Study design and setting
We conducted a cross-sectional survey between November 2018 and April 2019 of MSM living in urban areas of Chengdu, China. MSM living in Chengdu are greatly affected by HIV, with the prevalence reported to be approximately 7.5%, compared to the national prevalence of 5.7% in China [26]. The large population of MSM in Chengdu means it is a key city in the emerging MSM epidemic, as it attracts a large number of migrants and it is a major stop on the heroin trafficking route through China. Therefore, heroin is more accessible in Chengdu than in other parts of China [27]. MSM who are drug users (e.g., heroin or crystal methamphetamine users) tend to have a greater amount of unprotected sex and more severe mental disorders than MSM who do not use drugs, which increases the risk of HIV infection in drug-using MSM and leads to worse health outcomes [28, 29].
Participants and recruitment
Participants were recruited from clients of a local lesbian, gay, bisexual and transgender (LGBT)-friendly CBO that provides HIV-related services. Two staff members of the CBO contacted all potential participants from the CBO’s list of clients by phone to screen for their eligibility. The clients were eligible to participate if they were i) male at birth, ii) 18 years or older, iii) self-reported having had anal intercourse with at least one man in the past 6 months, and iv) had access to at least one HCP during their lifetime. No one was excluded based on HIV status. We identified 868 individuals that were eligible for participation in the study, and all of them were invited to attend the CBO to complete a questionnaire. During the study period, 689 participants (response rate of 79.4%) completed the survey in person. A research assistant was stationed at the CBO during the study period for progress monitoring, data collection and quality control.
Participants were briefed about the study purpose and procedure, and they provided written informed consent before starting the survey questionnaire, which was anonymous and self-administered using an iPad. The questionnaire took an average of 28 (standard deviation [SD] = 20) minutes to complete, and participants were offered USD7 in cash to compensate for the time taken to participate in the study. Confirmatory HIV testing was offered to all participants, for which separate consent was obtained. Ethical approval was obtained from the Ethics Committee of Sun Yat-sen University ([2018] 049).
Measures
All questions used in the survey were pilot-tested on 43 eligible participants, who were excluded from the formal survey. Minor revisions were made based on the pilot results and the comments from participants.
Background
The following socio-demographic information was collected: age, ethnicity, local household registration (hukou), educational level, relationship status, employment status, personal income, and self-rated health status. We also asked the participants about their self-identified sexual orientation and self-reported HIV status.
Same-sex behavior disclosure to HCPs
Information on same-sex behavior disclosure to the three types of HCPs (hospital clinicians, CBO peer educators, and CDC public health specialists) was collected separately. An example question was: ‘Have you disclosed your same-sex behavior to the hospital clinicians who provided services to you?’ The response options included: ‘all of them’, ‘some of them’, ‘none of them’, and ‘not applicable to me’ (the latter response was given by participants who had not had access to any hospital clinicians during his lifetime). Participants who responded with ‘all of them’ or ‘some of them’ were combined and classified as Y = 1, while those who responded with ‘none of them’ were classified as Y = 0. Participants who responded with ‘not applicable to me’ were excluded from the analyses.
HIV testing history
We asked the participants if they had their HIV status tested before the survey (lifetime HIV testing) and also if they tested HIV in the past 6 months (recent HIV testing). The response options were ‘yes’ or ‘no’.
Sexual behaviors
Participants were asked to recall the total number of partners with whom they had had sex in the past month. Participants who reported having had sex with more than one partner in the past month were classified as having “multiple sexual partnerships”. Participants who did not use condoms with all partners in the past month were classified into “inconsistent condom use”.
Immediate ART awareness
We asked the participants to determine if the statement ‘A person who is newly diagnosed with HIV should start ART immediately’ was correct. The response options were ‘correct’, ‘incorrect’, and ‘I don’t know’. Those who responded with ‘incorrect’ or “I don’t know” were combined and classified as unawareness of immediate ART.
U=U awareness
We asked the participants to determine if the statement ‘A person with an undetectable viral load cannot transmit HIV to others’ was correct. The response options were ‘correct’, ‘incorrect’, and ‘I don’t know’. Those who responded with ‘incorrect’ or ‘I don’t know’ were combined and classified as unawareness of U=U.
PrEP awareness
Participants were asked if they had heard of any type of PrEP (daily oral PrEP, on-demand oral PrEP, or long-acting injectable PrEP), although none of these PrEP options were available in China at the time of the survey. Participants who answered ‘yes’ were classified as having PrEP awareness, whereas those who did not were classified as having no PrEP awareness.
Statistical analysis
Bivariate associations were assessed using binary logistic regression to examine same-sex behavior disclosure to each of the three types of HCPs, with the four outcomes being HIV testing, immediate ART awareness, U=U awareness, and PrEP awareness. Bivariate analyses were performed to assess the association between background variables (e.g., age, educational level, income, self-identified sexual orientation, and sexual behaviors) and the four outcomes described above. The measures of association are presented as unadjusted odds ratios (ORu) with 95% confidence intervals (95% CIs). Subsequently, variables found to have a significant effect (P < 0.05) in the bivariate analyses were included in a multivariable logistic regression analysis, and the measures of association are presented as adjusted odds ratios (AORs) with 95% CIs. All statistical analyses were performed using SPSS Statistics (version 26; IBM, Armonk, NY, USA), and a two-tailed P value < 0.05 was considered statistically significant.