Study design, duration and population
Between September 2017 and December 2018, we conducted a descriptive cross-sectional study. MSM respondents were deemed eligible to participate in the study if they were: (i) over 18 years old, (ii) MSM who reported having had anal or oral sex within the previous six months, (iii) MSM who agreed to be tested for HIV and disclose information about their most recent experience of sexual intercourse, and (iv) were willing and able to sign a written informed consent document. Subjects who had previously tested HIV-positive were excluded from the study.
Study setting
This study was completed in Shenyang, a politico-economic-cultural center in northeastern China and the provincial capital of Liaoning Province. Shenyang’s gross domestic product (GDP) ranked 34th of the 100 largest cities in China in 2018. Shenyang has a population of over 8.1 million, with an estimated 140,000 MSM [22]. 225 VCT clinics provide free HIV testing for MSM, FSWs, and other high-risk groups. Two HIV/AIDS designated hospitals to provide HARRT treatment for over 19000 HIV infected people, and our VCT clinic is an outpatient department from one of the HIV/AIDS designated hospitals. The predominant HIV transmission route of annual newly reported HIV/AIDS cases in Shenyang was via the male-to-male sexual route, which accounted for 80.3% (712/887) cases in 2017. The reported HIV incidence was 6.9 (95% confidence interval (CI): 4.9–9.3)/100 person-years among MSM in Shenyang in 2013 [23, 24].
Sample size and sampling procedure/technique
The MSM respondents were recruited via a mixed-method, including outreach recruitment by community volunteers in places such as bars, parks, and baths, peer referrals, and recruitment on gay websites and gay chat rooms. This study was conducted at the First Affiliated Hospital of China Medical University. The questionnaire was administered by professionally trained staff in one-to-one, face-to-face interviews. After obtaining written informed consent from each research participant, we collected their sociodemographic information, sexual behaviors, and RD behaviors. The sociodemographic information included age, place of residence, educational level, highest educational degree obtained, marital status, and ethnicity. Sexual behaviors included the main channel for finding sexual partners, sexual roles, and usage of condoms and nitrite inhalants during their most recent sexual intercourse. For RD behaviors in the most recent experience of anal intercourse, we asked MSM participants if they had cleansed their rectum before and/or after RAI, the type of liquid used for douching, the type of douching product, the reasons for RD, and their serosorting behavior, defined as the practice of agreeing to have unprotected anal intercourse only with partners of the same HIV status, which is becoming increasingly popular among MSM in general [7]. All MSM who participated in this study received a HIV consultation before and after HIV testing, as well as condoms, lubricant, and educational materials on HIV.
The required sample size was calculated using an estimation formula based on the difference between two sample rates for the cross-sectional survey study [25]. The significance level was 0.05, and power was 0.8. To compare the difference of HIV prevalence between the RD and non-RD group, the sample size calculated referred to a published study in which HIV prevalence was 4.3%(11/258) in the non-RD group and 14.9%(47/315)in RD group[11]. After calculations, each group should have a minimum sample of 121 participants. We utilized PASS (Power Analysis & Sample Size) software version 15 to calculate the sample size.
Laboratory tests
Following completion of the questionnaire, 10ml of venous blood was collected from each research participant for HIV-1 antibody screening, HIV-1 Pooled-RT-PCR, Western blot (WB), and Treponema pallidum (TP) testing.
HIV antibodies were screened using Biomérieux’s Human Immunodeficiency Virus (HIV 1/2) antibody diagnostic kit (ELISA method). Secondary screening was performed using Abbott Laboratories’ Human Immunodeficiency Virus Antibody Rapid Detection Kit (colloidal gold method). Once HIV-positive status was confirmed by the screening, the HIV confirmation test was performed using Gene Lab’s serum HIV Western blot (WB) method. Antibody-negative specimens were tested with the 24 mini-pool nucleic acid amplification test (NAAT). Blood samples giving a positive ELISA result but a negative or indeterminate WB result were tested with NAAT individually without mixing, using COBAS AMPLICOR HIV-1 MONITORTM Test, v1.5 diagnostic kit (Roche, 21118390123), according to the COBAS AmpliPrep/COBAS TaqMan HIV-1 Test method [26].
For the detection of plasma TP, a rapid serotonin ring card test (RPR) was used for screening (Shanghai Kehua Bio-Engineering Co., Ltd.), and those who tested positive were further confirmed with Treponema pallidum particle agglutination (TPPA) (Fuji Corporation, Japan). Those who were screened and confirmed as positive were deemed to be TP-positive.
Data analysis
Questionnaire and laboratory test results were recorded by two research assistants twice using EpiData 3.02 software until the data from both questionnaires were fully consistent. Descriptive statistical analysis was performed using mean ± standard deviation or the median and interquartile range (IQR) as a measure of statistical dispersion. Categorical variables were described using percentages of its frequency. Chi-square tests were used to compare differences in the categorical variables (e.g., marital status, education, and information related to the most recent anal sex). We used the median and interquartile range to describe the central tendency and dispersion of age, respectively.
HIV infection-related factors were analyzed using multivariable logistic regression model analysis. On the basis of univariate analysis, variables with p < 0.2 were screened out and entered into a multivariable logistic regression model analysis. Both Model 1 and Model 2 controlled age, education, and household registration. Model 2 added sexual role as an additional covariate. The interaction between RD and CRAI was evaluated by the above two multivariate logistic regression models. The inspection level α was set at 0.05. Statistical analysis was performed using SPSS 17.0 statistical software (IBM).