Introduction

Transgender and gender expansive individuals are those whose gender identity differs from, or expands beyond, the sex assigned to them at birth [1]. Transgender (hereafter: trans) adults are 32 times more likely to acquire HIV than their cisgender peers [2,3,4,5]. Although trans and gender expansive youth and young adults (TGEYYA) are disproportionately affected by HIV and other health disparities, they are understudied and underserved [6,7,8,9,10,11,12]. Furthermore, most HIV prevention research with TGEYYA has focused on trans women, with few studies investigating the unique life circumstances and needs among subgroups of trans adolescents [13, 14]. A recent scoping review on determinants of pre-exposure prophylaxis (PrEP) implementation for HIV prevention with trans populations [15] identified only four peer-reviewed manuscripts on PrEP studies that explicitly included trans men–of which two had samples sizes of trans men too small to evaluate their barriers and facilitators to PrEP uptake [16, 17]–and only one PrEP study that explicitly included nonbinary participants [18].

There are many challenges in evaluating HIV prevention research among TGEYYA, such as inconsistent language regarding the population, flawed or nonexistent population-based estimates, and inattention to or exclusion of trans men and nonbinary individuals who are often mistakenly classified as low risk for HIV acquisition [9, 19,20,21,22]. Estimates published in 2022 suggest that there were more than 1.6 million trans individuals over age 13 in the United States [21]. Furthermore, young adults ages 18–24 years old are more likely to identify as trans or nonbinary than other age groups [23, 24].

TGEYYA endure multiple intersecting forms of marginalization that contribute to mutually reinforcing structural and behavioral syndemics, which cumulatively heighten susceptibility to HIV infection [25]. Violence and social stigma against TGEYYA often start early and extend across the life course [26]. Frequent family rejection can lead to cycles of housing instability, food insecurity, and poverty, amplified by institutional discrimination in education and employment, which can lead to a reliance on informal work, such as exchange sex to earn money [27,28,29,30]. It is estimated that 1 in 2 young trans women and 1 in 3 young trans men report exchanging sex for money or other needs, such as a place to stay [31,32,33,34]. TGEYYA are frequently profiled by law enforcement for their gender expression and incarcerated at higher rates than cisgender peers, exacerbating mental health and substance use disparities [35,36,37]. Approximately 1 in 5 trans people in the U.S. have spent time in prison [38]. Substance use is 2.5 to 4 times higher and polysubstance use is 4 times higher among TGEYYA compared to cisgender adolescents [39]. Few studies have evaluated differences in structural and behavioral risks among TGEYYA to determine which groups are most susceptible to HIV and in need of preventive interventions, such as PrEP [10, 15, 40].

TGEYYA may be reticent to seek out HIV prevention services because of a history of discrimination, stigmatization, and pathologizing by health professionals. Indeed, many TGEYYA do not share their gender identity or sexual behaviors with healthcare providers out of fear of discrimination or being outed to others [10, 41,42,43]. Among trans adults, lower rates of healthcare utilization have been documented due to anticipated discrimination, and while not assessed among youth, TGEYYA must also consider privacy concerns if they receive health insurance through a parent [30, 44, 45]. These barriers to healthcare might limit TGEYYA’s perceived need for PrEP as a tool for HIV prevention, and impair their ability to move through the PrEP cascade and HIV prevention continuum from no engagement, to HIV testing, to linkage to care, and finally to retention in routine healthcare and HIV prevention services [40, 46,47,48,49,50,51].

HIV prevention research with TGEYYA has almost exclusively focused on trans women, and less is known about risk and protective behaviors among trans men, trans masculine individuals, and nonbinary adolescents, irrespective of sex assigned at birth [15]. Over a decade of research has reliably estimated the prevalence of HIV among primarily adult trans women in the U.S. to be at least 1 in 5 [4, 52]. Although national estimates among trans masculine and nonbinary populations are not available, a few small non-probability samples of primarily adult trans men have demonstrated rates of HIV from 2 to 10%, much higher than the national HIV prevalence of 0.4% in the U.S. [5, 12, 53, 54].

Trans masculine and nonbinary individuals assigned female at birth who have sex with men are at increased risk for HIV transmission if they have condomless receptive anal or vaginal sex while not on pre-exposure prophylaxis (PrEP). Furthermore, transmission risk is elevated among those taking testosterone, which can cause vaginal atrophy and decreased lubrication [55,56,57,58]. Nonetheless, trans masculine individuals have been excluded from all PrEP trials despite high rates of eligibility for PrEP, and gender expansive and nonbinary populations assigned female at birth have been largely invisible in HIV research [13, 59,60,61]. Studies that informed CDC clinical guidelines on PrEP included very small numbers of trans women and no trans men; as such, efficacy data relevant for trans masculine and nonbinary populations are not available, and PrEP remains underutilized among all subgroups of TGEYYA [14, 47, 62, 63].

Additionally, previous studies have shown that while TGEYYA have significant needs for PrEP, uptake is differentially associated with both sex assigned at birth and gender identity, as well as access to healthcare [64]. Trans women tend to have high perceived need for PrEP aligned with their self-reported sexual behaviors, higher rates of HIV testing and positivity for HIV and STIs than other TGEYYA subgroups [65], and also higher PrEP awareness and uptake than other TGEYYA subgroups [66], yet they still face barriers such as stigma, medical mistrust, affordability, and accessibility [41, 43, 67]. Trans men report less exposure to targeted PrEP outreach, which affects their uptake and perceived need for PrEP. Nonbinary individuals tend to face greater barriers to care generally than trans women and trans men related to less visibility and provider knowledge of their identities and sexual behaviors, and lack of inclusive interventions to meet their needs [68,69,70,71].

The current analysis aimed to address gaps in the literature on PrEP use among TGEYYA, attending to differences by gender identity and sex assigned at birth. The objectives of the current analysis were to assess disparities in current and lifetime PrEP uptake among TGEYYA and potential mechanisms that might explain these disparities. In our investigation of mechanisms, we considered mediation by structural risk factors (housing instability, incarceration, low income, unemployment, and uninsurance), behavioral risk factors (condomless sex, multiple sex partners, exchange sex, intimate partner violence, and polysubstance use), perceived need for PrEP, and healthcare utilization based on the participant’s self-reported history of HIV testing, linkage to care, and retention in routine health care. We hypothesized that increased structural and behavioral risk factors would be associated with greater perceived need for PrEP, more healthcare utilization, and ultimately PrEP uptake.

Methods

The 2017–2022 cycle of the National Adolescent Medicine Trials Network for HIV/AIDS Interventions (ATN) consisted of three research networks (or U19s), and each U19 consisted of several study protocols. This cross-network analysis used baseline data from a pooled sample of TGEYYA participants (N = 477) in the CARES and TechStep study protocols within the ATN.

Pooled Study Sample and Inclusion Criteria

Participants who met the following criteria were included in the pooled sample for this cross-network analysis: 1) 12–24 years old at study enrollment; 2) seronegative for HIV infection at study enrollment; 3) gender identity different from sex assigned at birth; and 4) enrolled in either CARES (n = 197) or TechStep (n = 284) parent studies. Four participants were excluded due to missing information for sex assigned at birth. The final pooled sample included 477 TGEYYA participants. An overview of the CARES and TechStep parent study designs, eligibility criteria, participants, and procedures are provided below.

Parent Study Data and Samples

CARES Study Design

Participants for this cross-network analysis were drawn from CARES ATN Protocol 149, which enrolled seronegative youth into a four-condition randomized controlled trial for technology-based interventions to promote completion of each step of the HIV prevention continuum among youth at high risk for HIV [72].

CARES Participants

From May 2017 to August 2019, 1,482 seronegative participants between the ages of 12–24 years old were enrolled in ATN 149. Inclusion criteria were: (1) a negative rapid HIV result on a 4th generation test at screening (Alere, Waltham, Massachusetts) AND (2) self-reporting at least three HIV-related structural or behavioral risks among the following: past 12-month history of condomless anal sex, an HIV-positive sex partner, sharing needles for injecting drugs, illicit substance use (excluding marijuana), or a positive STI test; having ever been homeless, hospitalized for a mental health disorder, or been to jail or on probation in their lifetime; OR (3) identifying as trans or nonbinary/gender expansive – irrespective of sex assigned at birth – or as gay, bisexual, or other men who have sex with men (GBMSM); (4) able to read and speak English; and 5) age 12–24 years. Adolescents who self-identified as sexual or gender minority were weighted more heavily in eligibility scoring due to their higher representation in the domestic HIV epidemic. Baseline data from seronegative CARES participants who selected a gender identity different from their sex assigned at birth were included in this cross-network analysis (n = 197). These TGEYYA constituted 13% of the total 1,482 participants in ATN 149.

CARES Procedures

CARES was conducted in Los Angeles and New Orleans. Youth were recruited via: (1) community-based organizations (CBOs) and clinics that served gay, bisexual, and trans youth; homeless youth; and youth on probation or released from incarceration; (2) dating apps (e.g., Jack’d and Scruff); (3) community-based events (e.g., PRIDE, Teen PRIDE); and (4) referrals from other enrolled study participants [73]. Following consent, participants completed a tracking form and baseline assessment verbally administered by interviewers, and rapid diagnostic tests for HIV, STIs (syphilis, chlamydia, and gonorrhea), and substance use (alcohol, marijuana, opiates, methamphetamine, cocaine, and benzodiazepines). The study was approved by the Institutional Review Board at the University of California, Los Angeles, which served as the IRB of Record for all participating institutions.

TechStep Study Design

TechStep was Protocol ATN 160 within the University of North Carolina/Emory Center for Innovative Technology (iTech) within the Adolescent Medicine Trials Network for HIV/AIDS Interventions (ATN) [74]. The TechStep design included a three-condition, technology-based, randomized controlled trial, with a stepped care approach, for reducing sexual risk behaviors and increasing PrEP initiation among HIV-negative trans youth and young adults.

TechStep Participants

From October 2019 to September 2021, 284 participants completed the TechStep baseline Audio Computer Assisted Self Interview (ACASI) assessment. Inclusion criteria were: (1) self-identify as trans feminine, trans masculine, gender expansive, or any term that reflected incongruity between sex assigned at birth and current gender identity; (2) aged 15–24 years old; (3) self-report vaginal or anal sex (either insertive or receptive) with another person in the last 12 months; (4) verified HIV negative serostatus; (5) a mobile device with SMS and Internet access capabilities; and (6) able to read and speak English (as the interventions were built in English). Individuals were excluded if they did not meet all eligibility criteria. All 284 participants who completed the baseline assessment were included in this cross-network analysis because the eligibility criteria for TechStep required that all participants identified as trans or nonbinary/gender expansive.

TechStep Procedures

The TechStep study was conducted at five Subject Recruitment Venues (SRV) located in New York City, Philadelphia, Boston, Houston, and Los Angeles. Participants were recruited via (1) online ads; (2) flyers; (3) street- and venue-based outreach; (4) direct SRV referral or online screener; and (5) participant/peer referral. Following consent, participants completed the baseline ACASI assessment. Immediately following baseline procedures, participants were randomized 1:1:1 into one of three technology-based conditions: (1) WebApp (i.e., a website intervention optimized for a smartphone); (2) text messaging; or (3) an information only website. At month 3, participants who did not show improvement on the primary outcomes were re-randomized to potentially “step-up” to receive virtual e-coaching sessions. The study was approved by the Institutional Review Boards at the University of North Carolina, Chapel Hill, NC, which served as the IRB of Record for all participating institutions and SRVs.

Measures

Gender Identity and Sex Assigned at Birth

The independent variable for this analysis was gender identity, which was operationalized as a four-category variable: 1 = Trans woman/feminine, 2 = Nonbinary assigned male at birth (AMAB), 3 = Nonbinary assigned female at birth (AFAB), 4 = Trans man/masculine. Response options for gender identity differed between CARES (select one identity term) and TechStep (select all that apply). To ensure that categories were consistent across both studies, we created gender categorization rules for grouping individuals who selected more than one gender identity (See Table 1), accounting for both the participant’s current gender identity and sex assigned at birth. 72% of participants in the pooled sample selected one gender identity and 28% selected two or more identities.

Table 1 Gender categorization for ATN cross-network analysis of CARES and TechStep, 2017–2021, (N = 477)
Full size table

PrEP Uptake

The outcome for this analysis was PrEP uptake, which was operationalized as a dichotomous variable indicating if the participant had ever taken PrEP in their lifetime (Yes = 1, No = 0). Participants were also asked about current PrEP use (Yes = 1, No = 0), but this variable could not be used as an outcome for the analysis due to lack of variation, as only 7% of respondents reported currently taking PrEP at the time of baseline interview. Current PrEP use was included in the descriptive sample characteristics.

Mediators

Four mediators were tested in this analysis (structural risks, behavioral risks, no perceived need for PrEP, and healthcare utilization). These mediators were arranged serially along two pathways between gender identity and PrEP uptake to determine if PrEP disparities were explained by differences in both individual risk behaviors and perceptions, as well as by contextual environments and healthcare utilization. These two mediation pathways were designed based on Link and Phelan’s (1995) Theory of Social Conditions as Fundamental Causes [75], such that upstream structural risks were hypothesized to influence both the individual, behavioral/perception pathway and the contextual/healthcare utilization pathway. On the first pathway, we tested whether structural risks lead to behavioral risks, which in turn lead to perceived need for PrEP, and ultimately to PrEP uptake. This individual, behavioral/ perception pathway was informed by value expectancy theory from the Health Belief Model [76, 77]. On the second pathway, we tested whether structural risks lead to healthcare utilization and ultimately to PrEP uptake. This structural pathway was informed by Andersen's Model for Health Services Utilization [78]. See Figs. 1–3 for all mediation pathways tested in this analysis. The operationalization of each mediator is described below.

Fig. 1
figure 1

Generalized structural equation model for PrEP uptake among TGEYYA, model 1: unstratified, ATN cross-network analysis of CARES and TechStep, 2017–2021 (N = 477)

Full size image

Behavioral risk index was a count variable ranging from 0 to 5 for the number of behavioral risks reported by the participant in the last 3 months among the following dichotomous variables (1 = Yes, 0 = No): 1) condomless anal or vaginal sex, 2) multiple sex partners, 3) exchange sex, 4) intimate partner violence, and 5) polysubstance use.

No perceived need for PrEP was a dichotomous risk perception variable drawn from a list of reasons for not taking PrEP or stopping PrEP use. Participants were asked if they agreed with the following statement: “I didn’t think I needed PrEP” (1 = Yes, 0 = No).

Structural risk index was a count variable ranging from 0 to 5 for the number of structural vulnerabilities ever experienced by the participant among the following dichotomous variables (1 = Yes, 0 = No): 1) housing instability, 2) incarceration, 3) low income (< $500/month), 4) unemployment (not counting students), and 5) uninsurance. Type of insurance and access to insurance through parents could not be evaluated because insurance questions were not asked consistently across the two studies.

Healthcare utilization was a four-category, ordered variable ranging from 0 to 3, constructed from multiple questions related to the participant’s HIV testing history, linkage to health care, and retention in care. Tested was operationalized as ever having been tested for HIV in lifetime (1 = Yes, 0 = No). Linked was operationalized as currently having a primary care provider or regular source of care (1 = Yes, 0 = No). Retained was operationalized as having seen a doctor within the last 6 months (1 = Yes, 0 = No). When combining these variables, the four levels were: 0 = Never tested for HIV; 1 = Tested for HIV, but not linked to care; 2 = Tested for HIV, linked to care, but not retained; 3 = Tested, linked, and retained. The distribution across these four categories was reported in the descriptive sample characteristics. For the mediation analysis, this variable was dichotomized to indicate if the participant was tested, linked, and retained (category 3 vs. categories 0–2) because PrEP users must be confirmed seronegative (tested), have a PrEP prescriber (linked), and be regularly engaged in care for ongoing testing and monitoring (retained). Half of participants in the sample were categorized into this final level (tested, linked, and retained in care).

Control Variables

Several variables were included to control for confounding of gender identity with PrEP uptake and mediating variables. Demographic characteristics included sexual identity (Heterosexual (i.e., straight), Homosexual (i.e., gay or lesbian), Bisexual, or Pansexual/Other), race/ethnicity (African American, Latinx, White, or Other), age in years ranging from 12 to 24, city (Boston, Houston, Los Angeles, New Orleans, New York, or Philadelphia) and region (Northeast, West, or South), and education (Any College vs. None). Study enrollment was also included (CARES or TechStep). Finally, a PrEP Obstacles Index was constructed as a count variable ranging from 0 to 5 for the sum of the following dichotomous variables (1 = Yes, 0 = No): 1) never heard of PrEP, 2) concerns regarding PrEP side effects, 3) difficulty with schedule for taking PrEP, 4) difficulty getting PrEP medication, and 5) no one to support PrEP use. Due to lack of variation for each item in the PrEP Obstacles Index, these variables could not be included as individual covariates in the multivariate analyses. We also could not assess parental insurance nor parental support for PrEP because these variables were not consistent in both studies.

Statistical Analysis

Data were combined and analyzed using R statistical software (version 4.2.2; R Foundation for Statistical Computing, Vienna, Austria) [79].

Descriptive Statistics

We computed univariate statistics on the total pooled sample (N = 477). We calculated frequencies and percentages for categorical variables and means and standard deviations for continuous variables.

Bivariate Analyses

We conducted separate analyses to test for relationships between each covariate and the outcome variable (PrEP uptake), the proximal mediators (no perceived need for PrEP and healthcare utilization), and the distal mediators (structural and behavioral risk indices). For each categorical control variable, we performed a Pearson's chi-squared test (or Fisher’s exact test for small cell sizes < 5) to determine significant differences in PrEP uptake, no perceived need for PrEP, healthcare utilization, and the structural and behavioral risk indices. For continuous control variables, we performed two-tailed t-tests. The significance level was set at p < 0.05.

Structural Equation Models

After determining which variables were significantly associated with PrEP uptake and the mediators in bivariate comparisons, we built a generalized structural equation model (GSEM) using the lavaan package in R (version 0.6–10) to allow for discrete measures and test all mediation pathways simultaneously. We selected seven variables for the full model, each of which was associated with one or more outcomes/mediators below the significance level of p < 0.05. These variables were: gender identity, region, behavioral risk index, structural risk index, no perceived need for PrEP, healthcare utilization, and PrEP uptake. To ensure convergence, we limited the number of model parameters by dropping control variables which were not associated with PrEP uptake when adjusting for other factors (race/ethnicity, age, and education) and dichotomizing gender identity by sex assigned at birth (1 = Female, 0 = Male) and region (1 = South, 0 = West or Northeast).

We tested three GSEMs: Model 1 with the full pooled sample (N = 477), Model 2 with only participants assigned male at birth (n = 191), and Model 3 with only participants assigned female at birth (n = 286). In the stratified GSEMs, (Models 2 and 3), the independent variable was dichotomized to compare by gender identity among those with the same sex assigned at birth (1 = Nonbinary, 0 = Trans Woman or Trans Man). We reported the exponentiated estimates as adjusted odds ratios (AOR), 95% confidence intervals (CI), and p-values along each path. We calculated the likelihood ratio chi-squared test, goodness of fit index (GFI), comparative fit index (CFI), and root mean squared error of approximation (RMSEA) for each GSEM. Model fit was determined by a chi-squared p-value > 0.05, GFI > 0.95, CFI > 0.90, and RMSEA < 0.05.

As sensitivity analyses, we conducted five separate regressions of PrEP uptake and the four mediators on control variables to determine if coefficient values and statistical significance levels matched equivalent pathways in the GSEM. Because the GSEM and regression results were consistent, only the GSEM results are reported to limit redundancy.

Results

Sample Characteristics

See Table 2 for the descriptive characteristics of the sample in total and stratified by PrEP uptake, no perceived need for PrEP, and healthcare utilization.

Table 2 Sample characteristics of TGEYYA in total and by PrEP uptake, no perceived need for PrEP, and healthcare utilization, ATN cross-network analysis of CARES and TechStep, 2017–2021, (N = 477)
Full size table

Demographics

60% of the sample were assigned female at birth and 40% were assigned male at birth. Within each group, half identified as trans (trans woman or trans man), and half identified as nonbinary. Most participants (65%) identified as pansexual or bisexual, 26% identified as homosexual (i.e., gay or lesbian), and 9% identified as heterosexual (i.e., straight). For race/ethnicity, 44% were White, 20% African American/Black, 20% Latinx, and 17% Other. The mean age was 21 years old (SD: 2.0, range: 15–24), and the majority of participants were over 18 years of age (91%). Most had at least some college education (74%). By region, 37% of participants were recruited in the Northeast from Boston, New York City, or Philadelphia; 23% in the South from Houston or New Orleans; and 41% in the West from Los Angeles. Slightly more than half of participants (59%) were enrolled in the TechStep study and 41% in CARES.

PrEP Uptake

Only 16% percent of participants reported ever taking PrEP in their lifetime, and 7% reported that they were currently taking PrEP at study enrollment. In bivariate comparisons, the proportion who reported PrEP uptake in their lifetime was significantly higher among: trans women (29%) and nonbinary AMAB participants (23%) compared to trans men (12%) and nonbinary AFAB participants (6%); heterosexual and homosexual participants compared to bisexual and pansexual participants; Latinx participants; those who reported a perceived need for PrEP; those who had been tested, linked, and retained in health care; those who reported two or more behavioral risks, and in particular, multiple sex partners; those who did not report concerns regarding PrEP side effects or difficulty getting the medication; and those who had heard of PrEP.

No Perceived Need for PrEP

Almost half of participants (43%) reported no perceived need for PrEP. The proportion of participants reporting no perceived need for PrEP was significantly higher among: nonbinary AFAB participants and trans men vs. nonbinary AMAB participants and trans women; bisexual, pansexual, and homosexual participants compared to heterosexual participants; White participants and those of other races/ethnicities compared to African American/Black and Latinx participants; those with any college education; those recruited in the Northeast; those enrolled in TechStep vs. CARES; those who had never taken PrEP; those who had never been tested for HIV; those who reported fewer behavioral risks and fewer structural risks; and those who reported two or more PrEP obstacles, especially concerns regarding PrEP side effects.

Healthcare Utilization

Half of participants (52%) reported that they were tested, linked and retained in healthcare. Among those who were not retained in care, 20% had never been tested for HIV, 18% had been tested but did not have a regular source of medical care, and 10% had been tested and linked to medical care but had not seen a doctor in the last 6 months. The proportion of participants who were tested, linked, and retained in care was significantly higher among: trans women and trans men vs. nonbinary participants, irrespective of sex assigned at birth; Latinx participants; older participants; those recruited in the Northeast or West compared to the South; those enrolled in CARES vs. TechStep; those who had ever taken PrEP; those who reported perceived need for PrEP; those who reported two or more behavioral risks, especially exchange sex, multiple sex partners, and polysubstance use; those who had insurance; and those who had heard of PrEP.

Behavioral Risk Index

Participants reported an average of 1.6 behavioral risks (SD: 1.1) in the last 3 months, and 44% reported two or more behavioral risks. The majority of participants who reported two or more behavioral risks were assigned male at birth. The most commonly reported behavioral risks were condomless anal or vaginal sex (57%), having multiple sex partners (45%), and polysubstance use (35%). Exchange sex and violence were each reported by fewer than 12% of participants, but more commonly reported by trans women (23% and 16%, respectively). 67% percent of nonbinary participants assigned male at birth, 58% of trans women, 56% of trans men, and 49% of nonbinary participants assigned female at birth reported condomless anal or vaginal sex. There were no significant differences in those reporting multiple sex partners across all gender identities. Trans women reported significantly more behavioral risks on average compared to nonbinary participants assigned male at birth, trans men, and nonbinary participants assigned female at birth.

Structural Risk Index

Participants reported an average of 1.3 structural risks (SD: 1.3) in their lifetime, and 32% reported experiencing two or more structural risks. Most participants who reported two or more structural risks were assigned male at birth. The most commonly reported structural risks were being low income (44%), unemployed or disabled (30%), and housing instability (19%). Incarceration and being uninsured were each reported by fewer than 13% of participants. Trans women reported significantly more structural risks on average compared to trans men and nonbinary participants assigned female at birth.

PrEP Obstacles Index

Participants reported an average of 1.7 PrEP obstacles (SD: 1.0), and just over half (55%) reported two or more obstacles that contributed to them not using PrEP. The most commonly reported obstacle was concerns regarding PrEP side effects (23%). No other PrEP obstacle was reported by more than 10% of respondents. The vast majority of participants had heard of PrEP before the study (90%).

Generalized Structural Equation Models

The results of the GSEM analyses are displayed in Tables 3, 4, 5 and visualized in the diagrams in Figs. 1–3. As hypothesized, sex assigned at birth was directly associated with PrEP uptake and directly associated with all mediators in the model except healthcare utilization. Furthermore, all paths in Model 1 with the full pooled sample (see Fig. 1) were significant except the paths from sex assigned at birth and the structural risk index to healthcare utilization. Compared to participants assigned male at birth, participants assigned female at birth reported significantly fewer behavioral risks (Coef: −0.208, P Value: 0.032, 95% CI: −0.400 – −0.020), fewer structural risks (Coef: −0.582 P Value < 0.001, 95% CI: −0.800 – −0.360), were more likely to report no perceived need for PrEP (AOR: 1.258, P Value < 0.001, 95% CI: 1.150 – 1.363), and were less likely to have ever taken PrEP (AOR: 0.883, P Value < 0.001, 95% CI: 0.827 – 0.942).

Table 3 Generalized structural equation model of PrEP uptake among TGEYYA, unstratified, ATN cross-network analysis of CARES and TechStep, 2017-2021, model 1: full sample, (N = 477)
Full size table
Table 4 Generalized structural equation model for PrEP uptake among TGEYYA, stratified by sex assigned at birth, ATN cross-network analysis of CARES and TechStep, 2017-2021, model 2: assigned male at birth only, (n = 191)
Full size table
Table 5 Generalized structural equation model for PrEP uptake among TGEYYA, stratified by sex assigned at birth, ATN cross-network analysis of CARES and TechStep, 2017-2021, model 3: assigned female at birth only, (n = 286)
Full size table

In addition to the direct effects observed between sex assigned at birth and PrEP uptake, structural and behavioral risks significantly mediated the relationship between sex assigned at birth and perceived need for PrEP, which, in turn mediated the relationship between structural and behavioral risks and PrEP uptake. Those reporting more structural risks reported more behavioral risks (Coef: 0.219, P Value < 0.001, 95% CI: 0.140–0.300) and were less likely to report no perceived need for PrEP (AOR: 0.924, P Value < 0.001, 95% CI: 0.896 – 0951). Those who reported more behavioral risks were also less likely to report no perceived need for PrEP (AOR: 0.992, P Value < 0.001, 95% CI: 0.887–0.961), and those who reported no perceived need for PrEP were less likely to have ever taken PrEP (OR: 0.868, P Value < 0.001, 95% CI: 0.819–0.923).

Healthcare utilization significantly mediated the relationship between region and PrEP uptake but not between sex assigned at birth and PrEP uptake. Those who were tested, linked, and retained in care were more likely to have ever taken PrEP (AOR: 1.182, P Value < 0.001, 95% CI: 1.116–1.259), and those in the South were less likely to be tested, linked, and retained in care (AOR: 0.785 P Value < 0.001, 95% CI: 0.712–0.869).

When stratifying the model to include only those assigned male at birth (see Fig. 2), there were no significant differences observed between nonbinary participants assigned male at birth and trans women in behavioral risks, perceived need for PrEP, healthcare utilization, nor PrEP uptake. The only direct difference observed between these two groups was that nonbinary participants reported marginally fewer structural risks than trans women (AOR: −0.359, P Value: 0.048, 95% CI: −0.100–0.001).

Fig. 2
figure 2

Generalized structural equation model for PrEP uptake among TGEYYA, model 2: stratified, assigned male at birth only, ATN cross-network analysis of CARES and TechStep, 2017-2021, (n = 191)

Full size image

As observed in Model 1 with the full pooled sample, those reporting more structural risks again reported more behavioral risks (Coef: 0.161, P Value: 0.003, CI: 0.060–0.260) and were less likely to report no perceived need for PrEP (AOR: 0.951, P Value: 0.015, CI: 0.905–0.990); those with no perceived need for PrEP were less likely to have ever taken PrEP (AOR: 0.811, P Value < 0.001, 95% CI: 0.726–0.914); those who were tested, linked, and retained in care were more likely to have ever taken PrEP (AOR: 1.246, P Value < 0.001, 95% CI: 1.116–1.405), and those in the South were less likely to be tested, linked, and retained in care (AOR: 0.819, P Value: 0.007, 95% CI: 0.705–0.951).

When stratifying the model to include only those assigned female at birth (see Fig. 3), there were no significant differences between nonbinary participants and trans men in behavioral and structural risks nor PrEP uptake. However, nonbinary participants were more likely to report no perceived need for PrEP (AOR: 1.139, P Value: 0.018, 95% CI: 1.020–1.271) and less likely to be tested, linked, and retained in care (AOR: 0.819, P Value: 0.001, 95% CI: 0.733–0.914) than trans men.

Fig. 3
figure 3

Generalized structural equation model for PrEP uptake among TGEYYA, model 3: stratified, assigned female at birth only, ATN cross-network analysis of CARES and TechStep, 2017-2021, (n = 286)

Full size image

Again as observed in Model 1 with the full pooled sample, those who reported more structural risks also reported more behavioral risks (Coef: 0.279, P Value < 0.001, 95% CI: 0.160–0.400) and were less likely to report no perceived need for PrEP (AOR: 0.905, P Value < 0.000; 95% CI: 0.861–0.951); those who reported more behavioral risks were less likely to report no perceived need for PrEP (AOR: 0.905, P Value < 0.000, 95% CI: 0.861- 0.951); those who reported no perceived need for PrEP were less likely to have ever taken PrEP (AOR: 0.905, P Value: 0.003, 95% CI: 0.844–0.970); those who were tested, linked, and retained in care were more likely to have ever taken PrEP (AOR: 1.127, P Value < 0.001, 95% CI: 1.051–1.197); and participants from the South were less likely to be tested, linked, and retained in care (AOR: 0.726, P Value < 0.001, C95% I: 0.631–0.835).

Discussion

Findings from the current study identified potential mechanisms associated with disparities in PrEP uptake among 477 TGEYYA enrolled in two protocols in the National Adolescent Trials Network. Most participants in this sample would be considered PrEP candidates based on CDC guidelines [80, 81], due to recent sexual and substance use behaviors that increase the likelihood of HIV transmission; 85% reported at least one HIV risk behavior in the last 3 months. Most participants reported condomless vaginal or anal sex, almost half reported having multiple sex partners, and a third reported polysubstance use. Almost all participants were aware of PrEP as an HIV prevention option, but less than 1 in 3 participants assigned male at birth and 1 in 6 participants assigned female at birth had ever taken PrEP.

There was significantly higher lifetime PrEP uptake reported among those assigned male at birth compared to those assigned female at birth. Structural and behavioral risks and perceived need for PrEP mediated the relationship between sex assigned at birth and PrEP uptake, as hypothesized along the behavioral/perception pathway. Trans women reported the highest average number of both behavioral and structural risks of all gender identities. For those assigned female at birth, PrEP uptake was higher among trans men than nonbinary participants. Despite reporting lower behavioral and structural risks compared to those assigned male at birth, trans men and nonbinary participants assigned female at birth were engaging in some HIV transmission behaviors – such as having multiple concurrent sex partners – yet their perceived need for PrEP and actual use of PrEP was very low. This might be due in part to higher perceived susceptibility for HIV infection being associated with having partners who are cis male or perceiving one’s sexual or social networks to be high risk [17, 33]; however, this study was limited in its ability to assess peer influences and partner characteristics of participants, such as partner gender identity and sex assigned at birth due to incongruity in these partner variables between the CARES and TechStep studies. Social network and partner characteristics of TGEYYA will be important to investigate in future research on PrEP use among adolescents.

TGEYYA experience heightened susceptibility for HIV infection, yet our estimates of 16% lifetime PrEP uptake and 7% current PrEP use are lower than many other PrEP uptake estimates nationally. In 2021, the CDC estimated that 25% of the 1.2 million eligible people with indication for PrEP use received a PrEP prescription [81]. Our findings on PrEP uptake are also lower than a CDC study estimating 33.5% of trans women aged 18–29 years old had indicated PrEP use [82]. The inclusion of younger adolescents in our sample might contribute to these lower rates of PrEP uptake, as seen in other groups with PrEP indications, in which younger people have lower uptake [83]. Our findings are more in line with preliminary CDC data showing lower PrEP uptake among adolescents compared to adults [84]. Several barriers to PrEP use among adolescents and young adults have been described, including barriers related to cost and insurance, fear of disclosing gender identity or sexual orientation to parents and healthcare providers, and lower risk perceptions [85,86,87]. Yet, most of what is known about barriers and facilitators for PrEP use among adolescents are from cisgender male and trans women samples [15, 87]. For trans men and nonbinary adolescents and young adults who are PrEP candidates, more research is needed to understand their unique barriers to care and PrEP uptake.

Trans women reported the highest average number of both behavioral and structural risks of all TGEYYA, highlighting the importance of addressing syndemics among trans women to reduce new HIV infections among this highly impacted population. Our findings also demonstrate the need to intervene upstream (e.g., policies that allow gender affirming care in healthcare settings, anti-discrimination laws, etc.) to alter structural conditions that lead to elevated structural vulnerabilities and sexual and substance use behaviors among trans women [41, 42, 88]. This coincides with extensive research examining syndemic correlates of HIV transmission [3, 12, 27, 52], unmet structural needs [89], and PrEP uptake [48, 60, 90, 91] among trans women. Trans women experience many disparities in structural factors (e.g., discrimination, housing, incarceration), which impact behavioral risk (e.g., substance use, multiple sex partners, exchange sex, and condomless anal sex) for HIV transmission [42, 88].

HIV testing and prevention services often seek to engage trans women because they experience inequities in HIV infection rates [92]. Since HIV testing and linkage to care services would associate behavioral risks with the need for PrEP, those engaged in care might be more likely to view themselves as a PrEP candidate to reduce their HIV transmission risk [50]. Our findings confirm that sexual risk behaviors and perceived need for PrEP are important drivers for PrEP uptake generally [93, 94], and for trans women specifically [91]. Similar findings of higher PrEP uptake among trans women compared to trans men and nonbinary individuals have been found in national samples of young gender minorities [95], which is consistent with early PrEP trials, rollout, and marketing campaigns being focused almost exclusively on trans women compared to other gender minority groups [60]. However, these results contradict two previous studies, in which trans masculine youth were more willing to or more likely to be on PrEP compared to trans feminine youth [14, 96].

These results also highlight important information for nonbinary adolescents assigned male at birth. Our results found that nonbinary AMAB participants reported higher occurrence of condomless anal or vaginal sex compared to trans women and no difference in reporting multiple partners or polysubstance use. Yet trans women reported significantly more structural risks, which may be contributing to slightly higher mean number of behavioral risks among trans women compared to nonbinary AMAB participants. Emerging research comparing binary and nonbinary trans youth have found similar behavioral risk profiles between these gender identity groups[6]. Yet, few studies specifically look at the unique needs of nonbinary youth assigned male at birth and instead combine them with nonbinary youth assigned female at birth [89].

These findings also contribute to emerging research on PrEP uptake among adolescents assigned female at birth. While trans men reported fewer numbers of behavioral risks than trans women, a majority reported condomless vaginal or anal sex and almost half had multiple sex partners in the last 3 months. There were no significant differences across the four gender identity groups for having multiple sex partners. In comparison to trans women, trans men and nonbinary individuals assigned female at birth are understudied in research examining HIV transmission risk and PrEP uptake [48]. Researchers have often erroneously assumed that most trans men practice sexual behaviors that do not put them at risk for HIV due to conflation of gender identity and sexual identity assuming heteronormative preferences (e.g., trans men assumed to have a sexual preference for cisgender women) [58, 97]. However, for trans men who have sex with cisgender men or trans women and practice condomless sex in the absence of PrEP use, risk for HIV is elevated to the same extent as cisgender men who have sex with men [54, 98]. Recent research suggests that trans men also engage in elevated substance use behaviors and exhibit syndemic mental health conditions, including high levels of depression and anxiety, all of which are associated with increased sexual risk behaviors and high STI incidence [55, 56]. Despite these elevated risk behaviors, HIV testing and PrEP campaigns have largely ignored trans men and nonbinary individuals assigned female at birth, which have left these populations with lower utilization of HIV prevention services, lower risk perceptions for HIV, and less interest in taking PrEP [48]. In our study, nonbinary participants, in particular, were less engaged in care compared to trans individuals.

A pillar of many HIV prevention campaigns is the push to get those at most risk for HIV infection into medical care and HIV testing [47, 92]. Most participants in this sample had ever been tested for HIV and had a regular source of care, but only half were retained in routine medical care. Participants who had seen a medical doctor within the last 6 months were more likely to report lifetime PrEP use than those who were not recently engaged in medical care. Lower medical engagement has been associated with PrEP use among Black and Latina trans women [41]. Lower rates of HIV testing, linkage and retention in care among nonbinary participants of both sexes assigned at birth likely explained why no differences in healthcare utilization were observed based on sex assigned at birth in the unstratified model. Although no direct effects on PrEP uptake were observed by region, the indirect relationship between region and PrEP uptake mediated by healthcare utilization suggests the importance of reducing barriers to care for TGEYYA in the South as well as for nonbinary youth across the country [29].

Several limitations should be considered in light of these findings. First, we were unable to assess willingness and intentions to start PrEP because the PrEP Motivational Cascade was not used in the CARES study, a useful measure which might provide additional psychological insight into motivations for taking PrEP, while also considering the impact of structural and behavioral barriers to initiating and persisting on PrEP [51]. Second, PrEP uptake was assessed based on lifetime use due to low current use, so the lifetime outcome might not align with behavioral risk data collected in the last 3 months. Third, differences in the recruitment timing and procedures of the two studies (i.e., mostly in-person recruitment before the COVID-19 pandemic for CARES vs. mostly online recruitment during the COVID-19 pandemic for TechStep) may have led to different characteristics of each sample or different patterns of responses. Given that most TechStep participants enrolled during the COVID-19 pandemic, it is possible that unmet structural needs and fewer behavioral risks reported by TechStep participants were reflective of widespread trends of increased social isolation and staying at home across the U.S. rather than the specific life circumstances of the TechStep participants [99]. Fourth, we were unable to account for partner HIV status due to incongruity in the questions asked between studies, which is an important factor for determining PrEP eligibility and perceived need for PrEP. Nonetheless, the high prevalence of recent condomless anal and vaginal sex and having multiple sex partners suggests that most participants met CDC PrEP indication criteria even in the absence of information on partner HIV status [63].

Additionally, we limited the number of parameters in the GSEM models to maximize power and ensure convergence and interpretability, so some pathways, covariates, and moderators were excluded from the final models. As such, omitted variable bias could not be entirely ruled out. First, the pathway between no perceived need for PrEP and healthcare utilization was not included in the GSEM models because these variables were not associated in the unadjusted bivariate model (OR: 0.899, 95% CI: 0.626–1.291, p = 0.565) nor in the adjusted multivariable model when controlling for other covariates (AOR: 1.260, 95% CI: 0.813–1.953, p = 0.301). This lack of association could be due to the healthcare utilization variable being too general (i.e., linkage to a usual source of care and doctor visits for any reason), whereas a more specific measure related to utilization of HIV prevention services would be expected to correlate with perceived need for PrEP. Second, age was not included as a covariate in the GSEM models because age was not associated with PrEP uptake or the mediators. No observed correlations between age and any of the outcomes of interest for this analysis might be due to the narrow age range of adolescents in our sample with less than 10 years between the youngest and oldest study participants. The majority of participants (72%) were between 20 and 24 years old at enrollment and less than 10% were below 18 years old. Future studies with individuals at different life course stages should evaluate potential age disparities in PrEP use among TGEYYA, as well as disparities by other demographic characteristics such as race/ethnicity and educational attainment. Finally, the relationships between gender identity and mediators could have been conditional on other factors. Methods to incorporate interactions into GSEM were limited by the sample size and underscore additional directions for future research. However, the consistent results of our sensitivity analyses modeling the outcome and mediators separately using regression models with all covariates suggest that the inclusion of additional parameters in the GSEM models likely would not change the key findings of this analysis.

Despite these limitations, our study provides important insights on mechanisms of PrEP uptake among TGEYYA and multiple points of intervention to reduce PrEP disparities and improve uptake across the gender identity spectrum. This ATN cross-network analysis included a diverse sample of TGEYYA with multiple intersecting gender, sexual, and racial/ethnic minority identities, as well as a broad geographic reach from six major cities across the West, South and Northeast regions of the U.S. In addition to characterizing key drivers of PrEP use among TGEYYA across the country, our findings also highlight the need to reduce barriers to care for trans youth and young adults in the South and among nonbinary youth and young adults throughout the U.S. who remain underserved by existing HIV prevention efforts.