Characterizing the PrEP continuum for Black and Latinx sexual and gender minority youth

Abstract

Pre-exposure prophylaxis (PrEP) rollout efforts thus far have inadequately reached young people from underrepresented backgrounds. This study explores PrEP engagement among young Black and Latinx men who have sex with men, transgender women, and gender diverse individuals in three U.S. cities using an adapted PrEP continuum measure. We analyze data from the recruitment phase of an ongoing PrEP engagement intervention (n=319) using partial proportional odds logistic regression. Participants reported high willingness and intention to use PrEP, yet most (82%) were not currently taking PrEP. Being insured (aOR=2.95, 95%-CI=1.60–5.49), having one or more PrEP users in one’s sexual network (aOR=4.19, 95%-CI=2.61–6.79), and higher individual HIV risk scores (aOR=1.62, 95%-CI=1.34–1.97) were each associated with being further along the PrEP continuum. Strategies are needed to address barriers to healthcare access and leverage connections within social and sexual networks in order to bolster PrEP engagement among marginalized young people from diverse backgrounds.

The Ending the HIV Epidemic initiative identifies prevention as one of the four key pillars of the federal plan to address the HIV epidemic in the United States and highlights PrEP as a primary tool for HIV prevention.(1) Daily oral pre-exposure prophylaxis (PrEP) is an efficacious biomedical method of HIV prevention for young people (ages 15–24 years) at high risk for infection.(2) Despite the many benefits PrEP offers as an HIV prevention modality, PrEP uptake among young people has remained low: in 2018, people 13–25 years of age had the lowest rate of PrEP uptake of any age group aside from those older than 55.(3) Young people, including those from racial/ethnic, sexual, and gender diverse backgrounds, experience disproportionately high HIV incidence, yet they are prescribed PrEP at relatively low rates.(4) These disparities will continue to worsen if HIV prevention experts and providers do not maximize social and structural opportunities to facilitate PrEP access and adoption for underrepresented communities who could benefit from PrEP the most.

Parsons et al.(5) proposed a PrEP continuum that captures the stages involved in PrEP engagement and identifies where gaps emerge. This model is comprised of five stages (i.e., awareness of PrEP, willingness to use PrEP, intention to use PrEP, uptake of PrEP, and persistent use of PrEP) and is useful for exploring how various factors are linked to engagement at different stages of the continuum. Identifying the salient and intervenable factors that distinguish young people’s location at each stage will inform the scale-up of PrEP interventions for underserved HIV-affected communities.

Researchers have suggested that an ecological model is needed to contextualize HIV risk and PrEP engagement within social and structural conditions given the critical role that structural racism and homophobia play in shaping health outcomes.(6–8) This work has highlighted social and economic factors that contribute to co-occurring disparities in HIV incidence and PrEP uptake at the intersections of age, race/ethnicity, gender, and sexual orientation. Prior studies have linked historical and daily experiences of marginalization and discrimination experienced by Black and Latinx people as well as people from sexual and gender diverse backgrounds (e.g., men who have sex with men [MSM] and transgender women) to these communities’ limited engagement in HIV prevention services.(9) These stressors elicit reluctance, particularly among young people, to discuss their sexual orientation and behaviors with medical providers. At the interpersonal level, having supportive relationships within social and sexual networks have been associated with decreased sexual risk behaviors(10) and lower HIV incidence rates,(11,12) yet limited research exists that examines how networks are linked to engagement with PrEP.(13) Additionally, economic vulnerability can hinder access to HIV prevention and may be an important determinant of PrEP engagement along the continuum.(14) Economic factors like insurance coverage, cost of PrEP care, and prescription costs have also been identified as key barriers to PrEP uptake among young Black and Latinx MSM.(15,16)

Given the wealth of evidence that supports examining PrEP engagement across multiple ecological factors, our study aims to describe the PrEP continuum in a sample of young Black and Latinx MSM, transgender women, and gender diverse individuals and explore how economic vulnerability, sexual network-related factors, and individual HIV risk are associated with progress along the PrEP continuum. We hypothesized that 1.) participants with larger sexual networks and participants who perceived PrEP use as normative among their sexual network would show greater engagement with PrEP, 2.) participants with greater individual behavioral risk for HIV would show greater engagement with PrEP, and 3) markers of economic vulnerability would be associated with lower levels of PrEP engagement.

METHODS

Procedures

Data were used from the recruitment phase of the Providing Unique Support for Health (PUSH) study, an on-going randomized trial that tests the effectiveness of a case-management intervention on engagement in HIV treatment and prevention for young Black and Latinx MSM, transgender women, and gender diverse individuals who are living with HIV or are at risk for HIV. Recruitment for the study took place in Philadelphia (PA), Baltimore (MD) and Washington D.C. between August 2017 and March 2020. Participants were recruited via modified respondent-driven sampling with targeted seed selection from venue-based recruitment and online recruitment through social media; similar sampling techniques have been used in other studies of underrepresented populations.(17) Seeds were selected through direct (i.e., information about participation given directly to young people by study staff) and indirect (i.e., advertisements on social media) recruitment methods, and were invited to participate in the study with the option to recruit up to five of their peers. Potential participants were eligible for inclusion if they were: 1) 15–24 years old at the time of recruitment, 2) assigned sex male at birth, (3) self-identified as Black/African American or Hispanic/Latinx race/ethnicity, 4) reported oral or anal sex with a male partner in the past 12-months, and 5) spoke fluent English. The current analysis includes only participants who had a negative HIV test or self-reported HIV negative status at their baseline visit, and thus were potential candidates for PrEP.

A waiver of parental consent was obtained for this study and participants aged 15 to 17 were not required to obtain parental permission to join the study. At one site, the study team employed independent participant advocates to provide counseling to each participant throughout the informed consent process to ensure comprehension of risks and benefits of participating in the research. Consultation with an advocate at that site was required for participants not seeking sexual health services (i.e. HIV testing), and was optional for those seeking sexual health services. Study procedures were approved by the Institutional Review Boards at the Children’s Hospital of Philadelphia, Johns Hopkins University, and Children’s National Medical Center. Full details about recruitment and study protocols have been published elsewhere.(18,19)

Measures

PrEP Continuum

Participants were categorized into five mutually exclusive levels on the PrEP continuum describing their degree of engagement with PrEP. A participant was considered PrEP Unaware (PrEP continuum=0) if they reported that they had never “heard about ‘PrEP’ (pre-exposure prophylaxis) for the prevention of HIV infection in people who are HIV-negative”. Participants who were aware of PrEP but reported being “Very unwilling” or “Unwilling” to use PrEP were considered be in the PrEP Aware stage of the continuum (PrEP continuum=1). Those who reported being aware of PrEP and being “Somewhat willing”, “Willing”, or “Very willing” to take PrEP but reported that they were “Very unlikely” or “Unlikely” to be taking PrEP in the next three months were considered to be in the PrEP Willing stage of the continuum (PrEP continuum=2). Those who reported that they were aware of PrEP, willing to use PrEP, and “Somewhat likely”, “Likely”, or “Very likely” to be taking PrEP in the next three months but were not currently taking PrEP were considered to be in the PrEP Intending stage of the continuum (PrEP continuum=3). Those who reported taking PrEP in the last 30 days were considered to be in the Current PrEP User stage of the continuum (PrEP continuum=4). Participants who had previously taken PrEP but were not currently taking the medication (n= 45, 14.2%) were classified according to their current motivational stage (i.e. PrEP Aware, PrEP Willing, or PrEP Intending).

Network-related Factors

Two factors related to sexual networks were measured in this study. We assessed whether a participant had someone in their sexual network who was using PrEP by asking “Have any of your sexual partners ever told you that they are taking PrEP to prevent HIV-infection?” (0=No, 1=Yes). The size of a participant’s sexual network was measured by assessing the total number of sexual partners, of any gender, reported by the participant in the last three months.

HIV Risk Score

An HIV risk score was calculated based on the various behavioral risk factors suggested by the Centers for Disease Control and other experts for evaluating PrEP indication.(20,21) Our score gave participants a point for each of the following criteria that applied to them: (1) has had a sexual partner who was living with HIV or a partner whose HIV status was unknown to them within the last three months, (2) has ever tested positive for a sexually transmitted infection, (3) has ever had condomless anal intercourse, (4) has had infrequent condom use (i.e., reports using condoms “half of the time” or “never”) with any sexual partner in the last three months, (5) has had multiple sexual partners in the last three months, or (6) reported using injection drugs within the last three months. This yielded a score that could range from 0 to 6, with higher numbers representing greater behavioral risk for HIV risk.

Economic Vulnerability

Economic vulnerability was measured using four distinct metrics. Perceived family wealth was measured by asking participants “How well-off (in a good financial situation) do you think your family is?” and was dichotomized for multivariate analysis (0= “Not at all well off” or “Not so well off”, 1= “Average”, “Quite well off”, or “Very well off”). Housing instability was assessed by asking “In the past 12 months, have you been without a regular place to stay?” (0=No, 1=Yes). Participation in transactional sex was assessed by asking “Have you ever had sex with a male in exchange for money, a place to stay, or food?” (0=No, 1=Yes). Insurance status was collapsed into two categories: 0=No insurance and 1=Any insurance (inclusive of being covered by a parent’s insurance plan).

Demographics

Participants reported their race/ethnicity by selecting one category from the following that best represented their racial/ethnic identity: Black Arabic, Black African (e.g., Nigerian, Ethiopian), Black Caribbean (e.g., Jamaican, Trinidadian etc.), Black Latino, Hispanic, Black Asian, Black White, or African American/Black. These racial/ethnic categories were collapsed into three categories due to small numbers. Those identifying as African American/Black were categorized as “Black”; those identifying as Hispanic were categorized as “Latinx”; those identifying as Black Arabic, Black African, Black Caribbean, Black Latino, Black Asian or Black White were categorized as “other Black identity or multiracial”. Gender identity was collapsed into three categories: cisgender men, transgender women, and gender diverse individuals (inclusive of those identifying as genderqueer, gender questioning, and gender non-conforming). For regression analysis, age was collapsed into three categories: 15 to 18 years, 19 to 21 years, and 22 to 24 years of age based on developmental considerations.

Statistical Analysis

The baseline dataset included 319 participants. Of these, we identified three participants who had missing data for at least one of our variables of interest. We employed list-wise deletion to exclude these missing cases, yielding a final analytic sample of N=316. This sample includes 195 (62%) seed participants who were recruited through venue-based and online recruitment, and 121 (38%) peer-referred participants who returned a coupon they had received from an enrolled participant. Respondent-driven sampling weights were not applied to descriptive analyses nor regression models. This analytic decision was made for several reasons: 1.) the primary purpose for using modified RDS in this study was to engage youth and not to generate population-based estimates, 2.) the lack of scientific consensus on the use of weights in regression models, and 3.) weighting would be inappropriate for this analysis, given the restriction of participants to youth not living with HIV. (17,22)

We calculated descriptive statistics to summarize the sample and characterize participants’ placement along the PrEP continuum. Using the mutually exclusive classification, we compared PrEP continuum placement by gender identity and by racial ethnic identity using Fisher Protected Least Significant Difference tests for multiple comparisons.(23)

Given the ordinal nature of the PrEP continuum variable, we tested a partial proportional odds cumulative logit model to explore the effects of economic vulnerability, sexual network-related factors, and HIV risk score on progress along the PrEP continuum using the five mutually exclusive categories defined. This regression model is an extension of logistic regression that collapses the levels of an ordinal variable with J levels and J-1 cut-points into two categories for each cut-point: all levels at or below the cut-point, and all levels above the cut-point.(24) In this model, we started by assuming that predictor variables had equivalent slopes for each cut-point (i.e. the proportional odds assumption). In order to test the proportional odds assumption, we used the Brant test as well as visual inspection of models that allowed different slopes at different cut-points for one variable at a time.(25,26) For variables that violated the proportional odds assumption, we estimated the slopes for each cut-point of the outcome variable in our final model, yielding a partial proportional odds model. Additionally, we performed a sub-analysis of participants who identified as transgender or in the gender diverse category. In this model, all variables satisfied the proportional odds assumption; thus, a proportional odds model was used for the sub-analysis with a common slope for all cut-points. The inclusion of study site as a control variable did not alter the results of either analysis; as such, it was not included in the final models.

Analyses were performed in R 3.5.1(27) using the ordinal package(28) clm function to fit cumulative logit models for ordinal regression.

RESULTS

Participant characteristics

Most participants identified as non-Hispanic Black or African American (n=168, 53%), a small number identified as Hispanic/Latinx (n=31, 10%), and many identified with other Black racial identities or as multiracial (n=117, 37%). The majority of participants were cisgender men (n=277, 88%), although some identified as transgender women (n=27, 9%) or as gender diverse (n=12, 4%). The median age of participants in this sample was 21 years (interquartile range [IQR]= 19–23) and 19% (n=60) were 18 years of age or younger. Nearly one in four participants reported recent housing instability (n=69, 22%) or engagement in transactional sex (n=59, 19%). Many participants reported having a sexual partner who was using PrEP (n=131, 42%). Most participants reported at least one HIV risk factor (n=299, 95%). A summary of participant characteristics can be found in Table I.

Table 1.

Summary of participant characteristics

	Overall (n=316)
Age, years (median [IQR])	21.00 [19.00, 23.00]
Racial/Ethnic Identity
African American (n,%)	168 (53.2)
Black African (n,%)	58 (18.4)
Hispanic (n,%)	31 (9.8)
Black Latino (n,%)	19 (6.0)
Black White (n,%)	15 (4.7)
Black Caribbean (n,%)	11 (3.5)
Black Arabic (n,%)	8 (2.5)
Other (n,%)	5 (1.6)
Black Asian (n,%)	1 (0.3)
Gender Identity
Cisgender men (n,%)	277 (87.7)
Transgender women (n,%)	27 (8.5)
Gender diverse (n,%)	12 (3.8)
Metropolitan Area
Philadelphia (n,%)	101 (32.0)
Baltimore (n,%)	100 (31.6)
Washington D.C (n,%)	115 (36.4)
Perceived Family Wealth (%)
Not at all well off	19 (6.0)
Not so well off	59 (18.7)
Average	154 (48.7)
Quite well off	58 (18.4)
Very well off	26 (8.2)
Housing Instability (n, %)	69 (21.8)
Transactional Sex (n, %)	59 (18.7)
Insured (n, %)	270 (85.4)
PrEP use in Sexual Network (n, %)	131 (41.5)
Size of Sexual Network (median [IQR])	2.00 [1.00, 4.00]
HIV Risk Score (median [IQR])	3.00 [2.00, 4.00]
Employed (n, %)	182 (57.6)
HIV negative by self-report only (n, %)	68 (21.5)
PrEP Continuum Stage*
PrEP Unaware (n, %)	62 (19.6)
PrEP Aware (n, %)	47 (14.9)
PrEP Willing (n, %)	37 (11.7)
PrEP Intending (n, %)	113 (35.8)
Current PrEP User (n, %)	57 (18.0)

^*participants classified into mutually exclusive groups based on the last step in the PrEP continuum that they have reached.

PrEP continuum

Figures 1 and 2 show the progress of participants along the PrEP continuum, both in the overall sample, and stratified by racial/ethnic identity and gender identity. Overall, 80% of participants were aware of PrEP (n=254), 66% were willing to use PrEP (n=207), 54% intended to use PrEP in the near future (n=170), and 18% were currently using PrEP (n=57). Reported adherence was high among the 57 participants currently taking PrEP: only four reported missing three or more doses in the last week. Latinx participants were more likely to be willing to use PrEP when compared to participants in the “Other Black or multiracial” racial/ethnic category (odds ratio [OR]= 3.85, confidence interval [CI]=1.33–13.75), and were more likely to be currently using PrEP when compared to Black participants (OR= 4.51, CI=1.78–11.31) and when compared to participants in the “Other Black or multiracial” racial/ethnic category (OR= 3.27, CI=1.27–8.39). There were no other significant differences in progress along the PrEP continuum by racial/ethnic identity. We found no differences at any level of the continuum by gender identity.

Figure 1:

Cascade plot of progression along PrEP continuum for overall sample and racial/ethnic subgroups. Numbers on bars denote percentage of sample that is at or beyond each level of the continuum. A) Overall sample (N=316), B) African American/Black participants (n=168), C) Latinx participants (n=31), and D) participants of other Black racial/ethnic identities or who identified as multiracial (n=117).

Figure 2:

Cascade plot of progression along PrEP continuum for overall sample and gender identity subgroups. Numbers on bars denote percentage of sample that is at or beyond each level of the continuum. A) overall sample (N=316), B) cisgender men (n=277), C) transgender women (n=27), and D) gender diverse participants (n=12).

Ordinal regression

Being insured (adjusted odds ratio [aOR]=2.95, 95% CI=1.60–5.49) and having had a sexual partner who uses PrEP (aOR= 4.19, 95% CI=2.61–6.79) were each associated with greater engagement across the PrEP continuum with proportional odds at each level of the continuum. Higher HIV risk score was also associated with greater engagement across the PrEP continuum with proportional odds at each level of the continuum (aOR= 1.62, 95% CI= 1.34–1.97). Participants who were 15–18 years of age had greater odds of PrEP willingness (aOR= 2.30, 95% CI= 1.10–4.83) and lower odds of current PrEP use (aOR= 0.31, 95% CI= 0.11–0.90) compared to those aged 22–24. Participants who reported housing instability had lower odds of PrEP awareness (aOR= 0.39, 95% CI= 0.19–0.81). No significant differences in engagement with the PrEP continuum were observed by racial/ethnic identity, gender identity, perceived family wealth, engagement in transaction sex, or sexual network size (Table II).

Table 2.

Partial proportional odds ordinal logistic regression for engagement with pre-exposure prophylaxis (PrEP) continuum (n=316)

	Adjusted odds ratio, proportional odds (aOR, 95% CI)a	Adjusted odds ratio, PrEP awareness (aOR, 95% CI)b	Adjusted odds ratio, PrEP willingness (aOR, 95% CI)b	Adjusted odds ratio, PrEP intention (aOR, 95% CI)b	Adjusted odds ratio, current PrEP use (aOR, 95% CI)b
Age, >=22	Ref	Ref	Ref	Ref	Ref
Age, 19–21	---	1.29 (0.65, 2.56)	1.27 (0.70, 2.32)	0.88 (0.49, 1.57)	0.51 (0.24, 1.06)
Age, <19	---	1.14 (0.48, 2.71)	2.30 (1.10, 4.83)*	1.66 (0.83, 3.32)	0.31 (0.11, 0.90)*
African American/Black	Ref	Ref	Ref	Ref	Ref
Latinx	1.26 (0.59, 2.77)	---	---	---	---
Other/multiracial	0.69 (0.44, 1.09)	---	---	---	---
Cisgender men	Ref	Ref	Ref	Ref	Ref
Transgender women	1.49 (0.72, 3.15)	---	---	---	---
Gender diverse	0.63 (0.22, 1.79)	---	---	---	---
Perceived family wealth >= average	1.20 (0.73, 1.97)	---	---	---	---
Housing instability	---	0.39 (0.19, 0.81)*	0.80 (0.41, 1.55)	1.57 (0.83, 2.99)	1.95 (0.91, 4.16)
Transactional sex	1.32 (0.75, 2.35)	---	---	---	---
Insured	2.95 (1.60, 5.49)***	---	---	---	---
PrEP use in sexual network	4.19 (2.61, 6.79)***	---	---	---	---
Size of sexual network	1.00 (0.97, 1.03)	---	---	---	---
HIV risk score	1.62 (1.34, 1.97)***	---	---	---	---

^aAdjusted odds ratios for variables that satisfied the proportional odds assumption

^bAdjusted odds ratios at each cut-point for variables that did not satisfy the proportional odds assumption

^*p < 0.05,

^***p < 0.001.

In a sub-analysis of participants identifying as transgender or as gender diverse (n=39), engaging in transactional sex (aOR= 6.73, 95% CI= 1.36–39.58) and having had a sexual partner who uses PrEP (aOR= 5.05, 95% CI= 1.17–24.42) was each associated with greater engagement across the PrEP continuum with proportional odds at each level of the continuum. Participants in the “Other Black or multiracial” racial/ethnic category had lower engagement across the PrEP continuum when compared to Black participants (aOR= 0.11, 95% CI= 0.02–0.53). No significant differences in engagement with the PrEP continuum for this sub-sample were observed by age, perceived family wealth, housing instability, insurance status, sexual network size, or HIV risk score (Table III).

Table 3.

Proportional odds ordinal logistic regression for engagement with pre-exposure prophylaxis (PrEP) continuum among transgender and gender diverse participants (n=39)

	Adjusted odds ratio, proportional odds (aOR, 95% CI)
Age, >=22	Ref
Age, 19–21	0.35 (0.06, 1.79)
Age, <19	1.76 (0.22, 14.34)
African American/Black	Ref
Latinx	6.84 (0.33, 174.49)
Other/multiracial	0.11 (0.02, 0.53)**
Perceived family wealth >= average	1.82 (0.41, 8.52)
Housing instability	0.92 (0.17, 4.93)
Transactional sex	6.73 (1.36, 39.58)*
Insured	4.78 (0.46, 57.43)
PrEP use in sexual network	5.05 (1.17, 24.42)*
Size of sexual network	0.94 (0.82, 1.04)
HIV risk score	1.53 (0.82, 2.94)

^*p < 0.05,

^**p < 0.01

Small sample size in this sub-analysis produced wide confidence intervals for some variables.

DISCUSSION

Our study sought to examine young Black and Latinx MSM, transgender women, and gender diverse individuals’ engagement along the PrEP continuum contextualized by levels of HIV risk, sexual network characteristics, and economic vulnerability. Our sample exhibited moderate PrEP awareness, high levels of willingness and intention to use PrEP, low levels of current PrEP use, and a substantial number of participants who had discontinued PrEP use. Economic and network-related factors were found to be associated with engagement across the PrEP continuum. Our findings reflect the ongoing challenges that HIV prevention experts have in adequately reaching young people from marginalized communities, populations that may benefit the most from PrEP use.

Insurance coverage was positively associated with engagement across all levels of the PrEP continuum. These findings suggest that for the young people in this sample, lack of financial access to PrEP may dampen motivational factors to get linked to PrEP-related medical care and pose logistical barriers to care. Inadequate health insurance and other cost-related barriers can limit access to PrEP among those who are economically vulnerable. For those without health insurance, the anticipated cost of medical appointments and laboratory fees can be prohibitive, regardless of whether assistance programs exist to cover these costs.(15) Even for those who are insured, costs related to using PrEP, especially the ancillary costs of provider visits and laboratory fees, can still present barriers to access due to co-pays and deductibles.(29–31) There is also evidence of emerging racial disparities in PrEP discontinuation among young MSM,(16) and structural barriers such as loss of health insurance or inadequate resources for navigating complex health systems may be at the root of these disparities.(32) Programs and policies should be developed that address these barriers to PrEP access, adoption, and maintenance, including expanded access to Medicaid insurance as well as efforts to increase awareness and acceptability of PrEP access programs for those without insurance which cover medication and ancillary care costs.(3) These efforts will become even more critical as next generation HIV prevention modalities become available (e.g., long-acting injectable PrEP). Equitable implementation of these new technologies will require a focus on bolstering awareness, ensuring widespread accessibility, and providing the resources that young people need to navigate complex medical systems.

The role of social and sexual networks in PrEP engagement is emerging as a priority area of HIV prevention research.(33) We found that sexual partnerships with current PrEP users were positively associated with participants’ engagement with PrEP across all levels of the continuum. Sexual network size was not statistically linked to participants’ PrEP engagement. Our findings are consistent with the theory of diffusion of innovation, which postulates that behavior change can propagate through networks via social ties. This theory suggests that widespread behavior change is sparked by early adopters who model a behavioral innovation and can influence the behavior of those connected to them in a network.(34) Previous research has found that having PrEP users in one’s social network is associated with greater PrEP engagement,(35) and given the potential overlap between social and sexual networks, our results partially support this idea. The association between PrEP use in sexual networks and personal engagement with PrEP found in our study sample highlights the power of young people to promote HIV prevention among their peers. However, we saw that large networks alone were not sufficient to promote PrEP engagement. Instead, the presence of a PrEP user in one’s sexual network appears to be critical in promoting engagement. A small number of HIV prevention programs focused on young Black and Latinx MSM have used popular opinion leader and peer-change agent methods to promote PrEP engagement with these high priority populations.(36,37) PrEP interventions should leverage the connections that exist within social and sexual networks to promote PrEP engagement among underserved groups in order to address disparities in PrEP uptake.

Conceptualizing PrEP engagement as a continuum highlights how different factors can have differential impacts on various stages of this continuum. We found that younger participants (ages 15–18) had greater odds of having reached the PrEP Willing stage, and lower odds of having reached the Current PrEP User stage, when compared to older participants (ages 22–24). This supports the findings of other researchers, who have noted that adolescents report high levels of PrEP willingness(38) but low rates of PrEP uptake.(39–41) Adolescents may be less aware of costs, follow-up appointment requirements, and logistical challenges associated with accessing PrEP when compared to older young adults; years of life experience may yield greater likelihood of achieving PrEP access, but also a more realistic assessment of the challenges associated with achieving access.(38) We also found that housing instability was associated with decreased odds of PrEP awareness, which corroborates previous research.(42) A recent study of homeless youth found that despite most participants having an indication for PrEP based on behavioral risk, less than one third of were aware of PrEP as a method of HIV prevention.(43) Efforts to promote PrEP engagement will need to attend to barriers to engagement across the continuum. Interventions are needed that connect economically vulnerable young people to information and resources related to HIV prevention, especially among those experiencing homelessness. Additionally, strategies are needed to address the structural drivers of economic vulnerability among marginalized young people, including policies and programs that help to prevent homelessness and rapidly connect those experiencing homelessness to stable housing with supportive medical services.

We found that among transgender and gender diverse participants, transactional sex was associated with greater engagement with the PrEP continuum, suggesting that the role of transactional sex in shaping HIV vulnerability may dominate its role as a marker of economic vulnerability for young gender minorities. Young people who engage in transactional sex must balance various competing priorities, including economic needs to pay for food, housing and medical expenses, with the need to protect their own sexual health. The potential health risks involved in transactional sex is one factor that weighs on this balance, and young people respond to this risk by adopting HIV prevention behaviors when they are feasible.(44) Policies which create economic opportunities for transgender and gender diverse youth paired with policies that protect rather than persecute people who engage in transactional sex would support youth in making decisions around sexual health.

This study was subject to several notable limitations. Our sample included a small number of transgender and gender diverse participants, allowing for coarse descriptions and analysis of the PrEP continuum for these underrepresented gender identities. Given that data on PrEP engagement for these populations is sparse, subsequent research should aim to enroll larger samples of participants with diverse gender identities. The number of Latinx participants was also small and our study was limited to only English speakers. Future research should aim to recruit more Latinx participants, especially those whose primary language is not English. Social networks, in addition to sexual networks, are likely to play an important role in shaping how attitudes and behaviors related to PrEP are communicated amongst peers. Researchers should measure characteristics of both social and sexual networks in future studies to gain a more complete understanding of how networks influence PrEP engagement. Finally, due to the cross-sectional design of this study, we were unable to assess the causal direction of the associations we found. Longitudinal research can help to better understand the causal pathways connecting individual, network, and structural factors with PrEP engagement.

CONCLUSION

The advent of PrEP has the opportunity to dramatically change the landscape of HIV prevention in the U.S. Despite the great promise of this highly effective biomedical prevention modality, those at greatest risk for HIV have yet to see the full benefit of this innovation. PrEP implementation efforts have been slow to reach marginalized groups, specifically young Black and Latinx MSM and transgender women, who are disproportionately burdened by HIV. Understanding the unique, multilevel barriers and facilitators to PrEP engagement that exist for young people facing structural marginalization will be paramount to effectively addressing inequities associated with the HIV epidemic. Our study highlighted important factors for intervention and implementation scientists to consider, namely social ties to PrEP users and inadequate health insurance coverage, in continued PrEP rollout efforts for marginalized young people.

Data availability

The data that support the findings of this study are available on reasonable request from the corresponding author, SB. The data are not publicly available due to their containing information that could compromise the privacy of research participants.