Study location, design, and population
The study is based on data collected in a behavioral and biological surveillance study entitled “A nationwide study of the behaviors, attitudes, practices, and prevalence of HIV, Syphilis and Hepatitis B and C among MSM in Brazil”, conducted from June to December 2016 in 12 state capitals: Manaus and Belém in the Northern region; Fortaleza, Recife and Salvador, in the Northeast; Brasília and Campo Grande in the Central-West; Belo Horizonte, Rio de Janeiro and São Paulo in the Southeast; and Curitiba and Porto Alegre in the South.
The sample consisted of men defined as biologically male, 18 years of age or older, reporting oral sex or anal intercourse with another man in the previous 12 months; and living, studying or working in one of the study cities.
RDS methods were used to recruit participants. Before initiation of the survey, formative qualitative research using both individual interviews and focus group discussions was conducted to explore participation in the study, including willingness to recruit, testing, study logistics, such as level and kind of incentive, siting, and working hours .
To start the recruitment process six MSM were purposively selected in each city. These individuals, known as seeds, had large social networks of MSM and represented different sociodemographic characteristics. Next, each of these seeds invited three other MSM from their social networks using three invitation coupons. This procedure was repeated in sequence until the desired sample size was reached, estimated a priori as 350 participants per city. Participants received BRL 25.00 (USD $7.40) for their participation and for each recruited participant (BRL 25.00 for each one up to a total of three new participants - BRL 100.00) as a means of reimbursing their expenses for transportation and food. Data were collected using computer assisted personal interviews (CAPI) at fixed office sites, where educational materials, condoms and lube were provided, and where counseling, blood draws and results were provided. A total of 4,176 MSM were recruited in the 12 cities, who signed separate consent forms for the interview and for testing. The Research Ethics Committee of the Federal University of Ceará approved the research project. Further details can be found in Kendall et al. .-
A descriptive analysis was performed for the following variables: (1) demographic and education: age (<25 years and ≥25 years), self-declared race (white, black, brown/mixed race, indigenous/native, Asian origin), schooling (primary and incomplete secondary education, complete secondary education, incomplete undergraduate education, and complete undergraduate education) and marital status (single, separated or widowed; married or living together); (2) socio-economic status with three categories: A-B for high; C for middle; D-E for low) based on the Brazilian Economic Classification Criteria ; (3) discrimination: self-reported discrimination due to sexual orientation during lifetime (yes, no); (4) health services: medical appointment in the 12 months before the study (yes, no), having taken part in any workshop about sexually transmitted infections (STI) and HIV/AIDS in the 12 months before the study (yes, no); and ever testing for HIV was determined by responses to the following question: “have you ever been tested for HIV/AIDS in your life?” (yes, no); (5) knowledge about HIV/AIDS (low, middle and high) was analyzed according to item response theory (IRT), measured using 12 items about transmission and prevention (i.e. there are medicines for HIV-negative people to take to prevent HIV; an HIV-infected person who is taking AIDS medication has a lower risk of transmitting the virus, etc.), as described in Guimarães et al. ; (6) self-perception of risk (low, middle and high), in response to the question: “how do you assess your chance of becoming infected with HIV during your lifetime?”; (7) participation in an LGBT nongovernmental organization (NGO) (yes, no); and (8) condomless receptive anal sex in past six months (always used condoms and irregular use of condoms).
The outcome HIVST acceptability (yes, no) was structured according to the following question: “would you use an HIV/AIDS diagnosis test that you applied to yourself?”. The reasons for the decision to use or not use an HIVST were also collected.
We excluded MSM who were aware of their positive HIV serostatus before the study, yielding 3,605 MSM for this analysis. The dependence between observations resulting from recruitment chains in RDS, i.e., the probability of unequal selection and the different sizes of each participant’s contact network , was taken into account. Gile’s estimator  was used to weigh the proportion estimates with a 95% confidence interval using RDS Analyst . We aggregated all data from the 12 cities into a single dataset with each city serving as a stratum.
The weighted data were analyzed using the complex sample routines in STATA 15 (StataCorp, College Station, TX, USA). Descriptive, bivariate analyses were conducted using frequency distributions of the variables of interest and the differences between the analyzed proportions using Pearson’s 𝜒² test. All analyses were stratified according to prior HIV testing. i.e., those who were never tested or those who were tested at least once in a lifetime and their HIV result was negative. A multivariate logistic regression model was constructed to estimate the Odds Ratio (OR) with the associated 95% confidence intervals (CI), as a measure of association between the potential associated factors and HIVST acceptability. Selection of variables for multivariate modeling was based on the bivariate analysis (p-value <0.15) and the epidemiological relevance, taking into account a previous literature review, concerning potential factors associated with HIVST acceptability. Those variables with a p-value of < 0.05 remained in the final model.