Summary
Background
In early 2025, the United States (U.S.) federal government issued several executive orders restricting healthcare access and non-discrimination protections for transgender and nonbinary (trans) populations. The current study investigated correlates of HIV prevention service disruptions among trans adults across five U.S. states.
Methods
We analysed cross-sectional survey data of trans adults (Feb–April 2025) from the Priority Assessment in Trans Health Across States (PATHS2). This multi-state survey was conducted in five varying policy environments (Connecticut, Georgia, New York, Utah, and Washington). Analyses were restricted to participants currently using HIV prevention services, including PrEP (oral or long-acting) or HIV testing in the prior 12 months (N = 693). Main outcomes were (a) any HIV prevention service disruption (delayed or denied HIV prevention medications (PrEP: oral or long-acting), lost insurance or financial assistance coverage, reduced service availability, increased healthcare stigma) and (b) disruption breadth, modelled as ordinal (none/some [1–2 types]/significant [≥3 types]). Multivariable and ordinal logistic regressions assessed correlates across sociodemographic, psychosocial, and policy-related factors.
Findings
Nearly two-thirds (64·5%, n = 447/693) of participants reported having experienced HIV prevention disruption; among this group. Among those reporting any disruption (n = 447), the most common disruption types included delayed/denied medications (53·2%, n = 238), reduced HIV service availability (50·3%, n = 225), lost coverage (40·9%, n = 183), and increased healthcare stigma (29·3%, n = 131). Compared with trans men, trans women had significantly higher odds of any disruption (aOR = 2·69, 95% CI 1·50–4·81) and greater disruption breadth (aOR = 1·85, 95% CI 1·16–2·93). Nonbinary individuals had elevated odds across all disruption types (aORs range = 2·99–6·58). Past-year discrimination was the strongest correlate of any disruption (aOR = 16·5, 95% CI 9·08–30·07) and breadth (aOR = 9·68, 95% CI 6·26–14·99). Black/African American participants had higher odds of any disruption compared with White participants (aOR = 2·17, 95% CI 1·06–4·46). State policy environment was not significantly associated with disruption in adjusted models.
Interpretation
Past-year discrimination was the dominant correlate of disruption. State-level protective policies did not independently buffer against disruption in this short post-issuance window, suggesting that a federal policy shock of this magnitude may have exceeded the buffering capacity of state-level protections. Targeted, actionable responses, such as sustained funding for Black-trans-led community health organisations as trusted PrEP access points, anti-discrimination enforcement in healthcare, navigation services to maintain continuity of PrEP and prescription assistance during coverage instability, and insurance marketplace protections (e.g., Affordable Care Act) and programs (e.g., Medicaid, Ryan White) that explicitly prohibit gender-identity-based denials of HIV prevention coverage, are urgently needed.
Funding
The Royalty Research Fund at the University of Washington.
Keywords: Transgender, HIV prevention, PrEP, Executive orders, Policy, Health equity, Service disruption
Research in context.
Evidence before this study
We searched PubMed and relevant grey literature for studies published prior to April 2025 (search start date unrestricted) examining HIV prevention service disruption among transgender populations in the context of US federal or state policy changes, using terms including “transgender,” “PREP,” “HIV prevention,” “executive order,” “policy,” and “healthcare disruption.” The search was not restricted by language. We identified a literature documenting the effects of anti-trans legislation on mental health, substance use, and healthcare avoidance, and emerging reports of grant terminations and programme closures following the 2025 executive orders (EOs). No peer-reviewed study had empirically characterised the prevalence, type, or correlates of HIV prevention service disruption among trans adults in the immediate aftermath of the 2025 federal EOs.
Added value of this study
This study provides an early empirical prevalence estimate of HIV prevention service disruption among trans adults in the weeks following the 2025 federal EOs, a period in which routine surveillance systems could not keep pace with the rate of policy change. Its contribution is descriptive and hypothesis generating: it identifies which subgroups report the highest burden of disruption and which factors co-occur most strongly with it, and it establishes a baseline against future studies that can further establish legislative impacts over time. It is not designed to establish that the EOs caused the disruptions observed, only associative. This study found that nearly two-thirds of participants experienced at least one disruption type, with discrimination as the dominant correlate. Disparities by gender identity, race/ethnicity, and mental health status are characterised, providing preliminary evidence for future targeted intervention and policy response.
Implications of all the available evidence
These findings, alongside existing evidence on structural determinants of HIV prevention engagement, document widespread HIV prevention service disruption among trans adults co-occurring with the 2025 federal EOs; the current study’s evidence with a cross-sectional design does not establish that the EOs caused these disruptions. State-level protective policies were not associated with lower disruption, although the five states sampled do not span the full policy spectrum. These estimates provide a baseline against subsequent surveillance that are longitudinal and more geographically expansive to assess change over time. Our findings point to ensuring that anti-discrimination enforcement, inclusive insurance coverage, and targeted support for trans individuals, particularly Black trans communities, warrant attention.
Introduction
In early 2025, the US federal government issued executive orders (EOs) redefining sex as binary and biological, removing federal recognition of gender identity from records and data collection, rescinding non-discrimination protections in healthcare, eliminating gender-affirming care from federal programs, and restricting inclusive language in federally funded research.1, 2, 3 Taken together, these actions represent one of the most sweeping rollbacks of federal civil rights protections for transgender and nonbinary (trans) people in US history; yet the downstream consequences for healthcare access remain poorly characterised,4 largely because the speed of implementation has outpaced the capacity of existing surveillance systems.5,6 This gap is especially consequential within HIV prevention continuum services,7,8 where even modest disruptions in medication access, insurance coverage, or provider availability can erode hard-won gains in pre-exposure prophylaxes (PrEP) uptake and HIV testing.5, 6, 7
Trans populations face disproportionate HIV burden. In the US, trans women, particularly Black and Latina trans women, have HIV prevalence estimates among the highest of any population subgroup,9 while trans men and nonbinary individuals represent populations with meaningful and under-addressed HIV risk.10,11 Trans people already encounter substantial structural barriers to HIV prevention services prior to 2025, including limited provider competency, insurance exclusions, and geographic inaccessibility.11 These barriers were not introduced by the 2025 EOs but, rather, layered onto long-standing pre-existing inequities in HIV prevention access; the EOs threatened to intensify these barriers by creating legal uncertainty around trans people's rights and healthcare as well as funding for HIV services, undermining provider willingness to treat trans patients,12,13 and destabilising HIV services coverage pathways, including PrEP and HIV testing.14
Prior research on anti-trans legislation, including bathroom bills and gender-affirming care bans, has demonstrated associations with adverse mental health outcomes, substance use, and reduced service utilisation,15, 16, 17, 18 and state-level protective policies have shown some buffering capacity.16,19 However, empirical evidence on HIV prevention service disruptions specifically in the context of rapid federal rollbacks remains absent. The present study addresses this gap, estimating the prevalence and correlates of HIV prevention service disruption among trans adults residing in varying policy environments following the 2025 EOs.
Methods
Study design and data source
We analysed cross-sectional online survey data from PATHS2 (Priority Assessment in Trans Health Across States), designed to examine correlates of HIV prevention service disruption among trans adults during the period concurrent with differing policy environments. Data were collected February–April 2025, concurrent with EO issuance and implementation.
Setting and participants
PATHS2 recruited participants across five US states representing a spectrum of policy environments: Connecticut, New York, and Washington (more protective) and Georgia and Utah (less protective), classified using an established index of transgender policy protections incorporating non-discrimination statutes, gender-affirming care access protections, and insurance mandate coverage.19 Participants were eligible if they were aged 18 or older; self-identified as transgender, nonbinary, gender non-conforming, or another trans identity. This study's analytic sample was limited to those reported no known HIV diagnosis; and reported current use of HIV prevention services, defined as PrEP (oral or long-acting) or HIV testing in the prior 12 months. Recruitment used targeted social media advertising and community-based outreach previously operationalised in formative PATHS studies.15,17 Given the use of online recruitment, we implemented a multi-layered fraud prevention protocol including CAPTCHA gating at survey entry, IP and geolocation cross-checks, duplicate-device detection, screening for inconsistent or implausible response patterns, screener-to-survey demographic consistency checks, and manual review of flagged cases by two independent team members.
Investigative team and advisory board
PATHS2 was conducted in partnership with a community advisory board comprising trans individuals representing diverse demographics, geographic contexts, and expertise (including policy advocates, community health workers, clinicians, and researchers). The board integrated lived experience across all phases of study design, measure development, and interpretation, including developing survey items assessing the relationships between policy and HIV prevention service disruption. The team met with board members monthly throughout data collection.
Survey administration and measures
The survey was programmed and administered via Qualtrics, with an average time of 20 min. All participants provided informed electronic consent prior to participation and received $25 USD compensation upon survey completion.
Outcome variables
Two primary outcomes were examined. The first was any HIV prevention service disruption (binary), derived from four items assessing whether, in the prior 12 months, participants experienced: (1) delay or denial of HIV prevention medications (PrEP, in oral or long-acting formulations); (2) loss of coverage (e.g., loss of insurance coverage, prescription assistance, or financial assistance programs covering HIV prevention services); (3) reduced service availability; or (4) increased healthcare stigma or discrimination. These domains were identified collaboratively with the advisory board as the most proximate pathways through which the 2025 orders were anticipated to affect HIV prevention access, consistent with community-based participatory research principles20 and prior recommendations for measure development in marginalised communities.21, 22, 23 Participants endorsing any item were classified as experiencing any disruption (0 = no, 1 = yes). The second outcome was disruption breadth, which we frame as a measure of how many disruption types a participant experienced rather than per-event impact, modelled as an ordinal variable: none, some (one to two types), or significant (three or more types or self-reported significant disruption breadth on care access).
Independent variables
State policy environment was a binary indicator distinguishing more protective (Connecticut, New York, Washington) from less protective (Georgia, Utah) states.19 Major experiences of discrimination was assessed using Williams Major Experiences of Discrimination scale, an 11-item measure of prior-year discriminatory experiences, which asks participants to report the number of past-year occurrences of major discriminatory events across employment (e.g., not being hired, not being promoted, being fired), education (e.g., being discouraged from higher education, being denied a scholarship), housing (e.g., being prevented from renting/buying in a desired neighbourhood), policing (e.g., being hassled by police), financial services (e.g., being denied a bank loan), service settings (e.g., being denied service by a store, restaurant, or service provider), and healthcare (e.g., being denied medical care). Total event counts were dichotomised into low (<9) versus high (≥9) discrimination events, with the ≥9 threshold based on within-sample median distribution.24 Perceived negative attribution of anti-trans policies was assessed using a six-item study-specific measure developed for the PATHS study, with our community advisory board, building on prior items from Restar et al. (2024).15 Psychometric properties were acceptable (Cronbach α = 0·74; KMO = 0·79; single-factor CFI = 0·98, RMSEA = 0·048, SRMR = 0·028). One item showed a weaker loading (0·20 vs 0·65–0·72 for the others) but was retained to preserve community-defined content coverage. Score range was 6–24; scores ≥18 (sample median) were classified as higher perceived anti-trans policy attribution. Instrumental support was assessed using a four-item short form adapted from the Medical Outcomes Study Social Support Survey (MOS-SSS),25 drawing one item each from the tangible, affectionate, positive social interaction, and emotional/informational subscales (e.g., availability of someone to help with daily chores when sick, to get together for relaxation, to understand your problems, and to love you and make you feel wanted). Total score range was 0–16; scores ≥8 (sample median) were classified as higher instrumental support. Anxiety was assessed using the GAD-2 subscale and depression using the PHQ-2 subscale of the four-item PHQ-4 instrument, with positive screens defined as subscale scores ≥3, consistent with validated cut-points.26 Sociodemographic covariates included age (continuous), race/ethnicity (Black/African American, Hispanic/Latino, other race/ethnicity [including Asian/Asian American, Middle Eastern/North African, Native Hawaiian/Pacific Islander, American Indian or Alaskan Native, and two or more races], and White [referent]), gender identity (trans women, nonbinary [each vs trans men as referent]), education (high school or less, some college, college graduate or higher), income (<$15,000, $15,000–$49,999, ≥$50,000), insurance status (insured, not insured), and having a primary care provider (yes/no) (full categorical definitions are provided in Table 1).
Table 1.
Sample Characteristics by HIV prevention service disruption (N = 693).
| Measures | Total |
No disruption |
Any disruption |
p-value | |||
|---|---|---|---|---|---|---|---|
| n | %a | n | %a | n | %a | ||
| 693 | 100·0 | 246 | 35·5 | 447 | 64·5 | ||
| Age | 0·013c | ||||||
| Mean (SD) | 29·7 (6·8) | 29·7 (6·4) | 30·5 (7·3) | ||||
| Race/ethnicity | 0·0006 | ||||||
| Black/African American | 154 | 22·2 | 39 | 15·9 | 115 | 25·7 | |
| Hispanic, Latino, Or Spanish Origin | 93 | 13·4 | 38 | 15·4 | 55 | 12·3 | |
| White | 274 | 39·5 | 87 | 35·4 | 187 | 41·8 | |
| Other Race/Ethnicitya | 172 | 24·8 | 82 | 33·3 | 90 | 20·1 | |
| Gender | 0·0003 | ||||||
| Trans woman | 207 | 29·9 | 53 | 21·5 | 154 | 34·5 | |
| Trans man | 207 | 29·9 | 92 | 37·4 | 115 | 25·7 | |
| Nonbinary | 279 | 40·3 | 101 | 41·1 | 178 | 39·8 | |
| Education (3 categories) | 0·07 | ||||||
| HS or less | 380 | 54·8 | 124 | 50·4 | 256 | 57·3 | |
| Some college | 108 | 15·6 | 36 | 14·6 | 72 | 16·1 | |
| College graduateb | 205 | 29·6 | 86 | 35·0 | 119 | 26·6 | |
| Income (3 categories) | 0·003 | ||||||
| <$15,000 | 133 | 19·2 | 55 | 22·4 | 78 | 17·4 | |
| $15,000–$49,999 | 277 | 40·0 | 112 | 45·5 | 165 | 36·9 | |
| $50,000b | 276 | 39·8 | 77 | 31·3 | 199 | 44·5 | |
| State policy environment | 0·35 | ||||||
| Less protective (GA, UT) | 302 | 43·6 | 113 | 45·9 | 189 | 42·3 | |
| More protective (CT, NY, WA) | 391 | 56·4 | 133 | 54·1 | 258 | 57·7 | |
| Insured | 0·67 | ||||||
| No | 88 | 12·7 | 33 | 13·4 | 55 | 12·3 | |
| Yes | 605 | 87·3 | 213 | 86·6 | 392 | 87·7 | |
| Has primary care provider | <0·0001 | ||||||
| No | 105 | 15·2 | 55 | 22·4 | 50 | 11·2 | |
| Yes | 588 | 84·8 | 191 | 77·6 | 397 | 88·8 | |
| Higher discrimination events (≥ 9) | <0·0001 | ||||||
| No | 393 | 58·2 | 202 | 82·1 | 191 | 42·7 | |
| Yes | 282 | 41·8 | 36 | 14·6 | 246 | 55·0 | |
| Higher perceived anti-trans policy impact (score ≥ 18) | <0·0001 | ||||||
| No | 194 | 28·0 | 76 | 30·9 | 118 | 26·4 | |
| Yes | 499 | 72·0 | 170 | 69·1 | 329 | 73·6 | |
| Higher instrumental support (score ≥ 8) | 0·24 | ||||||
| No | 267 | 38·5 | 102 | 41·5 | 165 | 36·9 | |
| Yes | 426 | 61·5 | 144 | 58·5 | 282 | 63·1 | |
| Positive anxiety screen (GAD-2 ≥ 3)b | <0·0001 | ||||||
| No | 352 | 50·8 | 160 | 65·0 | 192 | 43·0 | |
| Yes | 341 | 49·2 | 86 | 35·0 | 255 | 57·0 | |
| Positive depression screen (PHQ-2 ≥ 3)b | <0·0001 | ||||||
| No | 351 | 50·6 | 159 | 64·6 | 192 | 43·0 | |
| Yes | 342 | 49·4 | 87 | 35·4 | 255 | 57·0 | |
| Care disruption breadth (Model 2 outcome) | <0·0001 | ||||||
| No disruption | 246 | 35·5 | 246 | 100·0 | 0 | 0·0 | |
| Some disruption | 333 | 48·1 | 0 | 0·0 | 333 | 74·5 | |
| Significant disruption | 114 | 16·4 | 0 | 0·0 | 114 | 25·5 | |
Data are n (column %) unless otherwise indicated. Column percentages are calculated among valid (non-missing) responses within each group and therefore exclude missing data; counts and denominators for variables with missingness may not sum to the group totals. p values are from χ2 tests for categorical variables.
Anxiety was assessed using the GAD-2 subscale and depression using the PHQ-2 subscale of the four-item PHQ-4 instrument, with positive screens defined as subscale scores ≥3, consistent with validated cut-points.
Other race/ethnicity consists of: Asian/Asian American [n = 43, no disruption, n = 25 (10·2%), any disruption, n = 18 (4·0%)], Middle Eastern/North African [n = 43, no disruption, n = 16 (6·5%), any disruption, n = 27 (6·0%)], Native Hawaiian/Pacific Islander [n = 50, no disruption, n = 22 (8·9%), any disruption, n = 28 (6·3%)], American Indian or Alaskan Native [n = 25, no disruption, n = 13 (5·3%), any disruption, n = 12 (2·7%)], and Two or more race/ethnicity [n = 11, no disruption, n = 6 (2·4%), any disruption, n = 5 (1·1%)].
Statistical analysis
Analyses were restricted to participants reporting current HIV prevention service use (PrEP, HIV testing, or other HIV prevention-related care in the prior 12 months; N = 693). Sample characteristics were summarised using frequencies and percentages, with χ2 tests comparing characteristics by disruption status. We estimated two sets of multivariable models: binary logistic regression for any disruption (yes vs no), and ordinal logistic regression (ologit) for disruption breadth (none, some, significant). For the ordinal model, an adjusted odds ratio greater than 1 suggests higher odds of being in a more severe disruption-breadth category. The proportional-odds assumption was assessed using a generalised ordered logit model (gologit2); it was supported overall (Wald χ2 [12] = 14·83, p = 0·25) and for all covariates except the trans-woman indicator, and the proportional-odds specification was retained for the primary analysis given the limited departure and unchanged conclusions. Model goodness of fit was evaluated by comparing mean predicted category probabilities with observed category frequencies, which agreed within 4·4 percentage points across all three categories [no disruption (predicted 31·4% vs observed 35·8%), some disruption (45·9% vs 45·8%), and significant disruption (22·7% vs 18·4%), in the n = 528 complete-case sample]. Models were adjusted for age, race/ethnicity, gender identity, state policy environment, insurance status, perceived anti-trans policy impact, discrimination, instrumental support, and anxiety and depression screens. Adjusted models were estimated on participants with complete covariate data (n = 528); missingness was confined almost entirely to the perceived anti-trans policy impact measure. Additionally, four logistic models examined each specific disruption type. All models were adjusted simultaneously for the full covariate set. Results are presented as adjusted odds ratios (aOR) with 95% confidence intervals (CI). Analyses were conducted in Stata 19.27 Statistical significance was defined as p ≤ 0·05 (two-tailed).
Ethical considerations
PATHS2 was approved by the University of Washington IRB (Protocol #STUDY00021320; approved January 2025) and the University of Utah IRB (Protocol #00182648; approved December 2024). All participants provided electronic informed consent. No personally identifying information was collected. The study was conducted in accordance with the Declaration of Helsinki.
Role of the funding source
Our study funder had no role in the study design, data collection, analysis, data interpretation, manuscript writing, or the decision to submit the paper for publication.
Results
Sample characteristics
Full sample characteristics stratified by disruption status are in Table 1. The sample had a mean age of 29·7 years (SD = 6·8), with gender identity distributed across trans women (29·9%, n = 207), trans men (29·9%, n = 207), and nonbinary individuals (40·3%, n = 279). The sample was racially and ethnically diverse: White (39·5%, n = 274), Black/African American (22·2%, n = 154), Hispanic/Latino (13·4%, n = 93), and other race/ethnicity (24·8%, n = 172). Most attained high school or less (54·8%, n = 380). A majority (56·4%, n = 391) resided in more protective state policy environments. Bivariate comparisons revealed significant differences (p < 0·05) by disruption status for age, race/ethnicity, gender identity, income, having a primary care provider, discrimination events, perceived anti-trans policy, and anxiety and depression screens. State policy environment, insurance status, education, and instrumental support did not differ significantly.
Prevalence and types of HIV prevention service disruption
Among 693 participants, 447 (64·5%) experienced any HIV prevention service disruption and 246 (35·5%) did not. Among those with any disruption (n = 447), 74·5% experienced some and 25·5% experienced significant disruption. The most common disruption type was delayed or denied HIV-related medication (53·2%, n = 238), followed by reduced service availability (50·3%, n = 225), loss of insurance coverage, prescription assistance, or financial assistance programmes covering HIV prevention services (40·9%, n = 183), and increased healthcare stigma (29·3%, n = 131).
Disruption types varied significantly by gender identity (Table 2). Nonbinary individuals had higher prevalence of delayed or denied medication (62·4%) than trans women (49·4%) or trans men (44·3%) (p = 0·005); higher loss of coverage (62·9% vs 38·3% and 17·5%; p < 0·0001); higher reduced service availability (57·9% vs 44·3% and 46·1%; p = 0·034); and higher increased healthcare stigma (41·6% vs 19·1% and 22·7%; p < 0·0001).
Table 2.
HIV service disruption type overall and by gender (N = 447).
| Overall n | % | Trans women n | Trans women % | Trans men n | Trans men % | Nonbinary n |
Nonbinary % |
p-value | |
|---|---|---|---|---|---|---|---|---|---|
| Disruption: Delayed/denied HIV medication | 0·005 | ||||||||
| No | 209 | 46·8 | 78 | 50·6 | 64 | 55·7 | 67 | 37·6 | |
| Yes | 238 | 53·2 | 76 | 49·4 | 51 | 44·3 | 111 | 62·4 | |
| Disruption: Lost coveragea | <0·0001 | ||||||||
| No | 264 | 59·1 | 127 | 82·5 | 71 | 61·7 | 66 | 37·1 | |
| Yes | 183 | 40·9 | 27 | 17·5 | 44 | 38·3 | 112 | 62·9 | |
| Disruption: Reduced service availability | 0·034 | ||||||||
| No | 222 | 49·7 | 83 | 53·9 | 64 | 55·7 | 75 | 42·1 | |
| Yes | 225 | 50·3 | 71 | 46·1 | 51 | 44·3 | 103 | 57·9 | |
| Disruption: Increased stigma/discrimination | <0·0001 | ||||||||
| No | 316 | 70·7 | 119 | 77·3 | 93 | 80·9 | 104 | 58·4 | |
| Yes | 131 | 29·3 | 35 | 22·7 | 22 | 19·1 | 74 | 41·6 |
Loss of coverage or financial assistance loss of insurance coverage, prescription assistance, or financial assistance programs covering HIV prevention services.
Correlates of any HIV prevention service disruption
Results are in Table 3. Higher discrimination was the strongest correlate of any disruption (aOR = 16·5, 95% CI 9·08–30·07, p < 0·0001). Trans women had higher odds than trans men (aOR = 2·69, 95% CI 1·50–4·81, p < 0·0009). Black/African American participants had higher odds than White participants (aOR = 2·17, 95% CI 1·06–4·46, p = 0·033). Positive anxiety screen (aOR = 3·41, 95% CI 1·96–5·92, p < 0·0001), positive depression screen (aOR = 1·87, 95% CI 1·08–3·24, p = 0·025), and higher perceived anti-trans policy (aOR = 1·91, 95% CI 1·14–3·19, p = 0·014) were each independently associated with disruption. Older age was associated with lower odds (aOR = 0·92 per year, 95% CI 0·88–0·97, p < 0·0007). State policy environment, insurance status, and instrumental support were not significant.
Table 3.
Adjusted multivariable logistic regression results for correlates of any disruption.
| Adjusted odds ratio | 95% Confidence interval | p-value | ||
|---|---|---|---|---|
| Age | 0·92 | 0·88 | 0·97 | 0·0007 |
| Race/Ethnicity (White, ref) | ||||
| Black/African American | 2·17 | 1·06 | 4·46 | 0·033 |
| Hispanic, Latino, Or Spanish Origin | 1·22 | 0·47 | 2·17 | 0·69 |
| Other Race/Ethnicity | 0·69 | 0·34 | 1·40 | 0·31 |
| Gender (trans men, ref) | ||||
| Trans woman | 2·69 | 1·50 | 4·81 | 0·0009 |
| Nonbinary | 1·08 | 0·55 | 2·15 | 0·81 |
| State policy environment (less protective, ref) | 0·85 | 0·49 | 1·47 | 0·56 |
| Insurance status (not insured, ref) | 1·47 | 0·72 | 2·96 | 0·28 |
| Higher discrimination events (no, ref) | 16·5 | 9·08 | 30·07 | <0·0001 |
| Higher perceived anti-trans policy impact (no, ref) | 1·91 | 1·14 | 3·19 | 0·014 |
| Higher instrumental support (no, ref) | 0·85 | 0·51 | 1·41 | 0·53 |
| Positive anxiety screen (no, ref) | 3·41 | 1·96 | 5·92 | <0·0001 |
| Positive depression screen (no, ref) | 1·87 | 1·08 | 3·24 | 0·025 |
Bold = significant at p < 0.05.
Correlates of disruption breadth
Results are in Table 4. Higher discrimination was the strongest predictor of greater odds of disruption breadth (aOR = 9·68, 95% CI 6·26–14·99, p < 0·0001), that is, participants reporting higher discrimination were nearly ten times as likely to fall in a more severe category than a less severe one. Trans women had higher odds of disruption breadth than trans men (aOR = 1·85, 95% CI 1·16–2·93, p = 0·009); nonbinary individuals were elevated but not significant (aOR = 1·62, 95% CI 0·94–2·79, p = 0·08). Positive anxiety screen (aOR = 2·38, 95% CI 1·58–3·59, p < 0·0001), positive depression screen (aOR = 1·70, 95% CI 1·13–2·57, p = 0·012), and higher perceived policy (aOR = 1·58, 95% CI 1·05–2·38, p = 0·028) each predicted greater odds of disruption breadth. Older age was associated with lower odds of disruption breadth (aOR = 0·94 per year, 95% CI 0·91–0·98, p = 0·0009). State policy environment, insurance status, and instrumental support were not significant.
Table 4.
Adjusted ordinal logistic regression results for correlates of disruption breadth.
| Adjusted odds ratio | 95% Confidence interval | p-value | ||
|---|---|---|---|---|
| Age | 0·94 | 0·91 | 0·98 | 0·0009 |
| Race/Ethnicity (White, ref) | ||||
| Black/African American | 1·32 | 0·79 | 2·20 | 0·28 |
| Hispanic, Latino, Or Spanish Origin | 0·60 | 0·28 | 1·27 | 0·18 |
| Other Race/Ethnicity | 0·50 | 0·28 | 0·89 | 0·019 |
| Gender (trans men, ref) | ||||
| Trans woman | 1·85 | 1·16 | 2·93 | 0·009 |
| Nonbinary | 1·62 | 0·94 | 2·79 | 0·08 |
| State policy environment (less protective, ref) | 0·79 | 0·52 | 1·20 | 0·26 |
| Insurance status (not insured, ref) | 1·44 | 0·79 | 2·60 | 0·23 |
| Higher discrimination events (no, ref) | 9·68 | 6·26 | 14·99 | <0·0001 |
| Higher perceived anti-trans policy impact (no, ref) | 1·58 | 1·05 | 2·38 | 0·028 |
| Higher instrumental support (no, ref) | 0·81 | 0·55 | 1·20 | 0·29 |
| Positive anxiety screen (no, ref) | 2·38 | 1·58 | 3·59 | <0·0001 |
| Positive depression screen (no, ref) | 1·70 | 1·13 | 2·57 | 0·012 |
Adjusted odds ratios (aORs) from ordinal logistic regression for disruption breadth (none, some, significant); an aOR >1 indicates higher odds of a more severe category. The proportional-odds assumption was assessed using a generalised ordered logit model (gologit2); it was supported overall (Wald χ2[12] = 14·83, p = 0·25) and for all covariates except the trans-woman indicator, and the proportional-odds specification was retained given the limited departure and unchanged conclusions. Model fit was adequate, with mean predicted category probabilities agreeing with observed frequencies within 4·4 percentage points across all three categories (no disruption (predicted 31·4% vs observed 35·8%), some disruption (45·9% vs 45·8%), and significant disruption (22·7% vs 18·4%)). n = 528 with complete covariate data. Bold indicates p ≤ 0·05. CI = confidence interval.
Bold = significant at p < 0.05.
Correlates of specific disruption types
Results are in Table 5. Nonbinary individuals had consistently elevated odds across all four disruption types relative to trans men. They had higher odds of delayed or denied medication (aOR = 3·29, 95% CI 1·49–7·26, p = 0·003), lost coverage (aOR = 3·13, 95% CI 1·45–6·74, p = 0·004), reduced service availability (aOR = 2·99, 95% CI 1·43–6·29, p = 0·004), and more than six times the odds of increased healthcare stigma (aOR = 6·58, 95% CI 2·72–15·90, p < 0·0001). Trans women, by contrast, had lower odds of lost coverage than trans men (aOR = 0·24, 95% CI 0·12–0·49, p < 0·0001). Moreover, higher discrimination was associated with greater odds of delayed or denied medication (aOR = 3·06, 95% CI 1·69–5·54, p < 0·0002) but lower odds of reduced service availability (aOR = 0·52, 95% CI 0·30–0·91, p = 0·022). A positive anxiety screen was associated with higher odds of delayed or denied medication (aOR = 2·04, 95% CI 1·18–3·53, p = 0·011). Black/African American participants had lower odds of delayed or denied medication than White participants (aOR = 0·33, 95% CI 0·16–0·65, p = 0·002). Insured participants had higher odds of both lost coverage (aOR = 5·25, 95% CI 1·88–14·61, p = 0·002) and reduced service availability (aOR = 4·87, 95% CI 1·88–12·64, p = 0·001) than uninsured participants. For increased healthcare stigma, no covariates other than individuals with nonbinary identity were significant.
Table 5.
Adjusted logistic regression results for correlates of specific disruption types.
| Delayed/denied medication |
Lost coveragea |
Reduced service availability |
Increased stigma/discrimination |
|||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Adjusted odds ratio | 95% Confidence interval | p-value | Adjusted odds ratio | 95% Confidence interval | p-value | Adjusted odds ratio | 95% Confidence interval | p-value | Adjusted odds ratio | 95% Confidence interval | p-value | |||||
| Age (18–29, ref) | 1·04 | 0·98 | 1·09 | 0·19 | 0·97 | 0·92 | 1·02 | 0·26 | 0·94 | 0·90 | 0·99 | 0·033 | 0·99 | 0·94 | 1·04 | 0·82 |
| Race/Ethnicity (White, ref) | ||||||||||||||||
| Black/African American | 0·33 | 0·16 | 0·65 | 0·002 | 1·33 | 0·63 | 2·82 | 0·45 | 0·75 | 0·40 | 1·41 | 0·38 | 0·56 | 0·26 | 1·19 | 0·13 |
| Hispanic, Latino, Or Spanish Origin | 0·36 | 0·12 | 1·08 | 0·07 | 0·54 | 0·19 | 1·54 | 0·25 | 0·76 | 0·27 | 2·13 | 0·60 | 0·34 | 0·11 | 1·05 | 0·06 |
| Other Race/Ethnicitya | 0·64 | 0·28 | 1·50 | 0·31 | 1·02 | 0·45 | 2·33 | 0·96 | 0·85 | 0·39 | 1·84 | 0·68 | 0·73 | 0·31 | 1·73 | 0·48 |
| Gender (trans men, ref) | ||||||||||||||||
| Trans woman | 1·33 | 0·71 | 2·47 | 0·37 | 0·24 | 0·12 | 0·49 | <0·0001 | 1·62 | 0·89 | 2·95 | 0·11 | 1·40 | 0·67 | 2·94 | 0·37 |
| Nonbinary | 3·29 | 1·49 | 7·26 | 0·003 | 3·13 | 1·45 | 6·74 | 0·004 | 2·99 | 1·43 | 6·29 | 0·004 | 6·58 | 2·72 | 15·90 | <0·0001 |
| State policy environment (less protective, ref) | 0·86 | 0·48 | 1·54 | 0·62 | 0·80 | 0·43 | 1·48 | 0·48 | 0·91 | 0·53 | 1·56 | 0·74 | 0·84 | 0·45 | 1·56 | 0·59 |
| Insurance status (not insured, ref) | 0·85 | 0·31 | 2·31 | 0·75 | 5·25 | 1·88 | 14·61 | 0·002 | 4·87 | 1·88 | 12·64 | 0·001 | 2·62 | 0·86 | 7·93 | 0·09 |
| Higher discrimination events (no, ref) | 3·06 | 1·69 | 5·54 | 0·0002 | 0·92 | 0·51 | 1·66 | 0·77 | 0·52 | 0·30 | 0·91 | 0·022 | 1·25 | 0·65 | 2·41 | 0·50 |
| Higher perceived anti-trans policy impact (no, ref) | 0·91 | 0·48 | 1·71 | 0·76 | 0·88 | 0·47 | 2·13 | 0·70 | 1·57 | 0·88 | 2·82 | 0·13 | 1·66 | 0·82 | 3·34 | 0·16 |
| Higher instrumental support (no, ref) | 0·64 | 0·36 | 1·14 | 0·13 | 1·12 | 0·64 | 3·04 | 0·62 | 1·63 | 0·96 | 2·76 | 0·07 | 1·09 | 0·58 | 1·97 | 0·82 |
| Positive anxiety screen (no, ref) | 2·04 | 1·18 | 3·53 | 0·011 | 1·67 | 0·92 | 3·04 | 0·09 | 0·82 | 0·48 | 1·38 | 0·45 | 0·70 | 0·37 | 1·20 | 0·26 |
| Positive depression screen (no, ref) | 1·65 | 0·93 | 2·92 | 0·09 | 0·88 | 0·47 | 1·64 | 0·681 | 1·06 | 0·62 | 1·83 | 0·83 | 1·44 | 0·75 | 2·76 | 0·27 |
Bold = significant at p < 0.05.
Loss of coverage or financial assistance loss of insurance coverage, prescription assistance, or financial assistance programs covering HIV prevention services.
Discussion
To our knowledge, this study provides one of the first empirical descriptions of HIV prevention service disruption among trans adults in the immediate aftermath of the 2025 US federal EOs. It is critical to note that our design cannot answer the question of whether the 2025 executive orders had a downstream causal impact on HIV prevention service disruption; the data established in this study are associations observed at a single point in time. Drawing on data from five states collected within weeks of EO issuance, our findings document that disruptions were widespread, patterned by discrimination and perceived policy, and concentrated among trans women, nonbinary individuals, and Black trans people. These disruptions were reported by a population already bearing a disproportionate burden of HIV, during a period in which federal civil rights and healthcare protections for trans people were being rescinded. We position this analysis as a descriptive, hypothesis-generating account of that period rather than as an evaluation of executive-order effects.
First, disruptions were documented in the immediate aftermath of the 2025 EOs, with nearly two-thirds of trans adults who actively utilise HIV prevention services reporting at least one form of service disruption within the first three months, and one in four reporting significant disruption. These estimates are higher than healthcare access barriers observed in trans populations during prior policy periods.28 Moreover, since these reports were collected before most formal implementation guidance had even been issued,10,11 the EOs themselves (independent of any formal enforcement) may have generated sufficient uncertainty and fear to reshape how trans people engage with HIV prevention care. This is consistent with emerging accounts from healthcare providers documenting a chilling effect on the delivery of trans-affirming care over a similar timeframe.12,13 The present study corroborates and extends this evidence by capturing the client experience specifically within HIV prevention services. Critically, we found that disruptions spanned multiple areas of the HIV prevention continuum29 from medication access and insurance coverage to service availability and provider attitudes, indicating that reported disruptions spanned multiple domains of the continuum rather than being confined to one.
Second, past-year discrimination and perceived anti-trans policy were the dominant correlates of both disruption occurrence and breadth. Consistent with fundamental cause theory, discrimination plausibly drives HIV prevention service disruption through three concrete mechanisms, partly observable in our data and partly inferred from prior literature: (a) Direct denial or substandard care at the point of service (e.g., for instance, encounters with providers, refusal to prescribe PrEP, refusal to refer to HIV prevention services).30 (b) Anticipatory avoidance, in which prior discriminatory experiences lead patients to delay, ration, or forgo care that they otherwise need. Perceived anti-trans policy was also independently associated with both disruption occurrence and breadth, even after adjusting for enacted discrimination. This suggests awareness of anti-trans policy, even absent direct discriminatory encounters, appears to function as a distinct structural exposure shaping HIV prevention engagement. This is consistent with prior work documenting that anticipatory stress and perceived policy hostility shape adverse health and behavioral outcomes among trans populations,4,15,17 and extends this evidence to HIV prevention service access specifically.16,30 Finally, (c) institutional disruption, in which providers and facilities facing federal funding threats restrict, scale back, or close trans-affirming services entirely.12,13 The magnitude of discrimination association is large; while conceptual overlap between the discrimination measure and service-specific stigma items cannot be ruled out in cross-sectional data, the consistency across all model specifications (including the ordinal disruption breadth model), suggests a substantive underlying relationship. Taken together, these findings suggest that anti-discrimination enforcement both structurally and socially, not simply service availability, must be a central pillar of any policy or structural response.
Third, the absence of a significant association between state policy environment and either disruption occurrence or breadth warrants interpretation. Prior research has documented that state-level protective policies (e.g., non-discrimination statutes, gender-affirming care access protections, and insurance mandates) buffer against some adverse health and behavioural outcomes among trans populations.16 The null finding here may reflect the possibility that the 2025 federal EOs generated a policy shock of sufficient magnitude to overwhelm the buffering capacity of state-level protections, at least in the short term. When federal actions explicitly rescind non-discrimination protections, redefine sex in ways that exclude trans people from legal recognition, and create uncertainty about the legality of gender-affirming care, state-level frameworks may be insufficient to counteract the chilling effects on both providers and patients. If state-level frameworks alone cannot counteract federal chilling effects, the relevant policy layer is the one that sits between them, namely, the federally regulated insurance infrastructure itself. Medicaid, Affordable Care Act (ACA)-marketplace protections, as well as Ryan White infrastructure for HIV prevention access,31,32 that explicitly prohibit gender-identity-based denials of HIV prevention coverage, paired with navigation services that maintain continuity of PrEP and prescription assistance through periods of coverage instability, are likely better positioned to absorb a federal policy shock than state non-discrimination statutes operating downstream of federal coverage decisions. Longitudinal data and samples spanning more extreme ends of the state policy spectrum will be needed to fully characterise how state protections interact with federal rollbacks over time.
Fourth, trans women emerged as significantly more likely to experience any HIV prevention service disruption and greater disruption breadth compared with trans men. This finding is of particular relevance given that trans women, and especially trans women of colour, bear disproportionately high HIV burden and face the most substantial structural barriers to care in the United States.9 These results suggest that pre-existing structural vulnerabilities for trans women were further compounded in the context of the 2025 policy environment, at precisely the population most dependent on uninterrupted HIV prevention access. Notably, trans women had significantly lower odds of lost coverage (loss of insurance coverage, prescription assistance, or financial assistance programs covering HIV prevention services) compared with trans men in the disruption-type models, which may reflect differential pre-existing patterns of insurance instability or reliance on public coverage programs rather than employer-based insurance, which is a pattern documented in prior research on PrEP access barriers among trans women.29 Their elevated odds of disruption were also concentrated at the transition from no disruption to any disruption, rather than at the escalation from some to significant disruption, suggesting that trans women's heightened vulnerability operates primarily at the point of initial service disruption. This pattern reinforces the importance of sustaining Medicaid and ACA protections, and Ryan White infrastructure for HIV prevention access, since these are the coverage pathways on which trans women disproportionately rely.32
Fifth, the pattern of disparities for nonbinary individuals has received little attention in prior research. While nonbinary individuals did not differ significantly from trans men in overall disruption occurrence or breadth, they exhibited consistently elevated odds across every disruption type examined: delayed or denied medications, lost coverage, reduced service availability, and increased healthcare stigma or discrimination. In other words, nonbinary individuals and trans men were similarly likely to experience any disruption, but nonbinary individuals were significantly more likely to encounter each specific type of disruption, suggesting that their service disruptions are broader in scope. This may reflect the compounded invisibility of nonbinary identities in healthcare systems that have only recently begun incorporating trans-inclusive frameworks, with even less infrastructure to recognise and affirm nonbinary patients’ healthcare needs.33 Nonbinary individuals may also face greater administrative barriers in contexts where binary gender markers on insurance documents and medical records are required,33 making coverage loss particularly salient in a policy environment that explicitly reverts to binary sex definitions. These findings call for explicit attention to nonbinary individuals in HIV prevention program design and policy, rather than treating this sub-group as a residual category within transgender health research.
Additionally, race and mental health indicators as correlates of HIV service disruption warrant particular attention. Specifically, the racial disparity observed for Black/African American trans participants, who had more than twice the odds of any disruption compared with White participants, is consistent with the broader literature on intersecting structural racism and trans stigma as compounding barriers to HIV prevention care.34,35 Black trans people, particularly Black trans women, face some of the highest HIV prevalence of any group in the United States, and also encounter the most substantial barriers to care.34 These findings suggest that the 2025 policy environment may have further compounded these pre-existing inequities, though it is notable that this disparity was significant for disruption occurrence but not breadth, warranting further investigation. Concrete responses should include sustained funding for Black-trans-led community health organisations as trusted PrEP access points, culturally responsive provider training paired with mandated reporting of healthcare discrimination, and navigation services to support continuity of PrEP and prescription assistance during periods of coverage instability. The associations between mental health and disruption are bidirectional in nature such that anxiety and depression may reflect the psychosocial toll of existing in an increasingly hostile policy environment, which in turn erodes the capacity to navigate complex healthcare systems, advocate for oneself in clinical encounters, or persist through bureaucratic barriers to coverage.36 At the same time, disruption itself is likely to exacerbate mental health symptoms15 suggesting a potential feedback loop between policy stress, mental health, and care access that warrants longitudinal investigation.
Age, too, emerged as an independent correlate of disruption. Younger participants were more likely to experience disruption, and to face it across more domains. Several explanations are plausible. Young adulthood is a period of frequent insurance transitions; for instance, losing parental coverage at age 26 under the ACA, and facing gaps before stable employer-based coverage is in place, which leaves coverage more exposed when policies shift.37 Younger adults also tend to have fewer financial reserves to absorb out-of-pocket costs when coverage falters or formularies change. And younger trans adults are often still establishing their healthcare relationships and community ties, with less institutional history to fall back on when services are disrupted.38 These possibilities warrant dedicated attention in future waves of PATHS and similar cohorts.
Beyond these substantive associations, the pattern of covariate results across our three sets of models warrants explicit interpretation as some associations appear to differ between the aggregate and disruption-specific analyses. Tables 3 and 4 model aggregate disruption (any disruption, and disruption breadth), where each covariate's effect is summed across all four disruption types; Table 5 disaggregates the same outcome into four binary models, which substantially reduces statistical power per model and allows opposing effects across disruption types to be detected separately. A covariate can therefore show a strong association with overall disruption while showing mixed or even opposing associations with individual type (e.g., as seen for trans women having lower odds of lost coverage but higher overall disruption driven by other types), The negative association between trans women and lost coverage, in particular, is consistent with prior literature showing greater reliance among trans women on public coverage programs (Medicaid, Ryan White).32 The unexpected inverse “protective” association between higher discrimination and reduced service availability likely reflects selection: participants reporting extensive discrimination may have already disengaged from care, leaving little remaining service availability to be disrupted. Given these interpretive points and the limited power of the disaggregated models, we interpreted Tables 3 and 4 as our primary analyses and Table 5 as descriptive of where disparities concentrate.
And finally, two findings ran counter to expectation and warrant explicit comment. First, insured participants had higher odds than uninsured participants of both lost coverage (aOR = 5·25) and reduced service availability (aOR = 4·87). This is most plausibly structural rather than substantive, such that participants without insurance had no coverage to lose and were less likely to be engaged with services whose availability could contract. It may also reflect that insured participants are more embedded in formal care systems and therefore more exposed to administrative, formulary, and coverage changes. Second, Black/African American participants had lower odds of delayed or denied medication (aOR = 0·33) despite elevated odds of any disruption (aOR = 2·17). This divergence between the aggregate and type-specific estimates is informative rather than contradictory, and suggests that the disruption burden among Black participants was concentrated in domains other than medication access. One possibility is a differential service pathway, in that Ryan White–funded and community-based programmes with dedicated medication assistance may have greater reach among Black trans communities, buffering medication access specifically while other disruption domains remained elevated. An alternative is selection, in that those facing the most severe barriers may not have been engaged in medication access at all during this particular survey period. These interpretations warrant further investigation and corroboration with qualitative interviews and/or larger longitudinal samples and disaggregated data on service pathway.
Limitations
Several limitations should be noted. First, the cross-sectional design precludes causal inference; Plainly, the present design does not, and cannot, answer the question of whether the executive orders caused the observed disruptions; our study characterises co-occurring associations only. The 12-month recall window for the disruption items predates EO issuance, and we cannot disaggregate EO-attributable disruptions from pre-existing or unrelated structural barriers; the findings should therefore be interpreted as documenting the prevalence and correlates of HIV prevention service disruption at a specific moment in time when the EOs were the salient and most proximate policy event, rather than as a causal evaluation of EO-specific effects. Other concurrent factors (e.g., state-level legislative activity, broader political climate, and pre-existing erosion of healthcare access) cannot be disentangled from the EO-specific contribution in this design, and observed associations reflect ecological co-occurrence rather than individual-level causal exposure. Second, reverse causality is a substantive concern for our two most prominent correlates: experiences of HIV prevention service disruption may themselves elevate reports of past-year discrimination and of perceived anti-trans policy. The magnitude of the discrimination association (aOR 16·5) is consistent with bidirectional reinforcement rather than purely unidirectional causation, and longitudinal data with temporally separated exposure and outcome measurement are required to identify the direction of effect. Third, the sample was recruited via social media and community networks, limiting generalisability to populations with lower digital access. Fourth, the study was restricted to five states that do not represent the full spectrum of policy environments, and the binary protective/less-protective classification may not capture real-time policy variation; future research should examine disruption across a broader range of state contexts and develop more granular measures of structural policy exposure. Fifth, the discrimination cut-point of ≥9 was empirically derived from the within-sample distribution rather than externally validated, and the perceived anti-trans policy measure, while internally consistent and structurally sound in this sample, has not yet been externally validated outside the PATHS study; both should be evaluated in independent cohorts. Sixth, self-reported disruption data are subject to recall bias. Finally, the healthcare stigma outcome had a relatively small number of events, limiting statistical power for that model. More broadly, the relationships among these factors are likely interrelated and partly bidirectional: the federal policy environment plausibly shapes both perceived anti-trans policy impact and enacted discrimination, which in turn influence HIV prevention service disruption, even as the experience of disruption may itself heighten reported discrimination and perceived policy impact. State policy environment and sociodemographic vulnerability may further confound or modify these pathways. Disentangling this structure (particularly the direction of effect between discrimination, perceived policy impact, and service disruption) will require longitudinal data with temporally separated measurement, which our cross-sectional design cannot provide.
Conclusion
HIV prevention service disruption was highly prevalent among trans adults in the period concurrent with the 2025 US federal EOs. Past-year discrimination was the factor most strongly associated with both disruption occurrence and breadth. Trans women showed elevated odds of disruption in the aggregate models, whereas nonbinary individuals did not differ significantly from trans men in overall disruption occurrence or breadth but showed elevated odds across each specific disruption type examined. Black or African American participants had elevated odds of any disruption. State-level protective policies were not independently associated with lower disruption in this short post-issuance window. This null association has several possible explanations, including a federal policy shock large enough to outweigh state-level protections, and should be read alongside study limitations. Responses that warrant consideration include: sustained funding for Black-trans-led community health organisations as trusted PrEP access points; anti-discrimination enforcement in healthcare backed by accountability mechanisms; navigation services that maintain continuity of PrEP and prescription assistance through periods of coverage instability; and Medicaid and ACA-marketplace protections that explicitly prohibit gender-identity-based denials of HIV prevention coverage. Longitudinal surveillance of HIV prevention care engagement in trans populations will be critical to detecting further deterioration and guiding responsive intervention; a follow-up wave of PATHS is planned to support this need.
Contributors
AJR and TD conceptualised the study, led data interpretation, and drafted the manuscript. AO, SS, RAS, JB, AB, TRS, AM, AJL, TK, DO, KEG, and TD contributed to study design, data interpretation, and critical revision of the manuscript. All authors reviewed and approved the final version. AJR and TD directly accessed and verified the underlying data. AJR and TD were responsible for the decision to submit the manuscript for publication.
Data sharing statement
Data cannot be shared publicly per IRB agreement on data sharing and to minimise participant risk and maximise privacy and confidentiality as much as possible. Data are available from the University of Washington and University of Utah for researchers who meet the criteria for access to confidential data.
Declaration of interests
We declare no competing interests.
Acknowledgements
Along with our community advisory board members, some of whom are co-authors, we would like to thank Vanessa Granberry for her vision and providing insights for this study. Our gratitude also goes out to the participants of this study. The study was funded by The Royalty Research Fund at the University of Washington.
References
- 1.The White House Defending women from gender ideology extremism and restoring biological truth to the federal government. US president executive order 14168 of January 20, 2025. United States: Federal Register 90. 2025. https://www.federalregister.gov/documents/2025/01/30/2025-02090/defending-women-from-gender-ideology-extremism-and-restoring-biological-truth-to-the-federal Available at:
- 2.Nuruzzaman N., Hsiang E. Contextualizing President Trump's executive orders targeting transgender health and gender-affirming care. Am J Publ Health. 2026;116(2):256–260. doi: 10.2105/AJPH.2025.308265. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.The White House. Protecting Children from Chemical and Surgical Mutilation US President Executive Order 14187 of January 28, 2025. United States: Federal Register 90. 2025. https://www.whitehouse.gov/presidential-actions/2025/01/protecting-children-from-chemical-and-surgical-mutilation/ Available at:
- 4.Restar A.J., Layland E.K., Davis B., Thompson H., Streed C. The public health crisis state of transgender health care and policy. Am J Public Health. 2024;114:161–163. doi: 10.2105/AJPH.2023.307523. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Miller A.S., Liu M., Graham R., Berlin N.L., Carruthers K.H., Keuroghlian A.S. Termination of gender-affirming care grants at the National Institutes of Health. JAMA Pediatr. 2026;180(1):83–89. doi: 10.1001/jamapediatrics.2025.4653. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Greer S.L., Jarman H., Kulikoff R., Yaver M. Trump's second presidency begins: evaluating effects on the US health system. Lancet Reg Health, Am. 2025;48 doi: 10.1016/j.lana.2025.101173. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Stone J., Mutai K.K., Artenie A., et al. The impact of cuts in the US President's Emergency Plan for AIDS Relief funding for HIV pre-exposure prophylaxis in sub-Saharan Africa: a modelling study. Lancet HIV. 2025;12(10):e712–e721. doi: 10.1016/S2352-3018(25)00192-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Holtgrave D.R. Epidemiologic and economic impacts of halting national HIV prevention efforts. AIDS Behav. 2025;29(7):2033–2036. doi: 10.1007/s10461-025-04757-0. [DOI] [PubMed] [Google Scholar]
- 9.Becasen J.S., Denard C.L., Mullins M.M., Higa D.H., Sipe T.A. Estimating the prevalence of HIV and sexual behaviors among the US transgender population: a systematic review and meta-analysis, 2006–2017. Am J Publ Health. 2019;109(1):e1–e8. doi: 10.2105/AJPH.2018.304727. https://pubmed.ncbi.nlm.nih.gov/30496000/ Available at: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Reisner S.L., Poteat T., Keatley J., et al. Global health burden and needs of transgender populations: a review. Lancet. 2016;388(10042):412–436. doi: 10.1016/S0140-6736(16)00684-X. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Scheim A.I., Baker K.E., Restar A.J., Sell R.L. Health and health care among transgender adults in the United States. Annu Rev Publ Health. 2022;43(1):503–523. doi: 10.1146/annurev-publhealth-052620-100313. [DOI] [PubMed] [Google Scholar]
- 12.Hughes L.D., Hughto J.M., Wolfe H.L., et al. Provider adaptations to legislative efforts to ban gender-affirming care for adolescents. J Adolesc Health. 2025;78:504–512. doi: 10.1016/j.jadohealth.2025.11.005. [DOI] [PubMed] [Google Scholar]
- 13.Hughes L.D., Gamarel K.E., Restar A.J., et al. Adolescent providers' experiences of harassment related to delivering gender-affirming care. J Adolesc Health. 2023;73(4):672–678. doi: 10.1016/j.jadohealth.2023.06.024. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Dang M., Scheim A.I., Teti M., et al. Barriers and facilitators to HIV pre-exposure prophylaxis uptake, adherence, and persistence among transgender populations in the United States: a systematic review. AIDS Patient Care STDS. 2022;36(6):236–248. doi: 10.1089/apc.2021.0236. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Restar A., Layland E.K., Hughes L., et al. Antitrans policy environment and depression and anxiety symptoms in transgender and nonbinary adults. JAMA Netw Open. 2024;7(8) doi: 10.1001/jamanetworkopen.2024.31306. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Lee W.Y., Hobbs J.N., Hobaica S., DeChants J.P., Price M.N., Nath R. State-level anti-transgender laws increase past-year suicide attempts among transgender and non-binary young people in the USA. Nat Hum Behav. 2024;8(11):2096–2106. doi: 10.1038/s41562-024-01979-5. [DOI] [PubMed] [Google Scholar]
- 17.Baker A., DeAtley T., Ogunbajo A., et al. Impact of structurally transphobic policy environment on harmful alcohol and substance use among transgender and nonbinary adults. Sci Rep. 2025;15(1) doi: 10.1038/s41598-025-27266-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Goldenberg T., Reisner S.L., Harper G.W., Gamarel K.E., Stephenson R. State policies and healthcare use among transgender people in the US. Am J Prev Med. 2020;59(2):247–259. doi: 10.1016/j.amepre.2020.01.030. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Dechman D., Mercure M., Harmsworth E., et al. Movement Advancement Project. Equality maps: Transgender health care protections. Movement Advancement Project; 2025. 2023. https://mapresearch.org/equality/ Available at:
- 20.Restar A., Minalga B.J., Quilantang M.I., et al. Mapping community-engaged implementation strategies with transgender scientists, stakeholders, and trans-led community organizations. Curr HIV AIDS Rep. 2023;20(3):160–169. doi: 10.1007/s11904-023-00656-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Israel B., Schulz A., Parker E., et al. NIEHS; Washington, DC: 2000. Community-Based Participatory Research Principles. Rationale and Policy Recommendations, A Keynote Address at the Preliminary Meeting of Successful Models of Community–Based Participatory Research, Final Report; 2000; pp. 29–31.https://projecte3.com/download/Successful_Models_of_Community-Based_Participatory_Research_-_National_Institutes_of_Health.PDF Available at: [Google Scholar]
- 22.Restar A.J., Suen D.C., Morris M.K., et al. The necessity of including transgender adolescents in developing affirming gender identity research measures. Pediatr Res. 2025:1–4. doi: 10.1038/s41390-025-04664-z. https://pubmed.ncbi.nlm.nih.gov/41372627/ Available at: [DOI] [PubMed] [Google Scholar]
- 23.Restar A.J., Lett E., Menezes N.P., et al. Getting precise about gender and sex measurement: a primer for epidemiologists. Am J Epidemiol. 2024;193(12):1861–1867. doi: 10.1093/aje/kwae144. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Williams D.R. Measuring discrimination resource. Psychology. 1997;2(3):335–351. doi: 10.1177/135910539700200305. [DOI] [PubMed] [Google Scholar]
- 25.Sherbourne C.D., Stewart A.L. The MOS social support survey. Soc Sci Med. 1991;32(6):705–714. doi: 10.1016/0277-9536(91)90150-b. [DOI] [PubMed] [Google Scholar]
- 26.Löwe B., Wahl I., Rose M., et al. A 4-item measure of depression and anxiety: validation and standardization of the Patient Health Questionnaire-4 (PHQ-4) in the general population. J Affect Disord. 2010;122(1–2):86–95. doi: 10.1016/j.jad.2009.06.019. [DOI] [PubMed] [Google Scholar]
- 27.StataCorp . StataCorp LLC; College Station, TX: 2025. Stata Statistical Software: Release 19.https://www.stata.com/stata-news/news40-2/ Available at: [Google Scholar]
- 28.James S., Herman J., Rankin S., Keisling M., Mottet L., Ma A. The report of the 2015 US transgender survey. 2016. https://justiceresearch.dspacedirect.org/items/00c52701-7548-40e6-bb8e-f59f8d8476b2 Available at:
- 29.McNairy M.L., El-Sadr W.M. A paradigm shift: focus on the HIV prevention continuum. Clin Infect Dis. 2014;59(suppl_1):S12–S15. doi: 10.1093/cid/ciu251. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Hatzenbuehler M.L., Phelan J.C., Link B.G. Stigma as a fundamental cause of population health inequalities. Am J Publ Health. 2013;103(5):813–821. doi: 10.2105/AJPH.2012.301069. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Dawson L., Sobel L., Pestaina K., Kates J., Artiga S., Burns A. The Biden administration's final rule on section 1557 non-discrimination regulations under the ACA. KFF. 2024. https://www.kff.org/affordable-care-act/the-biden-administrations-final-rule-on-section-1557-non-discrimination-regulations-under-the-aca/ Available at:
- 32.Thomas J.A., Irvine M.K., Xia Q., Harriman G.A. Service utilization and HIV outcomes among transgender women receiving Ryan White Part A services in New York City. PLoS One. 2021;16(7) doi: 10.1371/journal.pone.0253444. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Bindman J., Ngo A., Zamudio-Haas S., Sevelius J. Health care experiences of patients with nonbinary gender identities. Transgender Health. 2022;7(5):423–429. doi: 10.1089/trgh.2021.0029. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Hughto J.M.W., Reisner S.L., Pachankis J.E. Transgender stigma and health: a critical review of stigma determinants, mechanisms, and interventions. Soc Sci Med. 2015;147:222–231. doi: 10.1016/j.socscimed.2015.11.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Lett E., Asabor E.N., Tran N., et al. Sexual behaviors associated with HIV transmission among transgender and gender diverse young adults: the intersectional role of racism and transphobia. AIDS Behav. 2022;26(11):3713–3725. doi: 10.1007/s10461-022-03701-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Puckett J.A., Matsuno E., Dyar C., Mustanski B., Newcomb M.E. Mental health and resilience in transgender individuals: what type of support makes a difference? J Fam Psychol. 2019;33(8):954. doi: 10.1037/fam0000561. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Stitt B., Finlayson T.L., Lipton B.J. Aging out of dependent coverage and health insurance trends, 2014–2019. Acad Pediatr. 2024;24(5):791–799. doi: 10.1016/j.acap.2023.09.021. [DOI] [PubMed] [Google Scholar]
- 38.Rodriguez A., Horvath K.J., Dowshen N., et al. Awareness and utilization of pre-exposure prophylaxis and HIV prevention services among transgender and non-binary adolescent and young adults. Front Reprod Health. 2024;5 doi: 10.3389/frph.2023.1150370. [DOI] [PMC free article] [PubMed] [Google Scholar]