodemographic information included age, city currently living in, residence (local or non-local), education, employment status, student status, household composition, marital status, and average monthly income (low and middle income, middle income, and high income) .
We collected detailed information on sexual health related behavior. Participants self-reported if they ever had a sexual health issue that interfered with sex (yes or no) as well as ever experienced sexual health issues and went to see HCPs (yes or no). We broadly defined HCPs based on classification from WHO  and MSM-specific guidelines , as either health professionals, or lay HIV counselors. The survey also asked participants about their frequency of HIV and STI testing and sexual risk behavior. Variables associated with sexual risk behavior within the past 6 months included number of male sex partners, HIV-positive male partner, condomless anal sex, commercial sex behavior with male sex partners, STI symptoms/diagnosis, recreational drug use, sexual positioning, and group sex with male sex partners. We assessed self-perception of HIV risk with the following questions: “In five years, how likely do you think you are to become infected with HIV” (ranging from 1 very unlikely to 4 very likely); “In the past two years, have any of your friends or people you know become infected with HIV” (no, yes, I don’t know); and “How worried are you of becoming infected with HIV” (ranging from 1 not worried at all to 4 very worried).
After iterative discussions with GPP stakeholders in four participating cities, we developed an MSM-specific sexual health service model that participants could evaluate by prioritizing the model’s hypothetical “ideal” dimensions. Participants ranked the relative importance of each dimension of this clinic-based sexual health service model using a 5-point Likert scale (ranging from 1 not at all important to 5 extremely important). The model comprised ten dimensions (e.g., positive reputation in lesbian, gay, bisexual, transgender [LGBT] community, services are available anonymously, clinic is not LGBT-specific; see Table 2).
To answer our research question about differences by concealment of same-sex behavior, we stratified descriptive statistics for socio-demographics, testing behaviors, sexual behaviors, risk perceptions, and the sexual health service model by whether participants had disclosed same-sex behavior to an HCP at least once in their lives when seeking advice for sexual health issues (disclosure vs. non-disclosure). For our first aim to assess factors associated with disclosure vs. non-disclosure, we performed bivariate analyses using T tests and chi-square tests to evaluate the difference between the two groups on socio-demographic and sexual health related behavior. Significant variables were then analyzed in multivariate logistic regression (odds ratios, 95% confidence intervals) to evaluate independent factors associated with disclosing same-sex behavior to HCPs. Variables with p-values < .10 in bivariate analyses were selected for inclusion in an initial logistic regression model and odds ratios with p-values < .05 were included in the final multivariate model. We calculated mean scores for each dimension and across all 10 dimensions of sexual health service and likewise stratified by disclosure for bivariate analyses. All data analyses were completed using IBM SPSS Statistics 20 (IBM, Armonk, NY, USA).