Introduction

In Latin America and the Caribbean (LAC), an estimated 2.5 million people were living with HIV by 2019, and more than 133,000 new infections were detected, of whom 22,500 were aged 15 to 24 years [1]. The HIV epidemic in LAC is concentrated in key populations. While the HIV prevalence in the adult general population is 0.4% in Latin America and 1.1% in the Caribbean, rates among men who have sex with men (MSM) have been estimated as 12.6% in Latin America and 4.5% in the Caribbean from 2015–2019 [2]. HIV infection among MSM in the region has been associated with behavior (e.g., condomless anal sex, sex with multiple partners) and structural factors (e.g. economic instability, low socioeconomic status) [3,4,5,6,7,8].

Despite the emergence of new HIV prevention technologies and strategies such as treatment as prevention (TasP, endorsed by World Health Organization [WHO] in 2014), pre-exposure prophylaxis (PrEP, recommended by WHO since 2015), and HIV self-testing (recommended by WHO since 2017) [9], the number of new infections has decreased only slightly over the last decade and is on the rise among young key populations [2]. The overall number of people on PrEP across the region remains insufficient to have a significant impact on the epidemic. PrEP through the public health system is currently only available in Bahamas, Barbados, Brazil, and Cuba, and through private clinics, the Internet, nongovernmental organizations, and pilot studies in Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, Guatemala, Haiti, Jamaica, Mexico, Panama, Paraguay, Peru, Suriname and Uruguay [2, 10]. It is estimated that 23% of people living with HIV in LAC still do not know their status by 2019. In Brazil, among 16,667 MSM responding web-based surveys between 2016 and 2018, 17% reported have never been tested for HIV in their lifetime, with higher percentages in low-income cities [11]. In addition, the “undetectable equals untransmittable” (U = U) slogan is still not correctly understood by different populations including MSM [12], jeopardizing prevention benefits of TasP.

Data from the United States and Asian countries find that young MSM (YMSM) aged 15–24 are at higher risk of HIV infection compared to young heterosexual men and to older MSM [2, 13,14,15]. Lack of access to appropriate sexual education and preventive strategies, condomless and PrEP-less sex, use of alcohol and drugs during sex, transactional sex, and low perceived HIV risk are some of the factors associated with higher vulnerability to HIV among YMSM [11, 16,17,18,19]. Structural barriers in LAC may affect PrEP implementation and continuation among YMSM. Results from ImPrEP study (a large multicountry demonstration PrEP study currently ongoing in Brazil, Mexico and Peru) have shown that early continuation (attending 2 following-up visits) among 1,843 YMSM was 67%, being lower among those at higher social vulnerability, such as nonwhite and less educated [20]. A large web-based survey also identified an association between socioeconomic disparities and HIV prevalence among Latin-American MSM [21]. Homophobia, stigma, and fear of disclosing sexual orientation are other barriers YMSM face in accessing prevention and care services, as well as participation in research [22,23,24].

Gathering rigorous data on YMSM is increasingly needed to guide TasP, PrEP, and other effective interventions if we are to end the HIV epidemic in LAC by 2030 [25]. An important data challenge identified in LAC to date has been utilization of methods that produce representative samples of hidden and hard-to-reach populations in research on key populations like YMSM. In fact, data from non-probability-based samples constitute a large body of the literature on MSM. Probabilistic methods for sampling YMSM do exist. Respondent-driven sampling (RDS) is an approach that approximates probabilistic sampling and has been endorsed by the WHO, the Joint United Nations Program on HIV/AIDS (UNAIDS), Global Fund, and the Centers for Disease Control and Prevention (CDC). Time-location sampling (TLS) and true population-based surveys are sometimes conducted with MSM, though more costly and much less frequently. In this systematic review, we sought to synthesize the literature on HIV prevalence of MSM in LAC using data from probabilistic and non-probabilistic studies. We aimed to assess HIV prevalence in MSM over the years, particularly in YMSM, outcomes associated with the HIV care cascade, and factors associated with HIV infection in the YMSM population in LAC.

Methods

Protocol and Registration

This study was registered with the international database of systematic reviews in health and social care (PROSPERO). This study followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Statement [26].

Search of studies.

We performed a bibliographic search on MEDLINE, EMBASE, and Virtual Health Library (VHL, Pan American Health Organization/BIREME) restricted for publications between January 2010 and May 2020. Included publications had to be published from 2010 onwards but data collection could have been initiated prior. There was no language restriction. The VHL search was performed in English, Portuguese, Spanish, and French. The search strategy combined terms derived from four domains: (a) MSM; (b) HIV/AIDS; (c) HIV prevalence; and (d) the countries of LAC (see Supplementary file 1 for the full PubMed search strategy).

Eligibility Criteria

The literature search aimed to include studies reporting HIV prevalence in MSM living in LAC. Peer-reviewed publications and conference abstracts were included if they met all the following criteria: (i) published after 2010; (ii) conducted in LAC; and (iii) reported HIV prevalence. We excluded studies that (i) did not report the HIV prevalence specifically for MSM (studies that reported combined prevalence for MSM and trans women were included, however); (ii) repeated estimates reported in another study already included; (iii) did not provide original data (e.g., reviews). The systematic review did not include other sources of grey literature such as theses, dissertations, monographies, or reports. Two investigators (LEC, TST) reviewed all abstracts and full text publications independently to verify if they met the eligibility criteria.

Data Extraction

Two investigators (LEC, TST) independently extracted data from the selected publications using a predefined form and discussed disagreements to resolve discrepancies. Data extracted included: (i) recruitment setting (sites, cities, countries); (ii) year of data collection; (iii) study sample characteristics (sample size, proportion of MSM and trans women, age); (iv) HIV prevalence and 95% confidence intervals (95% CI) in MSM of all ages and among YMSM; (v) factors associated with HIV prevalence; and (vi) HIV care cascade outcomes (previous HIV testing, awareness of HIV status, antiretroviral therapy [ART] use, viral suppression). Results were stratified into probability and non-probability sampling methods. Investigators attempted to reach all publication corresponding authors to request YMSM prevalence estimates that were not included in the publications.

Results

Search Results

Our search found 47 unique studies, described in 49 publications, and conducted in 17 countries from 2006 to 2020 (Suppl Fig. S1), although no eligible studies were published from 2017 to 2020. Studies’ sample sizes varied from 41 to 7,823 participants. Most studies included only participants age 18 years and above; six studies included participants under 18 years. Nineteen studies included both MSM and trans women.

HIV Prevalence

Results of studies that used probabilistic sampling methods (N = 21), including RDS, TLS, stratified sampling of military conscripts, and stratified household sampling are shown in Table 1. Five of these 21 studies (23.8%) were conducted in Brazil. A total of 21,817 MSM were included in the 21 studies, with 11,118 (51.0%) from Brazil. Among those studies, HIV prevalence among all-age MSM ranged from 1.2% (95% CI 0.3, 3.6) in Santos, Brazil conducted in 2009 [27] to 32.6% (95% CI 18.0, 47.8) in Colón, Panama in 2012 [28]. Among cities that repeated studies at different time points, HIV prevalence tended to have increased over time in Belo Horizonte, Campo Grande, Curitiba, Manaus, Recife, and Salvador in Brazil and Quito, Ecuador [5, 27, 29, 30]. HIV prevalence tended to have decreased over time in Brasilia and Rio de Janeiro in Brazil and Bogota, Colombia [27, 30,31,32]. Nationally in Brazil, HIV prevalence was 1.23% among MSM who were conscripts in 2007 compared to 1.32% in 2016 [33, 34]. Overall, the proportion of YMSM (aged 18–24) included in these studies are shown in fifteen studies and ranged from 0.5% in Salvador, Brazil in 2009 to 78.4% in Brasilia, Brazil in 2016 [27, 30].

Table 1 Studies with probability-based sampling methods to recruit men who have sex with men, Latin America and the Caribbean, 2007–2016 (N = 21)
Full size table

Results of studies that used non-probabilistic methods of recruitment (e.g., targeted outreach, peer referral, clinic clients, or non-governmental and community-based organization clients) are shown in Table 2 (N = 26 studies). At total of 32,403 individuals were included in those studies with 12,857 (39.7%) from Peru. HIV prevalence among MSM ranged from 2.2% (95% CI 0.9, 5.3) in Haiti in 2016 [7] to 32.3% (95% CI 25.2, 47.9) in Jamaica in 2008 [8]. The proportion of YMSM included in these studies (N = 7) ranged from 61.7% in Jamaica in 2011 to 75.0% in Haiti in 2016 [7, 35].

Table 2 Studies using non-probability-based sampling methods to recruit men who have sex with men, Latin America and Caribbean, 2005–2016 (N = 26)
Full size table

Nineteen studies (15 that used probability-based sampling methods and 4 non-probability-based sampling) reported HIV prevalence estimates among YMSM (aged < 25) (Fig. 1). HIV prevalence was greater than 5.0% in more than a half (51%; 22/43) of the estimates. The highest HIV prevalence among YMSM was found in RDS surveys conducted in Panama City in 2012 (17.3%, 95% CI 9.6, 27.3) and in Colón, Panama in 2012 (15.6%, 95% CI 4.9, 32.4%) [28]. Among studies using non-probabilistic recruitment methods reporting HIV prevalence among YMSM, the highest estimates were in Jamaica using targeted outreach in 2008 (28.3%) [8] and peer referral in 2011 (24.1%) [35].

Fig. 1
figure 1

HIV prevalence among YMSM in Latin America and the Caribbean, 2006–2016

Full size image

HIV Care Cascade

HIV testing history and awareness of HIV status among all-age MSM are shown in Table 3. The proportion of MSM who had previously tested ranged from 26.8% among patients at three STI clinics in Escuintla province, Guatemala in 2007 [36] to 85.1% (95% CI 74.9, 91.6) in a national survey in Haiti nine years later in 2016 [7]. Few studies provided the proportion of awareness of HIV-positive status. In an RDS survey in San Salvador, El Salvador in 2008 [37], only 11.0% of HIV-positive participants were aware of their status, contrasting with 64.0% in Santiago, Chile in 2016 [38]. Only one study, a national survey in Haiti in 2016, reported the proportion of HIV-positive individuals on ART (86%, 5 out of 6) and those on ART with an undetectable viral load (80%, 4 out of 5) [7].

Table 3 HIV testing and status awareness among men who have sex with men, Latin America and the Caribbean, 2007–2016
Full size table

Specifically for YMSM, one study conducted from 2007 to 2009 in Buenos Aires, Argentina reported prior HIV testing in 30.5% of YMSM, and 33.3% who tested HIV-positive were aware of their HIV status (2 out of 6) [39]. In Brazil, a pooled analysis of RDS surveys in 10 cities in 2009 identified that 28.0% of MSM aged less than 20 years had a prior HIV test. In addition, never testing for HIV was independently associated with younger age [40].

Factors Associated with HIV Infection

Several studies characterized factors associated with HIV infection among MSM of all ages (Table 4), but none reported associations specifically among YMSM. In addition to various indicators of sexual risk behavior, older age, lower socio-economic status, trans women gender identity, drug use, and prior violence experiences were associated with higher HIV prevalence among MSM.

Table 4 Factors associated with HIV infection among men who have sex with men, Latin American and the Caribbean, 2008–2016
Full size table

Discussion

In this systematic review, we documented the high and potentially rising HIV prevalence among YMSM living in LAC. Estimates greater than 10% were observed in all sub-regions, from Meso and Central America, the Caribbean, the Andes, and the South American Cone [28, 30, 35, 41, 42]. The high HIV prevalence was consistent across survey methods. Although several RDS studies have evaluated HIV prevalence among MSM in the region over the past 10 years, none of them specifically targeted YMSM. Unfortunately, many reports of epidemiologic data on key populations do not disaggregate by age, as in most of the studies included in this review. Moreover, it is hard to estimate the population size of YMSM, and this group may be underrepresented in studies due to disparities between young and older MSM with respect to such factors as internalized and external homophobia, economic insecurity, and lack of social and parental support [43].

Globally, YMSM are disproportionately affected by HIV and are one of the few populations with worrisome increases in the numbers of new cases, despite stabilizing rates in other groups. HIV surveillance data from Bangkok, Thailand, has shown an increasing prevalence among MSM aged 22 years and under, from 13% in 2003 to 24% in 2014 [44]. In Nigeria, HIV incidence was four- and three- fold higher in MSM aged 16 to 19 years and 20 to 24 years, respectively, compared to MSM aged 25 years or older [45]. Data estimated a 2.5-fold higher HIV incidence in YMSM when compared to older MSM in the city of Atlanta, Georgia, United States [46], and even higher incidence among young Black MSM [47]. In Brazil, researchers detected an increase in HIV prevalence among MSM from 2009 to 2016 and hypothesized that this was due to a higher proportion of younger participants enrolled in the later surveys [48]. Also, a web-based survey conducted in 2018 among 7,055 YMSM found a high self-reported HIV prevalence, reaching 15.3%, 8.4% and 7.7% in Peru, Mexico, and Brazil, respectively [21].

YMSM may be more vulnerable to HIV infection than older MSM in LAC [2, 13,14,15]. Riskier sexual practices, low HIV risk perception, and adverse social economic disparities may be drivers of risk for this young population. In Brazil, pooled data from RDS surveys conducted in 12 cities in 2016 found that YMSM were more likely to have condomless receptive anal sex than older MSM (41.9% and 29.7%, respectively, p < 0.001) [49]. Among YMSM included in this analysis, condomless receptive anal sex was associated with poor self-rated health status, high risk perception, gay/homosexual sexual identity, and transactional sex. In addition, a temporal trend of increased risk behavior was observed among Brazilian YMSM from 2009 to 2016, including higher numbers of sexual partners, more frequent condomless insertive and receptive anal sex, and more frequent use of illicit drug [48]. Additionally, a Brazilian national web-based survey conducted in 2017 showed that YMSM were more likely to experience condomless receptive anal sex and transactional sex, but had lower perceived HIV risk, lower HIV testing, and lower PrEP awareness compared to older MSM [50]. Still in Brazil, younger age (18 to 24 years) increased the odds of high-risk behavior among 16,667 MSM who responded to web-based surveys from 2016 to 2018 [11]. Similar findings were observed in a Nigerian study, which observed an association between decreasing age and increased likelihood of condomless anal sex, no prior testing for HIV, and having rectal gonorrhea [45].

None of the studies included in this review evaluated factors associated with HIV prevalence specifically among YMSM, including socioeconomic disparities. Countries in LAC have among the highest wealth inequalities in the world [51], and social determinants are related to adverse health outcomes [52]. For example, low-income was associated with self-reported HIV prevalence among Brazilian and Peruvian YMSM [21]. Among PrEP users in a demonstration project in Brazil, Black YMSM with low income were less likely to have optimal ART adherence [53]. Also in Brazil, racial minorities status and lower education, which are often associated with poverty, were related to worse outcomes along the HIV care continuum, including HIV status awareness, being on ART, and virologic suppression [54]. Taken together, findings corroborate that social determinants most likely contribute to the increased HIV risk among YMSM. Echoing our findings, a systematic review in the United States identified that YMSM were more likely to experience depression, polysubstance use, low income, decreased health care access, and earlier sexual debut than older MSM. The authors also observed that YMSM living with HIV were less likely to be aware of their infection, on ART, or virologically suppressed [43].

Although several studies have found a positive association between older age and HIV infection among MSM in LAC [4, 5, 29, 31, 37]. This finding is expected as prevalence includes cumulative cases over longer periods of time for older persons. HIV prevalence among young persons may be a proxy for HIV incidence, considering that lifetime sexual exposure is shorter and risks leading to acquiring infection are more recent among YMSM. In support of this hypothesis, the study in El Salvador found a three-fold higher proportion of recent infection by the BED-CEIA assay among YMSM compared to older MSM (34.8% versus 9.6%, p < 0.01) [37].

This systematic review has several limitations. First, our original goal was to assess HIV prevalence among YMSM living in LAC and its trend over the last 10 years. The rationale used to select the systematic review study period was to capture data that could describe the state of HIV epidemic pre- and post- major HIV public health policies such as TasP and PrEP and how they impacted the HIV epidemic. Unfortunately, no studies were conducted solely with YMSM. Second, in order to have comparable research findings with probability-based methods, we initially limited our review to studies using RDS. We widened our search to other sampling and recruitment methods as none of the RDS studies targeted YMSM specifically. The wider inclusion allowed for other probabilistic methods (such as TLS, stratified sampling among military conscripts, and household surveys) and non-probabilistic methods. Third, the scarcity of data on YMSM did not allow to assess factors associated with HIV infection, nor to evaluate the HIV care continuum specifically among YMSM. Fourth, some of the studies included trans women. Their inclusion may have increased the HIV prevalence as trans women have an estimated global HIV prevalence higher than for MSM [55]. We acknowledge that trans women are a different population, often mistakenly grouped with MSM. Despite limitations, this study provides a compendium of knowledge on HIV prevalence among YMSM in LAC published in four languages during the last decade and sheds light to the gaps in the literature for this highly vulnerable population.

Conclusions

Current data demonstrate that HIV prevalence among YMSM in LAC is high. However, there is a dearth on studies specifically focusing on HIV prevalence and its associated factors specifically among YMSM. Areas requiring further investigation include socioeconomic and racial disparities, sexual behavior and sexual networks, and individual and structural barriers along the HIV care and prevention continuums. Studies aiming to address those gaps will need to specifically prioritize YMSM using rigorous sampling methods. There is an urgent need for larger epidemiological studies focusing on YMSM to support strategies and policies aiming to decrease HIV incidence in this highly vulnerable group.