HIV Risk Behaviors and the Use of Geosocial Networking Dating Apps among Men Who Have Sex with Men and Transgender Women in Lima, Peru: a Cross-Sectional Study

Abstract

Geosocial networking dating apps (GSN apps) are an increasingly widespread technology used by populations throughout the world to facilitate sexual encounters. Studies from a variety of settings suggest a possible association between GSN app use and HIV risk behaviors, including among sexual and gender minority populations such as men who have sex with men (MSM) and transgender women (TW). However, it remains unclear to what extent GSN apps play a causal role. We explored the relationship between GSN app use and sexual risk behaviors among MSM and TW in Lima, Peru by analyzing data from a multi-site cross-sectional survey assessing both general and partner-specific sexual behaviors. We performed bivariate analysis to estimate the association of GSN app use with different individual and partner-specific factors, then fit multivariable regression models adjusting for age and education. Among 741 total participants (698 MSM, 43 TW), 64% met at least one sex partner in the prior three months using a GSN app. GSN app users were significantly more likely to report engaging in HIV risk behaviors in general, including condomless receptive anal sex, group sex, transactional sex, and sex under the influence of alcohol or drugs. Having condomless anal sex with a given partner was not associated with meeting that partner via GSN app. These findings highlight GSN app users as a particularly vulnerable subpopulation among MSM and TW in Lima. GSN apps could provide a useful vehicle for targeted HIV prevention efforts for priority populations in Peru.

Introduction

Innovative strategies to promote engagement in HIV testing and prevention are crucially needed to address the ongoing high HIV incidence observed among men who have sex with men (MSM) and transgender women (TW) in Peru. As in most of Latin America, marginalized sexual and gender minority communities such as these are disproportionately impacted by the HIV epidemic. An estimated 10% of MSM and more than 30% of transgender people in Peru are living with HIV [1], compared to a prevalence of 0.4% in the general population [2]. The vast majority of new HIV cases also occur among MSM and TW, but rates of HIV testing – the gateway to both treatment and prevention services – remain low. One study found that as few as one in four Peruvian MSM or TW who are living with HIV have accessed testing and are aware of their status [3]. In June 2021, the United Nations Member States reaffirmed their commitment to end the HIV/AIDS epidemic by 2030, outlining a key goal of ensuring that 95% of people at risk for HIV have access to effective prevention options. The U.N. declaration highlights the importance of using data-driven approaches to understand local epidemiological context and inform targeted HIV prevention interventions tailored to the needs of priority populations at highest risk [4]. In order to maximize the reach of HIV testing and prevention services among MSM and TW in Peru, novel approaches are needed to identify and engage those who may be at particularly high risk for HIV acquisition or transmission.

In Peru, young MSM and TW are at higher risk for HIV infection and are also more likely to be utilizing modern modes of partner-seeking, such as geosocial networking applications (‘GSN apps’)[5]. Approximately 65% of new HIV diagnoses in Peru are in men under the age of 30, and approximately 40% of diagnoses in men younger than 25 [6]. Concurrently, these same populations are increasingly using online platforms to find sex partners [5,7,8]. Recent studies have shown a predominance of adolescents (<20 years of age) and young adults (20 to 25 years) using apps and online platforms, such as GSN apps, to find relationships, sex partners, and friends [5, 9,10]. GSN apps use a cellphone’s real time global positioning system (GPS) data to connect users based on geographic proximity, as well as other characteristics such as shared interests or social community membership [7]. Grindr is the most frequently used GSN app among MSM globally; developed in 2009, it now has over 13 million monthly users worldwide [11]. Notably, in their recent “Grindr Unwrapped” report, Peru was among the countries with the highest percentage of users identifying as a “pasivo” or “bottom” (i.e., receptive anal sex role) [12]. The strong association between receptive anal sex and HIV acquisition risk [13] motivates further exploration of the potential link between GSN apps and HIV-related sexual risk behaviors among GSN app users in Peru.

Previous investigations of the relationship between GSN app use and HIV risk behaviors have generated mixed findings, and there is a notable paucity of evidence from Latin America. A number of studies have demonstrated higher rates of HIV-related sexual risk behaviors [14–19], as well as bacterial sexually transmitted infection (STI) diagnoses [20,21], among MSM who use GSN apps. Interestingly, there is also evidence that condom use among GSN app users may be similar to, or even higher than, non-GSN app using counterparts [22–24]. A recent study by Knox et al of GSN app-using MSM in China found that individuals were less likely to engage in condomless sex in partnerships that were initiated online as opposed to in person [25]. Conversely, a study conducted in Peru found that MSM and TW who used social media (including GSN apps) for partner seeking were more likely to engage in high-risk behaviors such as increased number of partners, group sex, and condomless anal sex [5].

In Lima, as in much of Latin America, heteronormativity is pervasive and impacts both sexual behaviors and HIV prevention access among marginalized sexual and gender minority populations such as MSM and TW [26]. In a survey conducted among sexual and gender minorities in Peru, more than 70% of participants identifying as a gender or sexual minority reported that they had experienced violence or discrimination [27]. GSN apps can provide a sense of community and acceptance that may help to destigmatize sexuality for members of these communities [9,28]. Although LGBTQ+ virtual spaces have existed for decades, GSN apps allow individuals to meet potential partners without attending LGBTQ+ oriented physical venues, removing a significant barrier for many [28,29]. Moreover, GSN apps have the potential to not only identify high-risk populations but also act as a vehicle for HIV prevention interventions [10, 30–32]. In Peru specifically, studies have demonstrated the acceptability and feasibility of social media interventions and other online outreach for HIV prevention [33–35], supporting a possible role for GSN apps to deliver future interventions. These apps are now an integral component of the cultural landscape, and research is needed to better understand how their use may influence HIV risk behaviors in different populations with the goal of informing HIV prevention outreach strategies.

In this study, we analyzed data from surveys of 741 MSM and TW to explore how, and to what extent, self-reported GSN app use is associated with HIV risk behaviors (condomless anal sex, group sex, transactional sex, substance use around sex, having multiple partners) among MSM and TW in Lima, Peru. We assessed the relationship of GSN app use to sexual risk behaviors both at the level of the individual GSN app user and in the context of specific GSN app-initiated sexual partnerships.

Methods

Study Design, Setting, and Participants

Data were collected as part of a cross-sectional survey designed to evaluate patterns of social venue attendance and associated sexual behaviors, including the use of online platforms to meet sex partners, among MSM and TW in Lima, Peru. Eligible individuals were at least 18 years of age and self-identified as either MSM or TW. The survey was implemented in two different settings: 1) online, and 2) in-person at Vía Libre, a non-government clinic providing comprehensive HIV services, located in central Lima. Participants in the online survey were recruited from 11/26/2018 to 5/16/2019 through targeted outreach on social media platforms affiliated with Epicentro, a LGBTQ+ community-based organization located in the Barranco district of Lima. Online survey participation was anonymous and uncompensated; individuals accessed the survey link and completed the survey on their own personal devices. In-person survey participants were recruited from 2/19/2019 to 3/13/2020 among individuals presenting to Vía Libre for voluntary HIV testing and counseling. Participants completed the survey by computer-assisted self-interview (CASI) using a dedicated study tablet, prior to receiving their HIV test result from the clinic that day, and a small compensation (equivalent to ~5 US Dollars) was offered for participation. As a safeguard to prevent duplicate enrollment in the in-person survey, we also collected participants’ Peruvian national identity number and cross-checked this against the database of existing study records to verify the unique identity of new participants.

Data Sources and Measures

All data collection was performed in REDCap [36]. The same survey instrument was used in both the online and in-person study settings, and has previously been described in detail in the context of the primary analysis of the online survey data [37]. In brief, participants were asked about their sexual behaviors during the previous three months, including the use of online platforms to meet sex partners. In addition to asking about sexual behaviors in reference to one’s cumulative sexual experience over the 3-month recall period, participants were asked a series of partner-specific questions pertaining to each of their two most recent partners, including: partner type (e.g., casual, stable, transactional), how/where they met that partner, where they had sex, sexual role/positioning, and condom use. As such, the survey assessed participants’ sexual behaviors – including GSN app use – both in general (i.e., with any sex partner in the prior three months) and in reference to specific partners during the same 3-month recall period. This design enabled us to separately evaluate both the general, participant-level behaviors associated with being a ‘GSN app user’ as well as the specific, partner-level behaviors associated with having met a given partner using a GSN app. The survey also included questions about past experience with HIV testing and awareness of biomedical HIV prevention modalities (e.g., pre-exposure prophylaxis [PrEP], Undetectable = Untransmittable [“U=U”]). Of note, PrEP was not widely available in Peru at the time of this study; as of March 2020, when our study ended, it is estimated that fewer than 1,300 individuals, cumulatively, had ever received PrEP in the country [38]. Prior to 2023, when oral tenofovir-based PrEP was endorsed for the first time as part of Peru’s national HIV prevention guidelines, access was limited to demonstration studies or private sector purchase [39]. Additionally, for the in-person survey only, we obtained the results of participants’ HIV screening test administered at the Vía Libre clinic on the date of their survey. For online participants, HIV status was ascertained solely based on self-report.

To analyze the relationship between participants’ recent sexual behaviors and their use of GSN apps in general, we classified participants as either ‘GSN app user’ or ‘GSN app non-user’ based on whether or not they reported meeting any sex partner in the prior three months using a GSN app. Similarly, to analyze sexual partnership-level characteristics and behaviors associated with meeting a given partner via GSN app, the recent partners for whom participants reported partner-specific data (i.e., their last and/or penultimate partners in the prior three months) were classified as follows: ‘GSN partner’ (those met via GSN app), ‘online (non-GSN) partner’ (those met online, but in a venue not defined as ‘GSN app’), or ‘offline partner’ (those met in person at a physical venue). The following seven platforms were defined as ‘GSN app’: Grindr, Hornet, Scruff, Tinder, Badoo, Growlr, and Surge. All other online platforms were considered ‘non-GSN’; these included Manhunt, GayRomeo, PlanetRomeo, WhatsApp, Facebook, and Instagram.

Statistical Analysis

All analyses were performed using RStudio version 1.3.1056. We used descriptive statistics to summarize participant characteristics, including sexual behaviors reported during the prior three months. Differences in demographic characteristics comparing GSN app users to GSN app non-users were assessed by χ2 test for categorical variables and Student’s t-test for continuous variables.

The primary objective of this study was to assess whether GSN app users were more likely than non-users to engage in sexual behaviors known to be associated with increased HIV transmission or acquisition risk. We performed bivariate analyses to estimate crude prevalence ratios (PR) and 95% confidence intervals (CI) comparing the proportion among GSN app users relative to GSN app non-users who reported engaging in different HIV-associated sexual risk behaviors in the last 3 months (e.g., condomless anal sex [CAS], group sex, transactional sex, sex under the influence of alcohol or drugs, having multiple partners), and with respect to their past HIV testing history (including HIV status, stratified by survey setting) and awareness of biomedical HIV prevention modalities. In addition, we estimated adjusted prevalence ratios (aPR) and CI for the same set of comparisons, using generalized linear models with a Poisson distribution and robust standard errors, adjusted for age (± 30 years old) and education (± university degree).

As a secondary objective, we sought to evaluate the sexual partnership-level characteristics and partner-specific HIV risk behaviors (i.e., CAS reported with that partner) associated with meeting a given partner using a GSN app. With sexual partners as the unit of analysis, we fit generalized linear models using a Poisson distribution and robust standard errors to estimate crude and adjusted prevalence ratios and CI comparing GSN, online (non-GSN), and offline partners with respect to the following partner-level variables: partner gender, partner type, and whether CAS was reported with that partner. To account for autocorrelation resulting from the fact that each participant could report multiple partners (up to a maximum of two), we used generalized estimating equations (GEE) with an independent correlation structure. Multivariable models included participant age and education as covariables.

Sensitivity analyses were conducted to assess for potential sources of bias in our primary analyses assessing HIV risk behaviors among GSN app users versus non-users. First, to evaluate for potential confounding by gender in our estimates, we stratified by MSM versus TW. In addition, to assess for possible differences in these estimates based on survey setting, we conducted separate sub-analyses stratifying by online versus in-person participants.

Results

Participant Characteristics

A total of 751 individuals completed the survey, 10 of whom were excluded from the analysis population because they did not identify as either MSM or TW. Of the remaining 741 eligible participants included in this analysis, 397 (54%) completed the survey in the online setting and 344 (46%) in the in-person setting (Table 1). Overall, 698 (94%) self-identified as MSM and 43 (6%) as TW. A total of 477 (64%) participants reported meeting at least one sex partner using a GSN app in the prior three months. Compared to participants who did not report using a GSN app to meet a recent sex partner, GSN app users were younger, more likely to identify as MSM (vs. TW) and as homosexual (vs. bisexual or “other”), and on average had higher overall educational attainment. Income and living situation were similar between GSN app users and non-users. Among online survey participants, self-reported HIV prevalence was 26%; the vast majority of these individuals (94%) reported they were currently taking antiretroviral therapy (ART). Among in-person survey participants, HIV prevalence was 13% based on the results of contemporaneous HIV testing at Vía Libre.

Table 1:

Participant Characteristics

	OVERALL N=741	GSN APP USERS N=477 (64%)	GSN APP NON-USERS N=264 (36%)	p-value a
Survey setting, n (%)				<0.01
Online	397 (54%)	228 (48%)	169 (64%)
In-Person (VÍa Libre)	344 (46%)	249 (52%)	95 (36%)
Age (years), median (IQR)	28 (23–34)	27 (23–32)	31 (25–39)	<0.01
Gender category, n (%)				<0.01
Cisgender MSM	698 (94%)	461 (97%)	237 (90%)
Transgender woman	43 (5.8%)	16 (3.4%)	27 (10%)
Educational attainment, n (%)				<0.01
Less than secondary	37 (5.0%)	13 (2.7%)	24 (9.1%)
Complete secondary	127 (17%)	74 (16%)	53 (20%)
Partial post-secondary	250 (34%)	175 (37%)	75 (28%)
Complete post-secondary	327 (44%)	215 (45%)	112 (42%)
Monthly income, n (%)				0.91
≥1500 Soles b	252 (34%)	161 (34%)	91 (34%)
Living situation, n (%)				0.24
Lives alone	143 (19%)	89 (19%)	54 (20%)
With friend/roommate(s)	52 (7.0%)	39 (8.2%)	13 (4.9%)
With family (including spouse/partner)	546 (73%)	349 (73%)	197 (75%)
Sexual identity, n (%)				<0.01
Gay/homosexual	542 (73%)	366 (77%)	176 (67%)
Bisexual	159 (21%)	93 (19%)	66 (25%)
Other	40 (5.4%)	18 (3.8%)	22 (8.3%)

GSN, geosocial networking; IQR, interquartile range; MSM, men who have sex with men

^a.Calculated using Student’s t-test for continuous variables and chi-square test for categorical variables

^b.Exchange rate on 26 Nov 2018: 1 USD = 3.37 Peruvian Nuevos Soles (https://www1.oanda.com/currency/converter)

Association of GSN App Use with Sexual Behaviors, HIV Prevention Awareness, and HIV Testing History

Compared to GSN app non-users, GSN app users were significantly more likely to report a number of HIV-related sexual risk behaviors, including group sex, transactional sex, substance use before or during sex, and having one or more “casual” partner in the prior three months (Table 2). GSN app users also reported a higher number of total sex partners over this time compared to non-users. After adjusting for age and education, GSN app users were still more than twice as likely to have ≥5 recent sex partners, and were more likely to report group sex, transactional sex, having a “casual” partner, and substance use linked to sex (Table 2). In addition, GSN app users were more likely to report CAS (any form) with a recent partner – as well as receptive CAS specifically. However, when adjusted for age and education, this association remained statistically significant only for receptive CAS, and not for CAS in general. Awareness of biomedical HIV prevention strategies was also higher among GSN app users compared to non-users. GSN app users were more likely to indicate they had ever heard of PrEP, as well as to agree with a statement describing the concept of “U=U” when presented in the form of a ‘True/False’ question. With respect to prior experience with HIV testing, GSN app users were slightly more likely to report having ever received an HIV test before. Among participants who did not report an existing HIV diagnosis prior to the study, GSN app users were more likely to report having received HIV testing within the last 6 months. There was no significant difference in HIV prevalence between GSN app users and GSN app non-users in either the online or in-person survey setting.

Table 2 –

ASSOCIATION OF PARTICIPANT-LEVEL GSN APP USE WITH SEXUAL RISK BEHAVIORS, HIV PREVENTION AWARENESS, AND HIV TESTING HISTORY

	Overall (N=741)	GSN App Users (N=477)	GSN App NON-Users (N=264)	Crude Prevalence Ratio	Adjusted a Prevalence Ratio
SEXUAL BEHAVIORS (last 3 months)	n (% of N)	n (% of N)	n (% of N)	PR (95% CI)	aPR (95% CI)
≥5 Partners	388 (52%)	307 (64%)	81 (31%)	2.10 (1.73 – 2.55)	2.14 (1.76 – 2.61)
Group Sex	327 (44%)	255 (53%)	72 (27%)	1.96 (1.58 – 2.43)	1.96 (1.57 – 2.44)
Substance Use Before/During Sex	354 (48%)	248 (52%)	106 (40%)	1.29 (1.09 – 1.54)	1.28 (1.08 – 1.53)
Transactional Sex	292 (39%)	203 (43%)	89 (34%)	1.26 (1.03 – 1.54)	1.31 (1.06 – 1.60)
Casual Partner b	446 (60.2)	331 (69%)	115 (44%)	1.59 (1.37 – 1.85)	1.62 (1.39 – 1.89)
Condomless Anal Sex (CAS) c	381 (51%)	260 (55%)	121 (46%)	1.19 (1.02 – 1.39)	1.14 (0.97 – 1.33)
Receptive CAS	268 (36%)	192 (40%)	76 (29%)	1.40 (1.12 – 1.74)	1.28 (1.02 – 1.60)
HIV PREVENTION AWARENESS
Ever Heard Of PrEP	473 (64%)	341 (72%)	132 (50%)	1.43 (1.25 – 1.63)	1.44 (1.26 – 1.65)
Believes U=U Is True d	320 (43%)	224 (47%)	96 (36%)	1.29 (1.07 – 1.56)	1.28 (1.07 – 1.55)
HIV TESTING HISTORY
Ever Had HIV Test	639 (86%)	421 (88%)	218 (83%)	1.07 (1.00 – 1.14)	1.09 (1.02 – 1.17)
Last HIV Test ≤6 Months Ago e	338 (53%)	237 (57%)	101 (46%)	1.24 (1.05 – 1.46)	1.24 (1.05 – 1.47)
HIV STATUS f	Overall	GSN App Users	GSN App Non-Users
Online Survey	(N=397)	(N=228)	(N=169)
HIV-Positive (Self-Report)	103 (26%)	59 (26%)	44 (26%)	0.99 (0.71 – 1.39)	1.05 (0.75 – 1.47)
On ART (Of N=103 Self-Reported HIV+)	97 (94%)	54 (92%)	43 (98%)	0.94 (0.85 – 1.03)	0.96 (0.88 – 1.04)
In-Person Survey	(N=344)	(N=249)	(N=95)
HIV-Positive (Rapid Test)	43 (13%)	33 (13%)	10 (11%)	1.26 (0.64 – 2.47)	1.25 (0.61–2.57)

GSN, geosocial networking; PR, prevalence ratio; CI, confidence interval; PrEP, pre-exposure prophylaxis; ART, antiretroviral therapy

^a.Adjusted for age (± 30 years) and educational attainment (± completed university or higher)

^b.Refers to whether last/penultimate partner (in the past 3 months) was categorized as “casual”

^c.Refers to whether condomless anal sex was reported with last/penultimate (in the past 3 months)

^d.Responded “True” when presented with the following True/False statement: “It is very unlikely that a person with HIV will transmit the virus to their sexual partner if the person with HIV is taking antiretroviral therapy and the virus is undetectable in their blood”

^e.Among the n=638 who did not report having an existing HIV diagnosis

^f.Stratified by survey setting: for online survey, HIV status (and ART use) was ascertained by self-report; for in-person survey, used result of HIV screening test from Vía Libre clinic

Association of GSN App Use with Sexual Partnership Characteristics and Partner-Specific Risk Behaviors

Overall, the 741 participants included in our analysis population provided partnership-level data on a total of 1314 specific sexual partners. Of these, 44% were GSN partners, 17% were online (non-GSN) partners, and 39% were offline partners (Table 3). GSN partners, compared to online (non-GSN) or offline partners, were more likely to be cisgender men (98% vs. 91% vs. 90%, respectively), and to be described as a “casual” partner (62% vs. 35% vs. 51%). The prevalence of CAS was similar across the three partner categories (41% vs. 45% vs. 45%). In both the crude and adjusted GEE models, no significant association was observed between meeting a partner via GSN app and the likelihood of engaging in CAS with that partner.

Table 3:

ASSOCIATION OF SEX PARTNER MEETING VENUE WITH PARTNERSHIP-LEVEL CHARACTERISTICS AND CONDOMLESS ANAL SEX

Partnership-Level Characteristic		Sex Partner Meeting Venue			Prevalence Ratio Estimates (GEE)
	Overall N=1314	GSN App N=576 (44%)a	Online (Non-GSN) N=226 (17%)b	Offline N=512 (39%)c	Comparison	PR (95% CI)	aPR d (95% CI)
Partner Gender, n (%)
Cisgender Man	1232 (94%)	563 (98%)	206 (91%)	463 (90%)	(GSN vs. Offline)	1.09 (1.05 – 1.12)	1.08 (1.06 – 1.13)
					(Non-GSN vs. Offline)	1.01 (0.96 – 1.06)	1.01 (0.96 – 1.06)
					(GSN vs. Non-GSN)	1.07 (1.03 – 1.12)	1.07 (1.03 – 1.12)
Partnership Type, n (%)
“Casual” Partner	696 (53%)	355 (62%)	79 (35%)	262 (51%)	(GSN vs. Offline)	1.21 (1.09 – 1.35)	1.20 (1.07 – 1.34)
					(Non-GSN vs. Offline)	0.68 (0.56 – 0.83)	0.68 (0.56 – 0.83)
					(GSN vs. Non-GSN)	1.77 (1.46 – 2.13)	1.75 (1.45 – 2.12)
Condomless Anal Sex, n (%)
Had CAS with partner (any form)	569 (43%)	238 (41%)	102 (45%)	229 (45%)	(GSN vs. Offline)	0.93 (0.81 – 1.06)	0.90 (0.78 – 1.03)
					(Non-GSN vs. Offline)	1.01 (0.85 – 1.20)	1.00 (0.84 – 1.18)
					(GSN vs. Non-GSN)	0.92 (0.77 – 1.09)	0.75 – 1.07)

GEE, generalized estimating equations; GSN, geosocial networking; PR, crude prevalence ratio; aPR, adjusted prevalence ratio; CAS, condomless anal sex

^a.Specific apps used to meet n=576 ‘GSN app’ partners: Grindr 538 (93%); Tinder 22 (3.8%); Scruff 8 (1.4%); Badoo 6 (1.0%); Growlr 1 (<1%); Surge 1 (<1%)

^b.Specific online platforms used to meet n=226 ‘non-GSN’ partners: Facebook 137 (61%); WhatsApp groups 33 (15%); Manhunt 22 (9.7%); “chat” (e.g., Peru Gay Chat, Gay Chat, Chat Gay, Chat Peru Gay, El Chat) 13 (5.8%); GayRomeo 1 (<1%); other 17 (7.5%)

^c.Physical venues where participants met n=512 ‘offline’ partners: bar/disco 106 (21%); public outdoor space (e.g., plaza, park, mall, beach) 96 (19%); sauna 92 (18%); private party 46 (9.0%); at work/school 34 (6.6%); internet “cabinas” 27 (5.3%); gym or athletic field 23 (4.5%); sex club 19 (3.7%); public bathroom 17 (3.3%); adult movie theater 10 (2.0%); hair salon 7 (1.4%); other 35 (6.8%)

^d.All multivariable estimates are adjusted for following variables: age (± 30 years) and education (± completed university or higher)

Sensitivity Analyses

For our primary analyses assessing participant-level factors associated with GSN app use, stratifying by gender (MSM vs. TW) produced prevalence ratio estimates for MSM (n=698) that were similar to the overall combined estimates, whereas stratum-specific point estimates for TW (n=43) deviated more notably from the original combined estimates, consistent with the much smaller sample size of this group (Supplementary Table 2-a). The pattern of these changes, as well as the relatively wide confidence intervals of the stratum-specific estimates, which had considerable overlap with those of the unstratified estimates, was not indicative of confounding by gender. Rather, there was evidence of possible effect modification for certain factors (e.g., ≥5 sex partners). Considering the limited sample size and consequent wide confidence intervals around the TW-specific estimates, both TW and MSM were included in the final analysis. Similarly, stratifying by survey setting (online vs. in-person) generated relatively modest differences in the stratum-specific estimates (Supplementary Table 2-b), which in some cases fit a pattern of effect modification, but there was no evidence of confounding by this factor. Therefore, we did not adjust for survey setting in the multivariable analyses, and participants from the online and in-person survey settings were analyzed together.

Discussion

In this study evaluating the relationship between GSN dating app use and sexual behaviors among MSM and TW in Peru, we found that GSN app use was highly prevalent. Of 741 total survey participants across two study settings in Lima, nearly two thirds reported using a GSN app to meet a sexual partner in the previous three months. GSN apps were also the most frequently reported method of meeting a partner (among 1314 total last/penultimate partners referenced) as compared to either of the two other partner meeting venue categories we assessed (i.e., non-GSN online venues, or offline venues). GSN app users were significantly more likely to report they engaged in a variety of HIV-related sexual risk behaviors in the prior three months, including group sex, transactional sex, sex with a “casual” partner, CAS, substance use with sex, and higher total number of partners.

Interestingly, when we re-framed the analysis to evaluate partner-specific factors – focusing instead on the GSN app-initiated sexual partnership, rather than the individual GSN app user, as the unit of analysis – we found that the likelihood of engaging in CAS was not significantly different with GSN app partners compared to partners met via other non-GSN online platforms or those met offline – in contrast to the significantly higher prevalence of CAS among GSN app users in general that we observed in our primary analysis. These findings are consistent with a study by Knox et al., which found that when controlling for the user by employing a case-crossover study design, individuals were less likely to engage in CAS with partners met online compared to those met offline [25]. Additionally, previous studies have found a dose-response relationship between the overall number of venues (irrespective of venue type) used to meet partners and the frequency of CAS, suggesting that it is not the characteristics of the venue that determine engagement in CAS [8,22,40], but rather the characteristics and motivations of the individual (e.g., intention to have sex with many partners) that leads to GSN app use. Consistent with this explanation is our finding that GSN app users had a higher number of sexual partners, a trend also seen in other studies of GSN app users [16, 39–40]. In other words, it may not be that the GSN app itself influences behavior, but rather that the app attracts a subset of individuals who are looking for multiple partners, thereby leading to more opportunities for high-risk sexual behaviors such as CAS. Because GSN apps do not appear to be causal to engagement in high-risk behaviors, but rather their usage is a characteristic of populations that are at risk, there is an opportunity to leverage their popularity for HIV prevention efforts - a strategy that studies have found to be widely accepted in settings including Peru [31–35,43,44].

This study has several limitations, including its cross-sectional design and the inherently exploratory nature of our analysis. The survey instrument was structured in such a way that certain sexual behaviors – most notably, those pertaining to sexual role and condom use – were only ascertained in reference to the participant’s two most recent partners (limited to the prior three months), rather than considering all partners over this time. As such, our definition of participant-level CAS would not have captured cases of CAS with an antepenultimate (or earlier) partner – unless CAS also occurred with one’s last or penultimate partner. Because of this potential misclassification, we may have underestimated the true prevalence of CAS as determined based on the totality of one’s sexual encounters over the prior three months. Since GSN app users reported a significantly higher total number of recent partners relative to non-users – consistent with multiple other studies [14, 40, 41, 43] – any such misclassification would be expected to result in a more profound underestimation of CAS prevalence among GSN app users compared to non-users, thereby leading to a null bias in our estimates of association for participant-level CAS and GSN app use. Additionally, it worth recognizing that CAS is an imperfect measure of one’s true HIV transmission/acquisition risk, which is also influenced by the biomedical prevention strategies (e.g., PrEP, U=U) used by individuals and their sexual partners, among other factors. Therefore, our use of CAS as a proxy for HIV risk could potentially overestimate the true degree of increased risk among GSN app users relative to non-users – if, for example, GSN app users were also more likely to be taking PrEP (if HIV-negative) or virologically suppressed (if living with HIV). Although we did find evidence that GSN app users were more informed about existing HIV prevention strategies such as PrEP and U=U, considering the limited availability of PrEP [38] and relatively low knowledge of HIV status among MSM and TW in Peru at the time of this study [3], we expect that protection via these biomedical prevention methods was minimal in our study population. Another potential limitation is the relatively short 3-month recall period for which participants were asked to report their recent sexual behaviors. Although a short reference window can help to mitigate recall bias, the 3-month ‘snapshot’ we used may not accurately reflect the full range of one’s past or future sexual behaviors in general. Additionally, our survey instrument was not designed to ascertain the timing of partner-specific sexual behaviors (e.g., CAS) relative to when a given partnership was initiated during the 3-month recall period, limiting our ability to draw robust conclusions about the influence of partner meeting venue on sexual behaviors such as CAS. Future studies could address these limitations by eliciting information about additional sexual partners beyond just the two most recent, or by extending the recall period to more than three months. Alternatively, future investigations could include longitudinal prospective cohorts who report on their sexual encounters which would address these limitations and also minimize recall bias. Finally, because participation in the online survey was completely anonymous, there exists a small risk of duplicate survey responses since, in theory, a participant who completed the online survey could have again taken then survey in the in-person setting. However, we aimed to minimize this risk by having study staff ask prospective participants as part of screening and eligibility assessment procedures, to confirm that they had not already participated in the online survey.

With regard to the generalizability of findings, several additional considerations bear mentioning. First, although both MSM and TW were included in this study, only a small number of TW were ultimately recruited and available for analysis, leading to relatively wide confidence intervals around the TW-specific estimates in our gender-stratified sensitivity analyses. Therefore, the extent to which our results can be applied specifically to TW is unclear as experiences of TW in Lima may differ from those of MSM, for example due to heightened stigma and a larger proportion of the population that engage in sex work [46, 47]. Despite this limitation, overall this study serves to advance understanding of an increasingly widespread phenomenon – the use of GSN dating apps – among sexual and gender minority communities in Peru. The vast majority of previous studies on this topic have focused on populations outside of Latin America. Contextual factors, for example stigma experienced by MSM and TW in Lima, influence sexual behavior and are not uniform across geographic regions. Although there have been studies in other populations where stigma is pervasive [36], there is a need for data relevant to Latin America, and more specifically Peru, to appropriately guide regional HIV prevention interventions. It is also worth noting that our research was conducted before the COVID-19 pandemic, and the use of GSN-apps, as well as dating in general, were unquestionably influenced by the pandemic [42].

Additionally, this study has several strengths related to the methodology and analytic approaches employed, which enhance the validity and relevance of our findings. First, the potential for social desirability bias was minimized by the self-administered survey format used in both study settings, and by the anonymous nature of the online survey specifically. Second, recruiting participants in two distinct settings (in-person and online) likely enabled us to capture a more broadly representative sample of MSM and TW in Lima than would have been possible with either one of these approaches alone, thereby strengthening the external validity of our findings. The separate, albeit overlapping, recruitment timeframes in the online and in-person study settings may also account for some differences in these sample populations included in each of these settings. Although our recruitment methods represent an overall strength, it is worth noting that both approaches may select for populations with greater access to HIV prevention resources relative to the general population of MSM and TW in Lima, particularly within the in-person cohort, as participants were recruited from a subset of individuals seeking HIV testing. Notably, individuals outside of our study population may have less access to HIV prevention resources, and therefore could require distinct, though equally important, HIV interventions. Sensitivity analyses identified some evidence of effect modification by survey setting, although, overall, the sensitivity analyses served to confirm the robustness of our findings across substrata, representing a further strength of our analytic approach. Another notable strength is the multi-level analytic design we employed in this study. By collecting and analyzing data on both individual participant-level and sexual partnership-level factors, we were able to examine the relationship between GSN app use and HIV risk behaviors from two distinct, yet complementary, angles. This allowed us to generate a more nuanced picture of how GSN apps may (or may not) shape sexual interactions in this setting, providing additional insight into a potential mechanism.

In summary, among MSM and TW in Lima, Peru, we found that GSN app users were significantly more likely to report engaging in HIV-related risk behaviors, including CAS – although not necessarily with the same partners they met via GSN apps. Sexual partnerships initiated via GSN app were just as likely to feature CAS as those initiated offline or via non-GSN online platforms. Our results have important practical implications, indicating that GSN app use may be a useful marker to identify individuals at increased risk of HIV transmission or acquisition. GSN apps may provide an ideal medium through which to engage especially vulnerable members of LGBTQ+ communities in Peru, who may be less likely to access traditional HIV prevention services due to concerns about disclosure and public attitudes [48]. Studies from Peru and elsewhere have demonstrated that GSN apps are a feasible method for delivering information about HIV prevention [9,31–35,43,45]. Beyond delivering information, other more ambitious efforts have begun to emerge that leverage GSN apps to directly connect users with HIV services. Grindr, the most widely used GSN app in the United States, recently announced a program to send free HIV self-tests to users who sign up via the app [49,50]. Similar programs could be developed to support access to HIV testing and local prevention services for MSM and TW in Lima, for example, utilizing the apps’ geolocation feature to identify HIV prevention services within close proximity. GSN apps have become commonplace among diverse sexual and gender minority populations globally, and it is important to recognize that sub-groups of MSM and TW exist (early-adopters of GSN apps, bisexual men, ethnic minority groups, etc.) that may have distinct risk profiles and needs. Future studies could seek to better understand the nuanced use of GSN apps among said sub-groups, as well as the specific components of GSN apps (e.g., immediacy, location, photos, etc.) that influence high-risk, as well as protective, sexual behaviors. Ideally, such studies should control for observable confounding factors related to individuals’ patterns or intentions of engaging in high-risk sexual behaviors.

Overall, the results of this study highlight the important role that GSN apps play in facilitating sexual interactions among members of priority populations for HIV prevention in Peru. Our findings also indicate that GSN app users may represent a sub-population of MSM and TW at particularly high risk for HIV in this setting, and who should therefore be prioritized for HIV prevention interventions. Given their primary functionality as platforms for mobile communication, GSN apps may offer a useful medium through which to engage an especially vulnerable group with targeted HIV prevention initiatives.

Conclusions

Among MSM and TW in Peru, those who use GSN apps may be at particularly high risk for HIV, making GSN app users an important sub-population to prioritize for HIV testing and prevention. Future interventions should be developed with this population in mind and could be designed to capitalize on the ubiquity of GSN apps by using them as a platform to deliver tailored HIV prevention information. In Lima, as well as populations across the globe, it is imperative to explore innovative and socially relevant methods for disseminating available HIV prevention tools in order to reach UNAIDS’ goal of access to prevention for 95% of those at risk of HIV infection.

Availability of data.

The datasets analyzed during the current study are available from the corresponding author on reasonable request.