Abstract
Purpose
Adherence to pre-exposure prophylaxis (PrEP) during periods of PrEP-indication (i.e., prevention-effective adherence) is critical for preventing HIV. We sought to describe factors associated with prevention-effective adherence trajectories among transgender women (TW) to inform PrEP implementation strategies.
Methods
Using data from The LITE American Cohort (n=728), we performed group-based multi-trajectory modeling (GBMTM) to identify clusters of TW with similar trajectories of PrEP adherence and indication, and sociodemographic, biobehavioral, and structural correlates of each trajectory.
Results
We identified five trajectories: 1) consistent indication/no PrEP (15.3%), 2) initial indication/no PrEP (47.1%), 3) declining indication/discontinued PrEP (9.5%), 4) consistent indication/PrEP adherent (18.5%), and 5) increasing indication/initiated PrEP (9.6%). TW diagnosed with an STI were more likely to follow a consistent indication/no PrEP trajectory compared to consistent indication/PrEP adherent trajectory (adjusted Relative Risk Ratio [aRRR]:2.54; 95%CI:1.16–5.57). TW who experienced homelessness were more likely to follow PrEP discontinuation and initiation trajectories relative to PrEP adherence (aRRR:2.71; 95%CI:1.10–6.70 and 2.83; 95%CI:1.13–7.05, respectively).
Conclusions
Over a quarter of TW followed trajectories suggestive of prevention-effective adherence, while 15% did not initiate PrEP despite consistent indication. Findings highlight missed opportunities for PrEP engagement at STI diagnosis and suggest structural interventions addressing housing instability may improve prevention-effective adherence among TW.
Keywords: Pre-Exposure Prophylaxis, Transgender Women, HIV prevention, Group-based multi-trajectory modeling, United States
INTRODUCTION
Globally, transgender women (TW) are highly impacted by the HIV epidemic, with prevalence estimated at 19%.1 In the United States (US), approximately 14% of TW are living with HIV, with over half of new diagnoses occurring among TW in the South.2,3 Due to intersecting systems of oppression (e.g., racism, transphobia, xenophobia), 90% of incident infections are diagnosed among TW of color leading to HIV prevalence estimates of 44–62% and 26–35% among Black and Latina TW, respectively.2,3
Despite the established need for multi-level interventions to prevent HIV among TW, a recent review of HIV prevention interventions designed or adapted for transgender people identified only 13 evidence-based interventions, 11 of which were behavioral.8 The only evidence-based biomedical prevention intervention currently approved in the US for TW is pre-exposure prophylaxis (PrEP). However, the sole study reporting PrEP efficacy data specifically for TW, a subgroup intention-to-treat analysis of the iPrEx trial, found PrEP did not reduce HIV acquisition risk among TW due to suboptimal adherence.9
Qualitative research has described homelessness, lack of insurance coverage, discrimination in healthcare, violence, and other sociostructural factors as barriers to PrEP initiation and adherence among TW.4–6 The higher burden of discrimination, and social and economic marginalization among Black and Latina TW suggests that the groups most likely to benefit from PrEP also have the highest barriers, with the potential to reinforce existing HIV disparities if inequities in PrEP access are not directly addressed.7
A recent review of literature on PrEP outcomes in cisgender and transgender women in the US found just six studies with samples comprised exclusively of TW.10 An additional 40 studies included TW alongside other populations (predominately cisgender men who have sex with men), but the majority reported no findings specific to TW.10 One study utilizing PrEP programmatic data from a hospital-based PrEP clinic in Atlanta (N=42) reported 62% of TW who initiated PrEP persisted in care at six months post-referral.11 The use of programmatic data prevented authors from ascertaining changes in PrEP indication; thus it was assumed that TW initially prescribed PrEP had consistent PrEP indication, which does not account for time-varying HIV risk.21 Other studies have reported suboptimal PrEP uptake,12, 13 adherence,9, 14–16 and persistence17 among TW, but many had small sample sizes, short follow-up periods, failed to account for time-varying PrEP-indication, or reported on data from clinical trials, which make the generalizability of these findings to real-world settings unclear.18 Thus, there is a dearth of information on quantitative, longitudinal PrEP engagement outcomes among TW to guide public health interventions. Given the promise of PrEP for preventing HIV in TW alongside evidence of suboptimal engagement in the PrEP continuum, there is a need for differentiated implementation strategies for this population.15,19,20
One challenge of defining PrEP adherence targets for populations at increased risk for HIV acquisition (e.g., TW), is the need to consider time-varying HIV-risk.21 As Haberer and colleagues note, the adherence paradigm for HIV treatment is not appropriate for PrEP; unlike antiretrovirals for HIV treatment, which require sustained, lifelong adherence to achieve desired health outcomes, PrEP is effective so long as adherence is high during periods of exposure not protected by other effective strategies (e.g., condoms).21 Further, PrEP use during periods of non-indication may be associated with costs to the individual (e.g., unnecessary side effects) and health system (e.g., resource costs). Haberer and colleagues propose an alternative paradigm for understanding and measuring PrEP adherence, termed “prevention-effective adherence”.21 Following Haberer’s conceptualization of prevention-effective adherence, which measures PrEP adherence alongside dynamic sociobehavioral HIV risk indicators (i.e., PrEP indication), we sought to describe the distribution and correlates of prevention-effective adherence trajectories over 18 months among a large, multi-site cohort of TW.21
METHODS
Study design and participants
The LITE study is a multi-site, technology-enhanced epidemiologic cohort examining multi-level risks for HIV acquisition and other health outcomes among adult TW in the US. TW (assigned male sex at birth and identify as women or feminine gender identity) who are ≥18 years old, fluent in English and/or Spanish, and reside in the Eastern and Southern US are eligible. LITE is comprised of a site-based arm (with annual facility-based visits and the option of facility-based or remote visits for the interim, quarterly timepoints) and a fully remote arm (with semi-annual remote visits). Visits include sociobehavioral surveys and HIV testing, with site-based participants receiving annual testing for bacterial sexually transmitted infections (STIs; syphilis, gonorrhea, and chlamydia). Survey domains include sociodemographics, gender-affirming experiences, general physical and mental health, PrEP engagement, substance use, sexual health, and experiences of discrimination and violence. Due to a staggered launch and differences in visit frequency, the present analysis is restricted to site-based participants, who reside in six cities: Atlanta, GA; Baltimore, MD; Boston, MA; Miami, FL; New York, NY; and Washington, D.C. Detailed study procedures, including recruitment strategies and sample size calculations, are available in published protocols.22,23 All study procedures were approved by the Johns Hopkins School of Medicine Institutional Review Board (IRB). Written informed consent was obtained from all participants prior to enrollment.
Data sources and definitions
Surveys were self-administered; however, participants who screened positive for low literacy, needed assistance using the tablet, or requested a research assistant read questions aloud in their preferred language were provided that option. Data from the first seven quarterly time-points (i.e., baseline through 18-months of enrollment) were used in this analysis. Static indicators (e.g., race and ethnicity) were ascertained at baseline, while time-varying indicators (e.g., recent homelessness) were assessed quarterly.
The following self-reported measures from baseline were included in this analysis: age (dichotomized as 18–24 and ≥25 years old); race (Black or African American versus not Black or African American); ethnicity (Latina or Hispanic versus not Latina or Hispanic); census division (New England, Mid-Atlantic, and South Atlantic); education (high school education or less versus some college or more); time since last healthcare encounter (within the last 6 months versus ≥6 months); and anticipated discrimination in healthcare (determined by agreement with the statement “Because of who I am, a doctor or other healthcare provider might treat me poorly” from the Intersectional Discrimination Index-A).24 The following time-varying variables—homelessness, arrest by police, uninsured health insurance status, and current partner PrEP use—were dichotomized as “1” if a participant reported the experience during at least one of seven study visits versus “0” otherwise. We identified incident STIs through laboratory-confirmed testing at baseline and 12-month follow-up as described in the study protocol.23
The first outcome, PrEP adherence, indicates the number of reported PrEP doses taken in the last seven days and ranged from 0 (no PrEP use) to 7 (daily adherence). All participants who indicated they were not currently taking PrEP were coded 0. TW reporting current PrEP use were asked, “In the past 7 days, how many days have you missed taking a dose of PrEP?” This item was inverse coded so that TW reporting 0 missed doses were assigned an adherence value of 7, those reporting 1 missed dose were coded as 6, etc.
The second outcome, PrEP indication, is a composite variable based on current CDC PrEP prescribing guidelines, which we have previously adapted for TW.25 Although PrEP indication is not a perfectly sensitive indicator of HIV acquisition risk, it is a widely used metric to distinguish those at elevated risk of HIV acquisition, and important for characterizing risk heterogeneity within key populations. TW were indicated for PrEP if they had biologically-confirmed HIV-negative serostatus, a sex partner with flesh penis within the past 6 months, were not in a monogamous sexual partnership with an HIV-negative partner, and reported at least one of the following recent indicators (i.e., within past 3 months): a) condomless anal sex, b) STI diagnosis, c) sex work, d) use of postexposure prophylaxis, e) condomless anal or vaginal sex with a partner living with HIV or of unknown HIV status, and f) needle sharing. This outcome was dichotomized as PrEP-indicated or not PrEP-indicated at each visit.
Statistical analysis
We calculated the proportion of TW indicated for PrEP and the proportion using PrEP at any visit over follow-up. After restricting the analytic sample to only those who were either PrEP-indicated and/or using PrEP during at least one visit, we performed group-based multi-trajectory modeling (GBMTM), a data-driven approach for identifying latent longitudinal strata across distinct but related indicators, to determine whether there were distinctive patterns of prevention-effective adherence.26 Following the model selection process set forth by Nagin, we explored specific patterns in each of the outcomes (PrEP adherence and indication) individually before modeling the joint outcome.26 Model selection and adequacy was assessed following Nagin’s diagnostic criteria (see Supplementary file).27
Following specification of the final model, we explored sociodemographic, biobehavioral, and structural indicators by group. We used multinomial logistic regression to identify correlates of trajectory groups. Predictors of group membership that were statistically significant at p<0.10 in bivariate analyses were included in the final multivariable model, which also included race, ethnicity, and census division. All variables had missingness ≤5%, thus we did not impute missing values. As a sensitivity analysis to assess for statistically significant changes in prep use and indication during the COVID-19 pandemic, we conducted interrupted time series analysis, accounting for secular trends. Analyses were conducted using Stata version 14, including the traj package for GBMTM specification.28
RESULTS
We enrolled 728 TW between March 2018 and August 2020, of whom 526/728 (72.3%) were indicated for PrEP within 18 months of enrollment. Among Latina and Black TW, 88.4% (168/190) and 87.7% (206/235) were indicated for PrEP within 18 months of enrollment, respectively. The majority of those who were PrEP-indicated reported no PrEP use over follow up (62.9%; 331/526). Over the same period, 209/728 (28.7%) self-reported PrEP use, 14 of whom were not PrEP-indicated (Figure 1). The remaining 188/728 (25.8%) were neither PrEP-indicated nor using PrEP at any time over follow-up. This group was predominately non-Black, non-Latina, college-educated, engaged in healthcare, and had health insurance (Table 1). Thus 540 participants were PrEP-indicated and/or reported taking PrEP at least once during 18 months of follow-up and were therefore included in subsequent group-based multi-trajectory modeling.
Figure 1. PrEP indication and use among transgender women in the LITE cohort.
Note: Figure 1 represents the full study sample (N=728). Subsequent analyses (i.e., those reported in Table 2 and Figure 2) were restricted to participants who were PrEP indicated and/or PrEP users (n=540; represented by blue and/or maroon circles). Characteristics of those not enclosed in either circle and thus excluded from additional analyses (i.e., those never indicated for nor using PrEP; n=188) are summarized in Table 1 (see Group 0).
Table 1.
Sociodemographic, clinical, and structural characteristics of transgender women in the LITE cohort by prevention-effective adherence trajectory (N=728)
| Factor | Full sample | Group 0: Never indicated/no PrEP | Group 1: Consistent indication/no PrEP | Group 2: Initial indication/no PrEP | Group 3: Declining indication/discontinued PrEP | Group 4: Consistent indication/PrEP adherent | Group 5: Increasing indication/initiated PrEP | p-value (χ2 test) |
|---|---|---|---|---|---|---|---|---|
| n | 728 | 188 | 77 | 263 | 52 | 99 | 49 | |
| n (%) | n (%) | n (%) | n (%) | n (%) | n (%) | n (%) | ||
| Youth (Aged 18–24) | 214 (29.4) | 43 (22.9) | 16 (20.8) | 103 (39.2) | 16 (30.8) | 18 (18.2) | 18 (36.7) | <0.001 |
| Black | 235 (32.6) | 29 (15.7) | 30 (39.0) | 88 (33.7) | 23 (45.1) | 39 (39.4) | 26 (55.3) | <0.001 |
| Latina | 190 (26.5) | 22 (11.8) | 23 (29.9) | 78 (30.5) | 18 (36.7) | 35 (35.4) | 14 (29.2) | <0.001 |
| Census division | <0.001 | |||||||
| New England | 158 (21.7) | 63 (33.5) | 11 (14.3) | 55 (20.9) | 8 (15.4) | 13 (13.1) | 8 (16.3) | |
| Mid-Atlantic | 221 (30.4) | 47 (25.0) | 21 (27.3) | 86 (32.7) | 22 (42.3) | 31 (31.3) | 14 (28.6) | |
| South Atlantic | 349 (47.9) | 78 (41.5) | 45 (58.4) | 122 (46.4) | 22 (42.3) | 55 (55.6) | 27 (55.1) | |
| College education or higher | 464 (64.4) | 141 (75.0) | 45 (58.4) | 164 (63.1) | 20 (39.2) | 69 (71.1) | 25 (52.1) | <0.001 |
| Homelessness¶ | 163 (22.5) | 23 (12.3) | 20 (26.0) | 67 (25.6) | 19 (36.5) | 17 (17.2) | 17 (34.7) | <0.001 |
| Arrest ¶ | 42 (5.8) | 2 (1.1) | 10 (13.0) | 17 (6.5) | 2 (4.0) | 5 (5.1) | 6 (12.2) | <0.01 |
| Uninsured¶ | 80 (11.6) | 16 (8.7) | 15 (19.7) | 31 (12.8) | 4 (7.8) | 8 (8.9) | 6 (13.0) | 0.15 |
| Lab-confirmed STI¶ | 102 (14.0) | 0 (0.0) | 30 (39.0) | 27 (10.3) | 12 (23.1) | 21 (21.2) | 12 (24.5) | <0.001 |
| One or more current partner(s) using PrEP¶ | 215 (29.7) | 18 (9.7) | 22 (28.6) | 61 (23.3) | 26 (51.0) | 61 (61.6) | 27 (55.1) | <0.001 |
| Six months or longer since last healthcare encounter | 53 (7.4) | 10 (5.4) | 13 (17.1) | 19 (7.3) | 2 (3.8) | 4 (4.0) | 5 (10.2) | 0.01 |
| Anticipated discrimination in healthcare | 301 (42.2) | 90 (47.9) | 31 (40.8) | 103 (40.7) | 21 (42.0) | 43 (43.4) | 13 (27.7) | 0.23 |
At any point over follow-up. All other indicators were assessed at baseline only.
We identified five prevention-effective adherence trajectories, classified as 1) consistent indication/no PrEP (15.3%; 95% CI:11.3–19.3), 2) initial indication/no PrEP (47.1%; 95% CI:42.1–52.1%), 3) declining indication/discontinued PrEP (9.5%; 95% CI:6.8–12.5%), 4) consistent indication/PrEP adherent (18.5%; 95% CI:15.0–22.0%), and 5) increasing indication/initiated PrEP (9.6%; 95% CI:7.0–12.2%). This five-group multi-trajectory model exceeded a priori criteria for model adequacy (Supplementary file).
Declining indication/discontinued PrEP and consistent indication/PrEP adherent trajectories were characterized by strong concordance between indication and adherence, though adherence was suboptimal during periods of indication for both groups (Figure 2). The remaining three groups were marked by periods of PrEP indication and non-use (i.e., non-prevention-effective adherence). The first group, consistent indication/no PrEP, demonstrated non-prevention-effective adherence over the entire follow-up.
Figure 2.
Prevention-effective adherence trajectories among transgender women in the LITE cohort (N=540)
Sociodemographic, clinical, and structural characteristics of each trajectory group are compared in Table 1. The majority of TW assigned to the consistent indication/no PrEP trajectory (58.4%) resided in the South. This group had the highest proportion of uninsured TW (19.7%), STI diagnosis over follow-up (39.0%), and disengagement from healthcare at enrollment (17.1%). The modal trajectory, initial indication/no PrEP, was largely comprised of young (39.2%) and college-educated (63.1%) TW. Apart from the never indicated/no PrEP trajectory, this group had the lowest STI period prevalence (10.3%). TW with declining indication/discontinued PrEP trajectories were predominately TW of color (45.1% Black and 36.7% Latina), least likely to be college-educated (39.2%), and most likely to report homelessness over follow-up (36.5%). The majority of TW with consistent indication/PrEP adherent trajectories were ≥25 years old (81.8%), college-educated (71.1%), and reported at least one partner using PrEP over follow-up (61.6%). Finally, the majority of TW with increasing indication/initiated PrEP trajectories were Black (55.3%), college-educated (52.1%), and resided in the South (55.1%). This group also had the lowest proportion of anticipated discrimination in healthcare (27.7%).
Table 2 reports unadjusted and adjusted relative risk ratios (aRRR) from multinomial logistic regression models to identify correlates of trajectory group membership. The reference group for all pairwise comparisons is the consistent indication/PrEP adherent group. Participants with an STI diagnosis over follow-up and those who had not had a healthcare encounter within 6 months of enrollment were more likely to follow a consistent indication/no PrEP trajectory (aRRR: 2.54; 95% CI:1.16–5.57 and aRRR: 5.66; 95% CI:1.40–22.85, respectively). Participants reporting one or more PrEP-using partner(s) were significantly less likely to follow a consistent indication/no PrEP trajectory (aRRR: 0.18; 95% CI: 0.09–0.36). Those aged 18–24 were more than three times as likely to follow an initial indication/no PrEP trajectory (aRRR: 3.04; 95% CI:1.60–5.78). Finally, experiencing homelessness over follow-up was associated with significantly greater relative risk of membership in declining indication/discontinued PrEP and increasing indication/initiated PrEP trajectories (aRRR: 2.71; 95% CI:1.10–6.70 and 2.83; 95% CI:1.13–7.05, respectively). In our sensitivity analysis, there was not a statistically significant difference in PrEP indication nor use during the COVID-19 pandemic after accounting for secular trends (Supplementary file).
Table 2.
Relative Risk Ratios (RRR) for prevention-effective adherence trajectory among transgender women indicated for and/or using PrEP in the LITE cohort (N=540)
| Unadjusted Relative Risk Ratios | Adjusted Relative Risk Ratios | |||||||
|---|---|---|---|---|---|---|---|---|
| Reference Group: Group 4: Consistent indication/PrEP adherent | RRR | P-value | 95% CI | aRRR | P-value | 95% CI | ||
|
| ||||||||
| Group 1: Consistent indication/no PrEP | ||||||||
|
| ||||||||
| Youth (Aged 18–24) | 1.18 | 0.67 | 0.56 | 2.5 | 1.18 | 0.69 | 0.52 | 2.66 |
| Black | 0.98 | 0.95 | 0.53 | 1.81 | 0.62 | 0.25 | 0.28 | 1.40 |
| Latina | 0.78 | 0.44 | 0.41 | 1.48 | 0.47 | 0.06 | 0.21 | 1.04 |
| Census division (ref=New England) | ||||||||
| Mid-Atlantic | 0.80 | 0.66 | 0.30 | 2.12 | 0.92 | 0.88 | 0.31 | 2.69 |
| South Atlantic | 0.97 | 0.94 | 0.40 | 2.36 | 1.02 | 0.96 | 0.37 | 2.86 |
| College education or higher (ref=high school diploma/GED or less) | 0.57 | 0.08 | 0.30 | 1.07 | 0.63 | 0.22 | 0.30 | 1.31 |
| Homelessness¶ | 1.69 | 0.16 | 0.82 | 3.51 | 2.07 | 0.09 | 0.88 | 4.82 |
| Arrest¶ | 2.81 | 0.07 | 0.92 | 8.59 | 2.56 | 0.16 | 0.69 | 9.52 |
| Uninsured¶ | 2.52 | 0.05 | 1.00 | 6.32 | 1.47 | 0.48 | 0.51 | 4.24 |
| Lab-confirmed STI¶ | 2.37 | 0.01 | 1.22 | 4.61 | 2.54 | 0.02 | 1.16 | 5.57 |
| 1 or more current partner using PrEP¶ | 0.25 | <0.001 | 0.13 | 0.47 | 0.18 | <0.001 | 0.09 | 0.36 |
| 6+ months since last healthcare encounter | 4.90 | <0.01 | 1.53 | 15.71 | 5.66 | 0.02 | 1.40 | 22.85 |
| Anticipated discrimination in healthcare | 0.90 | 0.73 | 0.49 | 1.64 | 1.02 | 0.95 | 0.50 | 2.07 |
| Group 2: Initial indication/no PrEP | ||||||||
|
| ||||||||
| Youth (Aged 18–24) | 2.90 | <0.001 | 1.64 | 5.11 | 3.04 | <0.001 | 1.60 | 5.78 |
| Black | 0.78 | 0.32 | 0.49 | 1.26 | 0.62 | 0.14 | 0.33 | 1.17 |
| Latina | 0.80 | 0.38 | 0.49 | 1.31 | 0.71 | 0.27 | 0.39 | 1.30 |
| Census division (ref=New England) | ||||||||
| Mid-Atlantic | 0.66 | 0.26 | 0.32 | 1.36 | 0.73 | 0.45 | 0.32 | 1.67 |
| South Atlantic | 0.52 | 0.06 | 0.26 | 1.04 | 0.8 | 0.58 | 0.36 | 1.78 |
| College education or higher (ref=high school diploma/GED or less) | 0.69 | 0.16 | 0.42 | 1.15 | 0.9 | 0.75 | 0.49 | 1.67 |
| Homelessness¶ | 1.66 | 0.09 | 0.92 | 2.99 | 1.87 | 0.10 | 0.89 | 3.94 |
| Arrest¶ | 1.31 | 0.61 | 0.47 | 3.65 | 1.03 | 0.97 | 0.28 | 3.81 |
| Uninsured¶ | 1.50 | 0.33 | 0.66 | 3.4 | 1.52 | 0.38 | 0.59 | 3.91 |
| Lab-confirmed STI¶ | 0.43 | <0.01 | 0.23 | 0.8 | 0.45 | 0.03 | 0.21 | 0.94 |
| 1 or more current partner using PrEP¶ | 0.19 | <0.001 | 0.12 | 0.31 | 0.15 | <0.001 | 0.09 | 0.27 |
| 6+ months since last healthcare encounter | 1.88 | 0.26 | 0.62 | 5.67 | 2.1 | 0.29 | 0.54 | 8.21 |
| Anticipated discrimination in healthcare | 0.89 | 0.64 | 0.56 | 1.43 | 0.83 | 0.51 | 0.46 | 1.47 |
| Group 3: Declining indication/discontinued PrEP | ||||||||
|
| ||||||||
| Youth (Aged 18–24) | 2.00 | 0.08 | 0.92 | 4.36 | 1.48 | 0.38 | 0.62 | 3.56 |
| Black | 1.26 | 0.50 | 0.64 | 2.50 | 0.81 | 0.63 | 0.34 | 1.93 |
| Latina | 1.06 | 0.87 | 0.52 | 2.16 | 0.82 | 0.64 | 0.36 | 1.88 |
| Census division (ref=New England) | ||||||||
| Mid-Atlantic | 1.15 | 0.79 | 0.41 | 3.25 | 1.32 | 0.64 | 0.41 | 4.23 |
| South Atlantic | 0.65 | 0.40 | 0.24 | 1.78 | 0.88 | 0.83 | 0.28 | 2.80 |
| College education or higher (ref=high school diploma/GED or less) | 0.26 | <0.001 | 0.13 | 0.53 | 0.28 | <0.01 | 0.13 | 0.63 |
| Homelessness¶ | 2.78 | <0.01 | 1.29 | 5.99 | 2.71 | 0.03 | 1.10 | 6.70 |
| Arrest¶ | 0.78 | 0.78 | 0.15 | 4.19 | 0.91 | 0.92 | 0.15 | 5.62 |
| Uninsured¶ | 0.87 | 0.83 | 0.25 | 3.05 | 0.76 | 0.69 | 0.20 | 2.93 |
| Lab-confirmed STI¶ | 1.11 | 0.79 | 0.5 | 2.49 | 1.03 | 0.95 | 0.41 | 2.60 |
| 1 or more current partner using PrEP¶ | 0.65 | 0.21 | 0.33 | 1.28 | 0.59 | 0.18 | 0.28 | 1.27 |
| 6+ months since last healthcare encounter | 0.95 | 0.95 | 0.17 | 5.37 | 1.79 | 0.54 | 0.27 | 11.8 |
| Anticipated discrimination in healthcare | 0.94 | 0.87 | 0.47 | 1.88 | 1.01 | 0.98 | 0.46 | 2.23 |
| Group 5: Increasing indication/initiated PrEP | ||||||||
|
| ||||||||
| Youth (Aged 18–24 at enrollment) | 2.61 | 0.02 | 1.21 | 5.66 | 2.15 | 0.09 | 0.90 | 5.15 |
| Black | 1.90 | 0.07 | 0.94 | 3.84 | 1.5 | 0.37 | 0.62 | 3.66 |
| Latina | 0.75 | 0.46 | 0.36 | 1.59 | 0.64 | 0.33 | 0.26 | 1.56 |
| Census division (ref=New England) | ||||||||
| Mid-Atlantic | 0.73 | 0.58 | 0.25 | 2.17 | 0.81 | 0.74 | 0.24 | 2.80 |
| South Atlantic | 0.80 | 0.66 | 0.3 | 2.16 | 0.88 | 0.84 | 0.27 | 2.91 |
| College education or higher (ref=high school diploma/GED or less) | 0.44 | 0.03 | 0.22 | 0.9 | 0.65 | 0.31 | 0.28 | 1.49 |
| Homelessness¶ | 2.56 | 0.02 | 1.17 | 5.63 | 2.83 | 0.03 | 1.13 | 7.05 |
| Arrest¶ | 2.62 | 0.13 | 0.76 | 9.07 | 1.99 | 0.38 | 0.43 | 9.17 |
| Uninsured¶ | 1.54 | 0.45 | 0.5 | 4.73 | 1.25 | 0.72 | 0.36 | 4.32 |
| Lab-confirmed STI¶ | 1.20 | 0.65 | 0.54 | 2.71 | 0.71 | 0.49 | 0.27 | 1.87 |
| 1 or more current partner using PrEP¶ | 0.76 | 0.45 | 0.38 | 1.53 | 0.78 | 0.53 | 0.35 | 1.72 |
| 6+ months since last healthcare encounter | 2.70 | 0.15 | 0.69 | 10.54 | 3.44 | 0.14 | 0.67 | 17.52 |
| Anticipated discrimination in healthcare | 0.50 | 0.07 | 0.23 | 1.06 | 0.51 | 0.13 | 0.21 | 1.23 |
At any point over follow-up. All other indicators were assessed at baseline only.
DISCUSSION
Available evidence, though limited, has raised concerns regarding suboptimal engagement in PrEP care among transgender women.9, 11–17 Operationalizing Haberer’s prevention-effective adherence paradigm, we aimed to describe the distribution of prevention-effective adherence trajectories over an 18-month period among a large observational sample of TW.21 We found that just over one in four TW at risk for HIV demonstrated patterns of prevention-effective adherence, adhering to PrEP during periods of indication and discontinuing PrEP when no longer indicated. Further, less than 10% initiated PrEP. We estimate that 15% of TW remain unengaged in PrEP services despite consistent indication for PrEP—pointing to an urgent need for tailored PrEP implementation strategies to meet the needs of this subset of TW. Importantly, LITE study locations either have gender-affirming PrEP providers on-site or facilitate referrals to gender-affirming PrEP providers for TW who are interested. Thus, our study population likely has greater access to PrEP care relative to TW broadly--and yet multi-level barriers to engagement4–7 resulted in a substantial subset of TW underutilizing PrEP.
Our joint modeling of PrEP adherence and indication supports Haberer’s conceptualization of prevention-effective adherence by illustrating both periods of discordance and concordance. As additional PrEP delivery methods become available (e.g., bimonthly injections, yearly implants, and event-driven microbicides), there will be greater opportunity to align PrEP indication and adherence to ensure adequate preventative coverage for concentrated periods of indication (e.g., single events) or prolonged periods of heterogenous risk. Future research should explore how these forthcoming delivery methods will address the unmet prevention needs of TW.
Sociodemographic correlates of trajectory group align with epidemiologic evidence demonstrating concentrated HIV acquisition risk and reduced PrEP accessibility among young, Black, and Latina TW.3, 29 We found that youth are overrepresented in the initial indication/no PrEP and increasing indication/initiated PrEP groups and underrepresented in the consistent indication/PrEP adherent group. This highlights that PrEP indication and adherence are dynamic among young TW and suggests event-driven and/or long-acting options may be more appropriate modalities for youth. Given that HIV prevalence among TW aged 25–29 is nearly four times higher than prevalence among TW aged 18–24, interventions that optimize prevention-effective adherence during this life stage are critical.3
Black participants comprised over half of those who initiated PrEP. However, to address the inequities in new HIV diagnoses among Black TW, efforts to further increase access and support adherence (e.g., transportation support, telemedicine scale-up, provision of housing services, peer navigation programs, expanding access through community pharmacies) among Black TW are needed.3 Latina participants comprised a third of PrEP indicated TW and were relatively evenly distributed among trajectory groups. Notably, 87.7% and 88.4% of Black and Latina TW (respectively) were indicated for PrEP compared to 72.3% of the overall sample, demonstrating the importance of PrEP programming that is acceptable and accessible for Black and Latina TW.
Experiences of homelessness were prevalent across groups. TW who experienced homelessness were more than twice as likely to follow trajectories of discontinuation and initiation compared to consistent indication/PrEP adherence. Although further research is warranted to explore the mechanisms by which housing stability impacts prevention-effective adherence, findings point to housing as an important predictor of longitudinal PrEP engagement. TW lacking health insurance were overrepresented in the consistent indication/no PrEP group, which suggests lack of health insurance may be an important barrier. Policies and programs such as Medicaid expansion and low-cost care provision through Federally Qualified Health Centers may increase PrEP engagement.
We found high rates of arrest in our sample (5.8%), with sex work as the most common reason for arrest. A significantly higher proportion of TW reported recent arrest in the consistent indication/no PrEP and increasing indication/initiated PrEP groups (13.0% and 12.2%, respectively). Risk of arrest and HIV vulnerability share common drivers including discrimination, violence victimization, poor mental health, and poverty; policy interventions and human rights centered approaches have the potential to address these drivers, which disproportionately impact Black and Latina TW.30 As arrest and incarceration can lead to medication lapses, long-acting PrEP formulations may increase prevention-effective adherence among communities disproportionately impacted by over-policing and mass incarceration.
STIs were prevalent in all groups but were more than twice as common in the consistent indication/no PrEP group, which demonstrates that condomless sex is prevalent and PrEP need remains high. TW in this group are likely to benefit from combination prevention products in development such as microbicidal enemas that prevent HIV and STIs. TW who reported having a partner who uses PrEP were more likely to adhere to PrEP compared to those that do not engage with PrEP despite indication. This finding is consistent with existing literature demonstrating that social relationships and disclosure are associated with PrEP use; the role of dyadic and social network interventions warrant further exploration.31
Those who were disengaged from healthcare were overrepresented in the consistent indication/no PrEP group. There is evidence that providing higher priority services (e.g., gender-affirming hormones, mental healthcare, substance use treatment) can increase engagement in HIV prevention and care.11, 20 Anticipated discrimination in healthcare was prevalent overall (42.2%) but lower (though not statistically significant) in the PrEP initiation group (27.7%). Access to gender-affirming healthcare services that respect the needs and lived experiences of TW is critical for PrEP intervention success. This measure of anticipated healthcare discrimination is part of an intersectional discrimination scale which includes other forms of discrimination beyond transphobia (e.g., racism, sexism, xenophobia). Thus, while gender-affirming services are a requisite for successful PrEP programming, these healthcare services and environments must also be inclusive and affirming of the other intersectional identities among TW they serve, particularly Black and Latina TW.
Strengths of this analysis include the large prospective sample comprised exclusively of TW, the innovative application of GBMTM to operationalize prevention-effective adherence trajectories, and the diversity of the sample in terms of age, race, and ethnicity. However, the study had some limitations. All data (except STI diagnosis) were self-reported and therefore subject to social desirability bias. Some measures (e.g., anticipated healthcare discrimination and healthcare engagement) were only assessed at baseline. Additional research using biological endpoints for adherence and exploring the impact of time-varying measures of healthcare access on PrEP engagement is needed. In addition, the measures of association reported were correlations (relative risk ratios). The direction of the effect between sociodemographic, clinical, and structural characteristics and the prevention-effective adherence trajectories cannot be assumed. Further quantitative and qualitative research to test the direction of these effects, characterize causal pathways, and inform targeted interventions based on prevention-effective adherence trajectory is underway. Finally, this analysis includes data from participants who completed follow-up prior to March 2020 and those who contributed data after March 2020; it is possible that the COVID-19 pandemic resulted in changes in PrEP indication and adherence in this sample, as has been reported in other populations in the US, however we did not find statistically significant differences in our sensitivity analysis after accounting for secular trends.32
Our analysis identified the distribution of prevention-effective adherence trajectories over an 18-month period among TW in a large, diverse, multi-site cohort in the US. We identified sociodemographic, clinical, and structural correlates, which highlight the importance of linkage to PrEP at STI diagnosis, housing interventions to support PrEP engagement and adherence, and the potential of dyadic and social network interventions to increase PrEP use among TW.
Supplementary Material
Acknowledgments and funding
EEC is supported by a predoctoral fellowship from the National Institute of Mental Health (F31MH124582). The LITE study is jointly supported by the National Institute of Allergy and Infectious Diseases, the National Institute of Mental Health, and the National Institute of Child Health and Human Development of the National Institutes of Health under Award Number UG3/UH3AI133669 (ALW and SLR). Research reported in this publication was also supported by HIV/AIDS, Hepatitis, STD, and TB Administration (HAHSTA), Washington, DC, Department of Health. The LITE study is also appreciative of support from the CFAR at partner institutions, including JHU (P30AI094189), Emory University (P30AI050409), Harvard University (P30AI060354), DC CFAR (P30AI117970), and the University of Miami (P30AI073961). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or HAHSTA. The authors thank the transgender women who took part in this study. This study would not have been possible without their participation. The authors also acknowledge the work of the entire American Cohort To Study HIV Acquisition Among Transgender Women team: Andrea Wirtz (multiple PI; Johns Hopkins University (JHU)); Sari Reisner (multiple PI; Harvard University); Keri Althoff (JHU); Chris Beyrer (JHU); James Case (JHU); Erin Cooney (JHU); Oliver Laeyendecker (JHU); Megan Stevenson (JHU); Elizabeth Humes (JHU); Jeffrey Herman (JHU); Tonia Poteat (University of North Carolina); Kenneth Mayer (Fenway Health); Asa Radix (Callen-Lorde Community Health Center); Christopher Cannon (Whitman-Walker Institute); Jason Schneider (Emory University and Grady Hospital); Sonya Haw (Emory University and Grady Hospital); Allan Rodriguez (University of Miami); Andrew Wawrzyniak (University of Miami); the incredible research teams at each study site; and the LITE Community Advisory Board, including the following individuals: Sherri Meeks, Flora Marques, Sydney Shackelford, Nala Toussaint, and SaVanna Wanzer and those who have remained anonymous.
Footnotes
Declaration of interests
We declare no competing interests.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Data sharing
De-identified individual data, a data dictionary, and code will be made available upon reasonable request after approval of a proposal and signing of a data use agreement. There is a formal process for external users to request access to LITE data, which involves review and approval by Principal Investigators from each study site as well as the Community Advisory Board; further details and forms can be obtained by emailing Dr. Andrea Wirtz (awirtz1@jhu.edu).
References
- 1.Baral SD, Poteat T, Stromdahl S, Wirtz AL, Guadamuz TE, Beyrer C. Worldwide burden of HIV in transgender women: a systematic review and meta-analysis. Lancet Infect Dis. 2013;13(3):214–22. [DOI] [PubMed] [Google Scholar]
- 2.Becasen JS, Denard CL, Mullins MM, Higa DH, Sipe TA. Estimating the Prevalence of HIV and Sexual Behaviors Among the US Transgender Population: A Systematic Review and Meta-Analysis, 2006–2017. Am J Public Health. 2019;109(1):e1–e8. [Google Scholar]
- 3.Centers for Disease Control and Prevention. HIV Infection, Risk, Prevention, and Testing Behaviors Among Transgender Women—National HIV Behavioral Surveillance, 7 U.S. Cities, 2019–20.; 2021. [Google Scholar]
- 4.Sevelius JM, Deutsch MB, Grant R. The future of PrEP among transgender women: the critical role of gender affirmation in research and clinical practices. J Int AIDS Soc. 2016;19(7(Suppl 6)):21105. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Cahill SR, Keatley J, Wade Taylor S, Sevelius J, Elsesser SA, Geffen SR, et al. “Some of us, we don’t know where we’re going to be tomorrow.” Contextual factors affecting PrEP use and adherence among a diverse sample of transgender women in San Francisco. AIDS Care. 2020;32(5):585–93. [DOI] [PubMed] [Google Scholar]
- 6.Eaton LA, Kalichman SC, Price D, Finneran S, Allen A, Maksut J. Stigma and Conspiracy Beliefs Related to Pre-exposure Prophylaxis (PrEP) and Interest in Using PrEP Among Black and White Men and Transgender Women Who Have Sex with Men. AIDS Behav. 2017;21(5):1236–46. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Poteat T, Wirtz A, Malik M, Cooney E, Cannon C, Hardy WD, et al. A Gap Between Willingness and Uptake: Findings From Mixed Methods Research on HIV Prevention Among Black and Latina Transgender Women. J Acquir Immune Defic Syndr. 2019;82(2):131–40. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Poteat T, Malik M, Scheim A, Elliott A. HIV Prevention Among Transgender Populations: Knowledge Gaps and Evidence for Action. Curr HIV/AIDS Rep. 2017;14(4):141–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Deutsch MB, Glidden DV, Sevelius J, Keatley J, McMahan V, Guanira J, et al. HIV pre-exposure prophylaxis in transgender women: a subgroup analysis of the iPrEx trial. Lancet HIV. 2015;2(12):e512–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Baldwin A, Light B, Allison WE. Pre-Exposure Prophylaxis (PrEP) for HIV Infection in Cisgender and Transgender Women in the US: A Narrative Review of the Literature. Archives of Sexual Behavior. 2021:1–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Aldredge A, Roth G, Vaidya A, Paula Duarte A, Kundu S, Zheng Z, et al. Preexposure prophylaxis care continuum among transgender women at a patient-centered preexposure prophylaxis program in Atlanta, Georgia. Aids. 2021;35(3):524–6. [DOI] [PubMed] [Google Scholar]
- 12.Eaton LA, Matthews DD, Driffin DD, Bukowski L, Wilson PA, Stall RD. A Multi-US City Assessment of Awareness and Uptake of Pre-exposure Prophylaxis (PrEP) for HIV Prevention Among Black Men and Transgender Women Who Have Sex with Men. Prev Sci. 2017;18(5):505–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Reisner SL, Jadwin-Cakmak L, White Hughto JM, Martinez M, Salomon L, Harper GW. Characterizing the HIV Prevention and Care Continua in a Sample of Transgender Youth in the U.S. AIDS Behav. 2017;21(12):3312–27. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Kimani M, van der Elst EM, Chirro O, Wahome E, Ibrahim F, Mukuria N, et al. “I wish to remain HIV negative”: Pre-exposure prophylaxis adherence and persistence in transgender women and men who have sex with men in coastal Kenya. PLoS One. 2021;16(1):e0244226. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Wilson EC, Turner CM, Arayasirikul S, Lightfoot M, Scheer S, Raymond HF, et al. Disparities in the PrEP continuum for trans women compared to MSM in San Francisco, California: results from population-based cross-sectional behavioural surveillance studies. J Int AIDS Soc. 2020;23 Suppl 3(Suppl 3):e25539. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Grinsztejn B, Hoagland B, Moreira RI, Kallas EG, Madruga JV, Goulart S, et al. Retention, engagement, and adherence to pre-exposure prophylaxis for men who have sex with men and transgender women in PrEP Brasil: 48 week results of a demonstration study. Lancet HIV. 2018;5(3):e136–e45. [DOI] [PubMed] [Google Scholar]
- 17.Reback CJ, Clark KA, Runger D, Fehrenbacher AE. A Promising PrEP Navigation Intervention for Transgender Women and Men Who Have Sex with Men Experiencing Multiple Syndemic Health Disparities. J Community Health. 2019;44(6):1193–203. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Fontanari AMV, Zanella GI, Feijó M, Churchill S, Lobato MIR, Costa AB. HIV-related care for transgender people: A systematic review of studies from around the world. Social Science & Medicine. 2019;230:280–94. [DOI] [PubMed] [Google Scholar]
- 19.Deutsch MB. Pre-Exposure Prophylaxis in Trans Populations: Providing Gender-Affirming Prevention for Trans People at High Risk of Acquiring HIV. LGBT Health. 2018;5(7):387–90. [DOI] [PubMed] [Google Scholar]
- 20.Sevelius JM, Keatley J, Calma N, Arnold E. ‘I am not a man’: Trans-specific barriers and facilitators to PrEP acceptability among transgender women. Glob Public Health. 2016;11(7–8):1060–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Haberer JE, Bangsberg DR, Baeten JM, Curran K, Koechlin F, Amico KR, et al. Defining success with HIV pre-exposure prophylaxis: a prevention-effective adherence paradigm. AIDS. 2015;29(11):1277–85. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Wirtz AL, Cooney EE, Stevenson M, Radix A, Poteat T, Wawrzyniak AJ, et al. Digital Epidemiologic Research on Multilevel Risks for HIV Acquisition and Other Health Outcomes Among Transgender Women in Eastern and Southern United States: Protocol for an Online Cohort. JMIR Res Protoc. 2021;10(4):e29152. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Wirtz AL, Poteat T, Radix A, Althoff KN, Cannon CM, Wawrzyniak AJ, et al. American Cohort to Study HIV Acquisition Among Transgender Women in High-Risk Areas (The LITE Study): Protocol for a Multisite Prospective Cohort Study in the Eastern and Southern United States. JMIR Res Protoc. 2019;8(10):e14704. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Scheim AI, Bauer GR. The Intersectional Discrimination Index: Development and validation of measures of self-reported enacted and anticipated discrimination for intercategorical analysis. Soc Sci Med. 2019;226:225–35. [DOI] [PubMed] [Google Scholar]
- 25.Malone J, Reisner SL, Cooney E, Poteat T, Cannon CM, Schneider J, et al. Perceived HIV Acquisition Risk and Low Uptake of PrEP Among a Cohort of Transgender Women With PrEP Indication in the Eastern and Southern United States. J Acquir Immune Defic Syndr. 2021. Jul 14. [Google Scholar]
- 26.Nagin DS, Jones BL, Passos VL, Tremblay RE. Group-based multi-trajectory modeling. Stat Methods Med Res. 2018;27(7):2015–23. [DOI] [PubMed] [Google Scholar]
- 27.Nagin D Group-based modeling of development: Harvard University Press; 2009. [Google Scholar]
- 28.Jones BL, Nagin DS. A note on a Stata plugin for estimating group-based trajectory models. Sociological Methods & Research. 2013;42(4):608–13. [Google Scholar]
- 29.Sharpe JD, Guest JL, Siegler AJ, Sanchez TH, Sullivan PS. The spatiotemporal distribution of pre-exposure prophylaxis accessibility in the United States, 2016–2020. Annals of Epidemiology. 2021. Dec 1;64:102–10. [DOI] [PubMed] [Google Scholar]
- 30.Poteat TC, Malik M, Beyrer C. Epidemiology of HIV, Sexually Transmitted Infections, Viral Hepatitis, and Tuberculosis Among Incarcerated Transgender People: A Case of Limited Data. Epidemiol Rev. 2018;40(1):27–39. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Mehrotra ML, Rivet Amico K, McMahan V, Glidden DV, Defechereux P, Guanira JV, et al. The Role of Social Relationships in PrEP Uptake and Use Among Transgender Women and Men Who Have Sex with Men. AIDS Behav. 2018;22(11):3673–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Pampati S, Emrick K, Siegler AJ, Jones J. Changes in Sexual Behavior, PrEP Adherence, and Access to Sexual Health Services Because of the COVID-19 Pandemic Among a Cohort of PrEP-Using MSM in the South. J Acquir Immune Defic Syndr. 2021:639–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
De-identified individual data, a data dictionary, and code will be made available upon reasonable request after approval of a proposal and signing of a data use agreement. There is a formal process for external users to request access to LITE data, which involves review and approval by Principal Investigators from each study site as well as the Community Advisory Board; further details and forms can be obtained by emailing Dr. Andrea Wirtz (awirtz1@jhu.edu).