RDS has gathered robust samples of MSM in studies worldwide, including in Rio de Janeiro [5,6]. To reach and engage YMSM in Rio de Janeiro, Brazil, we will conduct an RDS-based study at the National Institute of Infectious Diseases Evandro Chagas (INI)-Fiocruz. The RDS will be adapted to include social media-based methods. The approach builds on research on young transgender women aged 15-24 years in San Francisco, United States [21]. Starting with initial “seeds” (ie, initial participants who start recruitment chains), successive waves of referrals will reach diverse social networks of YMSM. We anticipate that peer referrals through social media connections will particularly appeal to Brazilian YMSM who spend much of their time on mobile-based apps (eg, WhatsApp) and digital social media. Despite the discussion around the generalization of their data, RDS-based studies provide estimates of HIV risk among hard-to-reach populations [22] and allow the recruitment of these populations into longitudinal studies (ie, observational or intervention). Subsequently, we will offer all recruited individuals a prospective 48-week technology-based adherence intervention study to promote a rapid linkage of YMSM to HIV care or prevention and support and sustain adherence.

Individuals will be included if they (1) self-identify as men (cis or trans) or gender nonbinary, (2) report ever having engaged in anal sex with men or gender nonbinary persons with a penis, (3) are aged 18 to 24 years, (4) reside or spend most of the time in Rio de Janeiro metropolitan area, (5) did not previously participate in the study, and (6) possess an electronic referral coupon from a peer acquaintance participant. In this study, we collectively refer YMSM to young (18-24 years) cisgender men, transgender men, and non-binary or gender-nonconforming individuals. Exclusion criteria include individuals who self-identify as women (cis or trans).

RDS-based studies require a formative phase to verify theoretical assumptions and identify logistical constraints and solutions [23]. These objectives are met through focus group discussions, key informant interviews, and pilot testing. The theoretical RDS assumptions are: (1) members of the population know each other as members, (2) social networks are interconnected within a few degrees of separation, (3) sampling occurs with replacement, (4) network size is reported accurately, and (5) people recruit approximately randomly from their network. RDS also depends on sufficient network density to make long recruitment chains [24]. Furthermore, the characteristics of social networks guide the selection of the number and type of “seeds.” The formative phase also identifies potential “bottlenecks” (ie, social and physical barriers between networks) and variables to track “equilibrium” (ie, stability in sample composition as the chains grow). Other logistical questions include study site acceptability, transportation, online and social media apps for electronic referrals, safety, confidentiality, and appropriate incentives for participation and recruiting peers. In addition, the formative phase will inform the development and pilot testing of the questionnaire, as well as the refinement of the technology-based intervention.

The formative phase has already been conducted, comprised of two focus group discussions (aged 18-19 and 20-24 years) of up to 10 participants each and individual interviews with up to 20 YMSM. Group discussions and individual interviews lasted 1-2 hours. YMSM were referred from HIV clinics, LGBTQIA+ (lesbian, gay, bisexual, transgender, queer/questioning, intersex, and asexual/aromantic/agender) nongovernmental organizations, and peer counselors at Instituto Nacional de Infectologia Evandro Chagas, Fundação Oswaldo Cruz (INI-Fiocruz). We used a maximum variation sampling approach to include a diversity of participants by place of residence, race/ethnicity, and age [25]. Group discussions and individual interviews were audio-recorded with field notes captured by the interviewer. Files were transcribed in Portuguese and translated into English. Content analysis using transcribed content is underway and will focus on RDS theory, logistics, and strategies to improve the study. The study team will analyze qualitative data and discuss the preliminary results related to the density of the social network and logistics to reach a consensus on strategies to improve the implementation of RDS in this study. If social networks appear weak or diffuse, we will develop an additional seed recruitment plan and alternative approaches to improve recruitment based on the findings within a theoretical probability-based sampling framework. Questions raised during the formative phase will be incorporated into the RDS study instruments to be quantified. A chosen subset of participants (approximately10) will pilot the questionnaire. Participants will provide feedback regarding comprehensibility, missing and unnecessary constructs, wording, and timing.

Within an acceptable margin of error, the sample size of RDS surveys is powered to measure key indicators of HIV risk and prevention in the population of YMSM (eg, HIV prevalence, PrEP awareness, willingness to use PrEP, and never testing for HIV). Assuming a design effect of 2.0 (typical for RDS) [26] and a 95% CI, 400 participants are sufficient to measure RDS-adjusted estimates within SD 5% over a wide range of point estimates (ie, 5%-40%). For example, 400 participants could measure HIV prevalence at 5%, SD 2.1% or being unaware of PrEP at 39%, SD 4.7%. A sample size of 400 also provides 80% power at a 95% CI to detect significant odds ratios for effect sizes of 1.9 or greater for key outcomes (eg, HIV infection, PrEP awareness, and willingness to use PrEP) and predictor variables (eg, race/ethnicity, perceived HIV stigma, and family support).

RDS sampling begins similarly to “snowball sampling” (ie, with purposively selected “seeds” from diverse social networks who then refer peers to the study). Peer recruitment on RDS differs from snowball sampling on key procedural and theoretical factors that enable better estimates of disease prevalence [22]. Seeds will be chosen for their connections to other YMSM, enthusiasm for the research, and belonging to different social circles. As with all other participants, the seeds must be eligible and undergo all study procedures. After completion, the seeds will be trained to recruit their peers using a digital coupon. This process will create a link between the recruiter and recruits, providing information needed for statistical adjustment. Digital coupons (Figure 1) will be shared on mobile apps and web-based texting platforms [25,27]. Digital referrals significantly increased recruitment of a cohort of transwomen aged 15-24 years in San Francisco, United States [21,25]. Upon presentation of a digital coupon to the study site, recruits will undergo similar procedures if eligible. The process will continue until the sample stabilizes on key characteristics (“equilibrium”) and the sample size is met. RDS requires monitoring for “bottlenecks” (eg, chains that do not cross networks) and valid connection (eg, true acquaintances). The limited number of recruits per participant (typically 3) and incentives drive the propagation of long chains of referrals within and across social networks. Participants receive a “primary incentive” for their enrollment and a “secondary incentive” for each eligible participant they recruit. The value and type of incentives will be explored and determined in the formative phase.

All participants will be tested for HIV using an HIV rapid test and antibody/antigen 4th generation serology. Participants reporting recent anal condomless sex (<30 days) with a negative HIV rapid test will be screened for acute HIV infection through HIV RNA viral load (VL) testing.

We will also test for recent HIV infection with the limiting avidity assay [28] to identify seroconversions occurring in the last few months. Recency data can help to identify possible transmission clusters and permit the calculation of HIV incidence by providing the timing of seroconversion [28]. The detection of acute infections may contribute to identifying active clusters of HIV transmission. HIV VL reaches high levels in the short period after infection, increasing the risk of HIV transmission. Recent HIV acquisition events will allow the characterization of drivers of new infections and avoid recall bias because remembering sexual partners once in the last few weeks is easier than over a longer time.

We will perform HIV phylogenetic testing on specimens with detectable HIV VL. Phylogenetic data will identify HIV strain types, ART resistance mutations, and clusters of transmission.

The incorporation of biomarkers for acute and recent infection, as well as STIs, may be particularly valuable among YMSM, where the onset of behaviors leading to infection is more recent. We will evaluate STIs as markers of HIV sexual risk and causes of morbidity using nucleic acid amplification tests for oropharyngeal, urethral and rectal Neisseria gonorrhea and Chlamydia trachomatis (nontreponemal syphilis testing and if positive treponemal test), chronic hepatitis B (anti-HBs antibody, Hbs antigen, and total anti-HBc antibodies), chronic hepatitis C (anti-HCV), human papillomavirus testing and anal cytology. Participants diagnosed with STI will receive treatment according to Brazilian guidelines [29]. Participants screened positive for hepatitis B and C will undergo a complete diagnosis algorithm (ie, HCV viral load) [30] and will be referred to treatment. Participants without hepatitis B diagnosis and who have not been vaccinated will be referred for hepatitis B vaccination.

Trained staff will administer face-to-face questionnaires to participants using tablets. Due to the COVID-19 pandemic, all efforts will be undertaken to minimize SARS-CoV-2 transmission. Before study procedures, participants will be screened for COVID-19 symptoms, and visits will consider the safety of all participants and the study team. Our instrument will build upon questionnaires previously used for MSM in Brazil and the United States. Table 1 presents examples of measurements, hypotheses, and sources of questionnaires.

Biological data, particularly markers of recent HIV and STI infection, will permit closer linkage to the events, behaviors, partnerships, and social and structural conditions that lead to HIV acquisition. We will also identify whether there are infection clusters within specific racial/ethnic groups, as previously seen among Black MSM in the United States [31]. With robust biomedical markers of infection, we will be able to discern whether tight sexual networks constrained by racial bias and discrimination impact the spread of HIV among Black and Pardo Brazilians and whether this is a recent phenomenon or ongoing over the course of the Brazilian MSM epidemic.

All RDS participants will be invited to participate in a technology-based adherence intervention study. The intervention builds upon Health eNav, a text messaging HIV care linkage and retention intervention for young people living with HIV in San Francisco, United States, using an SMS-based platform to support ART adherence and provide HIV prevention counseling [21]. Based on a prior study, we anticipate 13% HIV prevalence [20] (N=52 individuals), 26 (50%) of whom will be newly diagnosed and 26 (50%) previously diagnosed (regardless of linkage to care). Therefore, we estimate that 65% (226/348) of HIV-negative YMSM, as previously described [15], will be interested and eligible for PrEP according to the Brazilian guidelines (eg, condomless anal sex, sex with HIV positive partner, and STI diagnosis in the last 6 months) [52].

HIV-negative participants enrolled in the intervention study will receive same-day PrEP, according to the Brazilian recommendations, plus a complete prevention package, including HIV/STI testing and treatment, counseling for HIV/STI risk reduction, hepatitis B vaccination, and condoms/lubricants distribution (Group 1: HIV-negative on PrEP). HIV-negative YMSM not eligible for PrEP or who do not accept PrEP may also be enrolled in the intervention study to receive the prevention package (including PEP, when indicated), risk assessments, and counseling (Group 2: HIV-negative not on PrEP).

HIV-positive YMSM enrolled in the intervention study will initiate same-day ART treatment according to Brazilian recommendations and will be further linked to care at the INI-Fiocruz HIV clinic (Group 3: HIV-positive on ART). HIV-positive YMSM previously diagnosed and who initiated ART prior to inclusion will also be invited to the intervention.

All participants enrolled will receive weekly reminders for 24 weeks via WhatsApp, the most common SMS platform in Brazil. Automatic messages will include scheduled medication reminders, if applicable, and posts. Periodic feedback from participants will adapt and improve the approach. (eg, adjustments or modifications on the content of electronic messages). The weekly reminders will be personalized. Participants will be able to select the timing of reminders and choose between having explicit medication reminders or factoid messages (not specific to medication), such as LGBTQIA+ community events, health facts, or a combination of both types of messages. All messages sent will solicit a response, so we will have backend data (eg, time, date, message content, etc) on whether the message was received and reviewed by participants. For example, we may send a medication reminder text, such as “Did you take your medication today?” Responses will be recorded in our database, but more importantly, we will know whether the participant was exposed to the text message regardless of receiving a formal response.

After enrollment, follow-up visits will occur at 4 and 12 weeks and then quarterly (24, 36, and 48 weeks). Briefly, during follow-up visits, we will collect information on study outcomes, such as HIV seroconversion, PrEP continuum endpoints, and PrEP adherence (using dried blood spots for TFV-diphosphate concentration among HIV-negative participants); HIV viral load suppression among HIV-positive participants; and STI incidence among all participants. The intervention study baseline visit procedures are the same as those conducted during the RDS visit. A detailed schedule of events and procedures is depicted in Table 2.

We will measure exposure to the intervention, including the number of received and read WhatsApp messages during the 24-week intervention period. Intervention exposure and outcome data will be analyzed using multivariable logistic generalized estimating equations (GEE) with exchangeable correlation structures to account for repeated measures over time for each participant. GEE models will assess the association between intervention exposure measures (eg, receipt of intervention and intervention dosage) and primary outcomes related to the HIV care and PrEP continuum.