Abstract
Background
Differences in heart failure (HF) phenotypes and trajectories among transgender and gender diverse (TGD) populations may provide insight into mechanisms of HF risk.
Objectives
The objective of the study was to evaluate associations between gender identity, incident HF, and trajectories between HF phenotypes.
Methods
We conducted a retrospective cohort study of 1,103,923 veterans in the Veterans Health Administration (October 1, 2010–September 30, 2019; median follow-up 9.4 years). TGD status (n = 42,157; 3.8%) was identified using natural language processing. Gender identity was categorized as cisgender men, cisgender women, transmasculine, transfeminine, or unclassified TGD. Incident HF was defined using administrative codes and classified by ejection fraction (EF) as HF with reduced EF (<40%), HF with preserved EF (HFpEF) (>50%), HFpEF with EF decline, and HF with improved EF. Multivariable Cox regression estimated associations.
Results
The mean age of TGD veterans was 46 years. 107,766 incident HF events occurred, including 3,078 among TGD veterans. Incident HF rates ranged from 6.22 to 8.89 per 1,000 person-years across TGD identities vs 3.87 in cisgender women and 13.05 in cisgender men. TGD status was associated with higher HF risk than cisgender women (HR: 1.53; 95% CI: 1.46 to 1.60), but lower than cisgender men (HR: 0.90; 95% CI: 0.86-0.93). HF with reduced EF rates were intermediate between cisgender men and women; associations with HFpEF were inconsistent after adjustment. Rates of HFpEF with EF decline and HF with improved EF varied (130.0-163.3 and 161.3-262.6 per 1,000 person-years, respectively).
Conclusions
Gender identity is associated with differences in HF incidence and trajectory among veterans. HF epidemiology may be incompletely characterized without inclusion of TGD populations.
Key words: cardiovascular epidemiology, electronic health records, health disparities, heart failure, transgender and gender diverse populations
Central Illustration
Heart failure (HF) is a leading cause of morbidity and mortality in the United States and exhibits well-established sex-specific differences in incidence and phenotype.1, 2, 3 However, the incidence, phenotype distribution, and progression of HF across gender diverse populations remain poorly characterized. An estimated 1.3 million adults in the United States identify as transgender and gender diverse (TGD),4 and the proportion of TGD adults is expected to rise.5 The cardiovascular health of TGD adults has historically been understudied, in part because standardized sexual orientation and gender identity (SOGI) data have not been routinely collected in cardiovascular research or clinical data sets.6, 7, 8, 9 Emerging evidence suggests that cardiovascular risk and outcomes may differ across gender identities; TGD people experience higher risk of atherosclerotic cardiovascular disease (CVD) and HF, particularly heart failure with preserved ejection fraction (HFpEF).10, 11, 12
HF is a common clinical endpoint of multiple forms of CVD and encompasses distinct phenotypes, including heart failure with reduced ejection fraction (HFrEF) and HFpEF, which differ in pathophysiology, clinical characteristics, and treatment approaches. In addition to these phenotypes, some patients experience transitions between HF states, including progression from HFpEF to HFrEF (HFpEF with ejection fraction [EF] decline) and recovery of systolic function among patients with prior HFrEF (HF with improved EF, heart failure with improved ejection fraction [HFimpEF]). Sex hormones influence myocardial remodeling, vascular function, and neurohormonal signaling, and contribute to well-described sex differences in HF phenotypes.13, 14, 15, 16 Studying HF across TGD populations exposed to varying hormonal environments may provide insights into mechanisms underlying HF risk and phenotype.
Because many longitudinal cardiovascular cohorts have not collected standardized SOGI data, electronic health record (EHR)-based data sets have become an important resource for studying cardiovascular outcomes in TGD populations.7 The Veterans Health Administration (VHA), the largest integrated health care system and one of the largest providers of health care for TGD people in the United States, provides a unique opportunity to evaluate CVD patterns among TGD veterans at a national scale.10,11,17, 18, 19
In this study, we used a national cohort of veterans receiving care in the VHA to compare incident HF phenotypes, including overall HF, HFrEF, and HFpEF, between TGD and cisgender adults, and to examine trajectories across HF phenotypes (HFpEF with EF decline and HFimpEF) among TGD veterans.
Methods
Data source
This analysis used EHR data from veterans receiving care through the VHA. We included veterans with at least 1 primary care encounter from October 1, 2010, through September 30, 2019, to align with the repeal of Section 654 of Title 10, U.S. Code (10 U.S.C. § 654, commonly referred to as “Don’t Ask, Don’t Tell”) which previously prohibited out sexual and gender minority (SGM) (eg, lesbian, gay, bisexual, transgender, and queer) people from serving in the U.S. military. Data were obtained from the Veteran Affairs (VA) Corporate Data Warehouse clinical and administrative data files. The VA Connecticut Healthcare System and VA Washington DC Institutional Review Boards approved this study.
This EHR-based cohort contains data from 1,144,148 veterans who were classified SGM or non-SGM by a natural language processing (NLP) algorithm, as previously described and elaborated on in the Supplemental Methods.10,11 A glossary of terminology is included in Supplemental Table 1. From this starting sample, we excluded 40,225 individuals with prevalent HF (defined as at least 1 inpatient or outpatient code for HF in the year before their baseline date) (Supplemental Table 2) for a final analytic sample of 1,103,923.
Independent variable
In primary analyses, gender modality (TGD or cisgender)20 and identity (cisgender man, cisgender woman, transmasculine, transfeminine, or unclassified TGD) were ascertained using NLP in combination with a validated algorithm based on International Classification of Disease (ICD) codes and past-year receipt of gender-affirming hormone therapy (GAHT) as previously described.10,21, 22, 23, 24, 25
When compared with self-reported data on gender identity among 148,409 veterans, this method of sequentially classifying both veterans’ gender modality and identity performed well as a group; the gender identities of TGD veterans were identified with 99.0% sensitivity and 96.4% specificity, and 31.1% positive predictive value (PPV). The low PPV was largely attributable to individuals who were categorized as TGD unclassified but self-report other gender identities; the reason for low algorithm performance in this instance was primarily related to receipt of both feminizing and masculinizing GAHT, presence of nonspecific gender dysphoria ICD codes, or presence of general keywords captured by the NLP without a corresponding gender (eg, “transgender” alone vs “transgender man”).
In contrast, this method of gender identity classification performed well in cases where specific directional classification could be made. Transfeminine veterans were identified with 98.8% sensitivity, 99.7% specificity, and 76.2% PPV and transmasculine veterans were identified with 99.0% sensitivity, 99.9% specificity, and 87.9% PPV. Additional details describing the algorithm are provided in the Supplemental Methods; NLP keywords are displayed in Supplemental Table 3 whereas GAHTs used to classify gender identity are provided in Supplemental Table 4.
Because veterans’ gender modality and identity were algorithmically classified rather than self-reported, all references to gender modality and identity herein are presumed unless otherwise specified. In sensitivity analyses, we limited the sample to veterans who self-identified their gender, and this self-identified variable served as the exposure of interest.
Dependent variable
The primary dependent variable was incident HF occurring between baseline and the end of follow-up (death or September 30, 2019). The baseline date was defined as 1 year after the index date, where the index date was the first primary care encounter from October 1, 2009, through September 30, 2019. Incident HF was identified by at least 1 inpatient or at least two outpatient ICD-9/10 or Current Procedural Terminology codes (Supplemental Table 2).
EF was obtained from the echocardiogram report closest to the HF diagnosis within a 180-day window and was used to classify HF as HFrEF (EF <40%) or HFpEF (EF >50%). Secondary outcomes included incident HFrEF, HFpEF, HFpEF with EF decline to EF <40%, and HFimpEF (HFrEF with EF improvement to EF >50%). For analyses of HF phenotype trajectories (HFpEF with EF decline and HFimpEF), baseline was the date of incident HF, and follow-up time was calculated as the interval between incident HF and the date of a repeat echocardiogram demonstrating transition between HF phenotypes.
Additional variables
Models included age, race, Hispanic ethnicity, sexual minority identity (ie lesbian, gay, or bisexual), body mass index (BMI), diabetes, hypertension, alcohol use disorder, anxiety, depression, HIV status, lipids, race-free estimated glomerular filtration rate smoking status, , homelessness, area deprivation index (ADI), Veterans Affairs enrollment priority group (based on income, medical conditions, and other factors), and a state-level tally of anti-TGD policies in veterans’ state of residence (higher values indicate more legal protections for TGD individuals) as covariates.26, 27, 28, 29, 30, 31, 32, 33 Values of these variables closest to the baseline date during the 1-year baseline period were used. Distribution of military sexual trauma and prior suicide attempts across gender identity were also examined. Additional details regarding data source and variable definitions are provided in the Supplemental Methods and relevant ICD codes are displayed in Supplemental Table 5.
Missing data
The frequency of missing data is shown in the footnote of Table 1. Multiple imputation by chained equations generated 10 complete data sets for missing covariates. Predictive mean matching was used to preserve biologically plausible values.34 Results were combined using Rubin rules.35
Table 1.
Baseline Characteristics by Gender Identity
| Cisgender Womena (n = 144,300) | Cisgender Menb (n = 917,466) | Transfemininec (n = 7,991) | Transmasculined (n = 7,129) | Unclassified TGDe (n = 27,037) | P Value | |
|---|---|---|---|---|---|---|
| Age, y | 40.4 (13.8) | 52.8 (16.9) | 43.8 (14.3) | 47.2 (14.8) | 46.7 (14.6) | <0.001 |
| Race | <0.001 | |||||
| Unknown | 11,508 (8.0) | 69,551 (7.6) | 393 (4.9) | 382 (5.4) | 1,400 (5.2) | |
| American Indian | 1,504 (1.0) | 6,350 (0.7) | 115 (1.4) | 74 (1.0) | 351 (1.3) | |
| Asian | 2013 (1.4) | 8,982 (1.0) | 80 (1.0) | 92 (1.3) | 320 (1.2) | |
| Black | 41,754 (28.9) | 174,541 (19.0) | 719 (9.0) | 988 (13.9) | 6,447 (23.8) | |
| Mixed race | 2,542 (1.8) | 8,984 (1.0) | 134 (1.7) | 112 (1.6) | 381 (1.4) | |
| Pacific Islander | 1,682 (1.2) | 8,441 (0.9) | 52 (0.7) | 76 (1.1) | 244 (0.9) | |
| White | 83,297 (57.7) | 640,617 (69.8) | 6,498 (81.3) | 5,405 (75.8) | 17,894 (66.2) | |
| Ethnicity | <0.001 | |||||
| Non-Hispanic | 128,074 (88.8) | 830,135 (90.5) | 7,524 (94.2) | 6,566 (92.1) | 24,866 (92.0) | |
| Hispanic | 12,362 (8.6) | 62,594 (6.8) | 442 (5.5) | 535 (7.5) | 1906 (7.0) | |
| Unknown | 3,864 (2.7) | 24,737 (2.7) | 25 (0.3) | 28 (0.4) | 265 (1.0) | |
| Marital status | <0.001 | |||||
| Married | 46,558 (33.4) | 433,361 (48.5) | 2,332 (29.4) | 3,444 (48.9) | 8,897 (33.4) | |
| Divorced/separated | 48,647 (34.9) | 253,653 (28.4) | 3,237 (40.8) | 1964 (27.9) | 9,780 (36.7) | |
| Never married/single | 39,421 (28.3) | 163,734 (18.3) | 2,139 (27.0) | 1,518 (21.5) | 7,074 (26.5) | |
| Widowed | 4,628 (3.3) | 42,503 (4.8) | 221 (2.8) | 122 (1.7) | 917 (3.4) | |
| Sexual minority identity | 41,625 (28.8) | 123,021 (13.4) | 2,317 (29.0) | 1,163 (16.3) | 5,808 (21.5) | <0.001 |
| VA enrollment priority | 3.4 (2.4) | 3.9 (2.4) | 3.8 (2.2) | 3.6 (2.4) | 3.9 (2.2) | <0.001 |
| BMI | 29.3 (6.4) | 29.7 (5.8) | 29.3 (6.3) | 31.4 (6.1) | 29.2 (5.9) | <0.001 |
| Hypertension | <0.001 | |||||
| None | 83,091 (66.7) | 364,966 (44.3) | 3,672 (49.1) | 3,329 (50.0) | 13,549 (54.9) | |
| Controlled | 26,067 (20.9) | 298,091 (36.2) | 2,668 (35.7) | 2,127 (32.0) | 6,624 (26.8) | |
| Uncontrolled | 15,356 (12.3) | 159,945 (19.4) | 1,142 (15.3) | 1,197 (18.0) | 4,513 (18.3) | |
| Smoker status | <0.001 | |||||
| Never | 69,399 (50.4) | 287,026 (32.5) | 3,023 (38.0) | 2,506 (35.3) | 7,735 (29.0) | |
| Former | 23,677 (17.2) | 238,736 (27.0) | 1977 (24.8) | 2,236 (31.5) | 5,793 (21.7) | |
| Current | 44,695 (32.4) | 358,024 (40.5) | 2,957 (37.2) | 2,354 (33.2) | 13,110 (49.2) | |
| Total cholesterol, mg/dL | <0.001 | |||||
| <200 | 57,296 (62.7) | 459,987 (69.8) | 3,798 (64.7) | 3,745 (67.7) | 13,013 (67.0) | |
| 200-239 | 23,507 (25.7) | 140,919 (21.4) | 1,477 (25.2) | 1,244 (22.5) | 4,509 (23.2) | |
| ≥240 | 10,640 (11.6) | 58,409 (8.9) | 596 (10.2) | 542 (9.8) | 1886 (9.7) | |
| HDL cholesterol, mg/dL | <0.001 | |||||
| <35 | 6,579 (7.2) | 164,049 (24.9) | 967 (16.5) | 1,428 (25.8) | 3,873 (20.0) | |
| 35-54 | 44,555 (48.7) | 378,082 (57.3) | 3,217 (54.8) | 3,214 (58.1) | 10,927 (56.3) | |
| ≥55 | 40,309 (44.1) | 117,184 (17.8) | 1,687 (28.7) | 889 (16.1) | 4,608 (23.7) | |
| LDL cholesterol, mg/dL | <0.001 | |||||
| <130 | 64,991 (71.1) | 490,553 (74.4) | 4,190 (71.4) | 4,031 (72.9) | 14,099 (72.6) | |
| 130-159 | 17,482 (19.1) | 112,947 (17.1) | 1,161 (19.8) | 995 (18.0) | 3,540 (18.2) | |
| ≥160 | 8,970 (9.8) | 55,815 (8.5) | 520 (8.9) | 505 (9.1) | 1769 (9.1) | |
| Statin use | 24,160 (16.7) | 333,610 (36.4) | 1791 (22.4) | 2,492 (35.0) | 6,786 (25.1) | <0.001 |
| Diabetes | <0.001 | |||||
| Controlled | 4,405 (3.1) | 56,403 (6.2) | 317 (4.0) | 465 (6.6) | 1,159 (4.3) | |
| Uncontrolled | 6,271 (4.4) | 105,299 (11.6) | 506 (6.4) | 728 (10.3) | 2,182 (8.1) | |
| None | 132,945 (92.6) | 748,324 (82.2) | 7,125 (89.6) | 5,896 (83.2) | 23,547 (87.6) | |
| eGFR, mL/min/1.73 m2 | 87.0 (20.6) | 81.5 (20.1) | 84.4 (20.3) | 84.4 (19.0) | 85.1 (19.6) | <0.001 |
| Chronic kidney disease | <0.001 | |||||
| Stage 1: eGFR > 90 | 47,893 (44.7) | 265,527 (36.2) | 2,685 (40.1) | 2,387 (39.6) | 9,248 (42.2) | |
| Stage 2: 60 < eGFR <9 | 52,486 (49.0) | 381,725 (52.0) | 3,433 (51.2) | 3,214 (53.3) | 11,096 (50.6) | |
| Stage 3a: 45 < eGFR <59 | 5,406 (5.0) | 61,218 (8.3) | 481 (7.2) | 362 (6.0) | 1,234 (5.6) | |
| Stage 3b: 30 < eGFR <44 | 1,083 (1.0) | 18,816 (2.6) | 82 (1.2) | 49 (0.8) | 263 (1.2) | |
| Stage 4: 15 < eGFR <29 | 260 (0.2) | 4,674 (0.6) | 12 (0.2) | 14 (0.2) | 70 (0.3) | |
| Stage 5: eGFR <15 | 86 (0.1) | 2069 (0.3) | 11 (0.2) | 7 (0.1) | 29 (0.1) | |
| Alcohol use disorder | 8,296 (5.7) | 100,002 (10.9) | 680 (8.5) | 514 (7.2) | 4,499 (16.6) | <0.001 |
| Homelessness | 12,644 (8.8) | 89,200 (9.7) | 1,124 (14.1) | 620 (8.7) | 5,243 (19.4) | <0.001 |
| Suicide attempt | <0.001 | |||||
| Not reported | 137,849 (95.5) | 882,572 (96.2) | 7,318 (91.6) | 6,779 (95.1) | 24,580 (90.9) | |
| >1 y after enrollment | 3,691 (2.6) | 18,395 (2.0) | 433 (5.4) | 217 (3.0) | 1,545 (5.7) | |
| Present at enrollment | 2,760 (1.9) | 16,499 (1.8) | 240 (3.0) | 133 (1.9) | 912 (3.4) | |
| Military sexual trauma | <0.001 | |||||
| Never screened | 16,218 (11.2) | 83,392 (9.1) | 577 (7.2) | 550 (7.7) | 2,128 (7.9) | |
| No | 74,340 (51.5) | 791,897 (86.3) | 5,372 (67.2) | 5,730 (80.4) | 20,398 (75.4) | |
| Yes | 53,742 (37.2) | 42,177 (4.6) | 2042 (25.6) | 849 (11.9) | 4,511 (16.7) | |
| Depression | 24,754 (17.2) | 84,027 (9.2) | 1,534 (19.2) | 1,026 (14.4) | 4,426 (16.4) | <0.001 |
| General anxiety | 4,942 (3.4) | 18,088 (2.0) | 315 (3.9) | 226 (3.2) | 914 (3.4) | <0.001 |
| HIV | 304 (0.2) | 15,985 (1.7) | 104 (1.3) | 138 (1.9) | 659 (2.4) | <0.001 |
| Area deprivation index, % | 54.4 (25.3) | 54.9 (26.6) | 53.0 (25.7) | 51.2 (26.1) | 54.7 (26.9) | <0.001 |
| State anti-TGD policies | 6.0 (11.4) | 6.3 (11.5) | 8.2 (11.7) | 7.6 (11.8) | 7.2 (11.6) | <0.001 |
Summary statistics for continuous variables are presented as mean and SD with P values from ANOVA; summary statistics for categorical variables are presented as number and column percentage with P values from chi-square tests.
BMI = body mass index; eGFR = estimated glomerular filtration rate; HDL = high-density lipoprotein; LDL = low-density lipoprotein; TGD = transgender and gender diverse; VA = Veteran Affairs.
Number missing: 5,046 (marital status), 1,123 (priority score), 20,585 (BMI), 19,786 (hypertension), 6,529 (smoking status), 52,857 (lipids), 679 (diabetes), 37,086 (eGFR), 19,783 (pain), 28,109 (ADI National Rank), 26,369 (state anti-TGD policies).
Number missing: 24,215 (marital status), 3,975 (priority score), 100,584 (BMI), 94,464 (hypertension), 33,680 (smoking status), 258,151 (lipids), 7,440 (diabetes), 183,437 (eGFR), 93,004 (pain), 117,555 (ADI), 100,981 (state anti-TGD policies).
Number missing: 62 (marital status), 11 (priority score), 537 (BMI), 509 (hypertension), 34 (smoking status), 2,120 (lipids), 43 (diabetes), 1,287 (eGFR), 503 (pain), 772 (ADI), 684 (state anti-TGD policies).
Number missing: 81 (marital status), 10 (priority score), 504 (BMI), 476 (hypertension), 33 (smoking status), 1,598 (lipids), 40 (diabetes), 1,096 (eGFR), 476 (pain), 608 (ADI), 542 (state anti-TGD policies).
Number missing: 369 (marital status), 48 (priority score), 2,481 (BMI), 2,351 (hypertension), 399 (smoking status), 7,629 (lipids), 149 (diabetes), 5,097 (eGFR), 2,204 (pain), 2,624 (ADI), 2,211 (state anti-TGD policies).
Statistical analyses
Summary statistics were calculated overall and by gender identity. Continuous variables are reported as mean (SD) and were compared using analysis of variance. Categorical variables are reported as frequency (percentage) and were compared using chi-square tests.
Unadjusted HF incidence rates were calculated by gender modality and identity as the number of events divided by total person-years and are reported per 1,000 person-years. After confirming the proportional hazards assumption using Schoenfeld residuals, Cox proportional hazards regression models were used to estimate the total and direct effect of gender identity on incident HF, with cisgender men and women serving as referent groups. We developed a directed acyclic graph (Figure 1) in accordance with the minority stress framework,36, 37, 38 and the conceptual model of cardiovascular health among TGD individuals.7,39 To estimate the total effect of gender identity on incident HF, multivariable-adjusted model 1 included adjustment for presumed confounders of the association between gender identity and HF: age, race, Hispanic ethnicity, and sexual minority identity (Supplemental Methods).
Figure 1.
Directed Acyclic Graph of Gender Identity and Heart Failure
This directed acyclic graph (DAG) depicts hypothesized pathways linking gender identity to incident heart failure (HF) within a minority stress framework. Gender identity is the exposure and HF the outcome. Gray nodes represent social and structural determinants, blue nodes represent measured clinical and behavioral factors, and the purple node represents minoritized gender identity. Green arrows indicate pathways contributing to the total effect, whereas red arrows denote key confounding pathways (eg, age, race and ethnicity, sexual orientation). The DAG informed covariate selection for multivariable analyses. TGD = transgender and gender diverse.
Then, to estimate the direct effect of gender identity on incident HF, multivariable-adjusted model 2 included additional adjustment for: 1) presumed mediators of this association which we directly measured in our data (hypertension, lipids, diabetes, BMI, estimated glomerular filtration rate, substance use, anxiety, depression, and HIV); 2) proxies for unmeasured mediators (homelessness, VA enrollment priority group, and ADI as proxies for individual-level and neighborhood-level socioeconomic status); and 3) an instrumental variable (the state-level anti-TGD policy index in state of residence) to account for any lingering unmeasured confounding (Supplemental Methods).40
Secondary analyses repeated the calculation of incidence rates and Cox proportional hazards regression modeling with separate analyses for HFrEF, HFpEF, HFpEF with EF decline, and HFimpEF. Sensitivity analyses were restricted to 145,474 veterans who self-reported their gender identity and repeated the primary analyses with self-identified gender as the exposure.
Analyses were performed in R (R Foundation for Statistical Computing, version 4.4.1)41 and a 2-sided P value <0.05 was considered statistically significant.
Results
Among 1,103,923 veterans, a total of 42,157 (3.8%) were classified as TGD. The mean age of TGD veterans (46 years) was between that of cisgender women (40 years) and cisgender men (53 years) (Table 1). The distribution of most traditional cardiovascular risk factors (eg, lipids, BMI, hypertension, and diabetes) were similar across gender identities but unclassified TGD individuals were the most likely to have currently or ever smoked and to have a recorded alcohol use disorder (Table 1). In addition, TGD veterans were more likely to experience homelessness, have attempted suicide, have a diagnosed anxiety disorder, have an HIV diagnosis, and live in states with greater legal protections for TGD individuals than their cisgender peers (Table 1).
Primary analyses: association of gender identity with any HF
During a median follow-up of 9.4 years (25th-75th percentile 5.7-10.1 years), 107,766 incident HF diagnoses were observed, 3,078 of which were among TGD veterans. Overall incident HF rates were lowest among cisgender women at 3.87 cases per 1,000 person-years (95% CI: 3.76-3.98), and highest among cisgender men at 13.05 cases per 1,000 person-years (95% CI: 12.97-13.13) (Table 2). Among TGD veterans, incident HF rates per 1,000 person-years ranged from 6.22 (95% CI: 5.66-8.83) in transfeminine people to 8.89 (95% CI: 8.52-9.27) in unclassified TGD people (Table 2). When estimating the total effect of gender identity on HF (ie, adjusting for presumed confounders), gender minority status was associated with a greater HF risk compared to cisgender women (Table 2) (HR [95% CI]: 1.53 [1.46-1.60]) but lower HF risk compared to cisgender men (Table 2) (HR [95% CI]: 0.90 [0.86-0.93]). On additional adjustment for presumed mediators (clinical comorbidities), proxies for unmeasured variables (ADI, VA enrollment priority group), and an instrumental variable (state-level tally of anti-TGD policies), the direct effect was slightly attenuated compared to the total effect, but retained statistical significance (Table 2). When disaggregating TGD identities, we failed to observe a significant difference in HF risk when comparing transmasculine individuals to cisgender women and when comparing unclassified TGD individuals to cisgender men (Table 2, Model 2).
Table 2.
Association Between Gender Identity and Any HF
| Gender Identity | Incident HF Events/N | Incidence Rate [95% CI]a | Model 1 (Total Effect)b |
Model 2 (Direct Effect)c |
||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Versus Cisgender Women |
Versus Cisgender Men |
Versus Cisgender Women |
Versus Cisgender Men |
|||||||
| HR [95% CI] | P Value | HR [95% CI] | P Value | HR [95% CI] | P Value | HR [95% CI] | P Value | |||
| Cisgender man | 99,905/917,466 | 13.05 [12.97-13.13] | -- | -- | 1.00 | -- | -- | -- | 1.00 | -- |
| Cisgender woman | 4,783/144,300 | 3.87 [3.76-3.98] | 1.00 | -- | -- | -- | 1.00 | -- | -- | -- |
| TGD | 3,078/42,157 | 8.19 [7.90-8.48] | 1.53 [1.46-1.60] | <0.001 | 0.90 [0.86-0.93] | <0.001 | 1.27 [1.21-1.33] | <0.001 | 0.92 [0.88-0.95] | <0.001 |
| Transmasculine | 508/7,129 | 7.78 [7.11-8.48] | 1.37 [1.25-1.51] | <0.001 | 0.83 [0.76-0.90] | <0.001 | 1.09 [0.99-1.20] | 0.074 | 0.79 [0.72-0.86] | <0.001 |
| Transfeminine | 446/7,991 | 6.22 [5.66-6.83] | 1.45 [1.32-1.60] | <0.001 | 0.79 [0.72-0.87] | <0.001 | 1.23 [1.11-1.36] | <0.001 | 0.82 [0.74-0.90] | <0.001 |
| Unclassified TGD | 2,124/27,037 | 8.89 [8.52-9.27] | 1.59 [1.51-1.67] | <0.001 | 0.94 [0.90-0.98] | 0.006 | 1.33 [1.26-1.41] | <0.001 | 0.98 [0.94-1.02] | 0.305 |
HF = heart failure; other abbreviation as in Table 1.
Calculated as number of events over total person-years and reported per 1,000 person-years. CI calculated using the Poisson distribution.
Adjusted for age, race, Hispanic ethnicity, and sexual minority identity.
Additionally adjusted for BMI, diabetes, hypertension, lipids, eGFR, substance use, anxiety, depression, HIV, homelessness, ADI, VA enrollment priority group, and state-level tally of anti-TGD policies in veterans’ state of residence.
Secondary analyses: association of gender identity with HFrEF, HFpEF, HFpEF with EF decline, and HFimpEF
Although results assessing the total and direct effect of gender identity on HFrEF were largely similar to those of any HF (Table 3), the association between gender identity and HFpEF was less clear. When adjusting for presumed confounders (age, race, Hispanic ethnicity, and sexual minority identity), gender minority status was associated with a greater HFpEF risk compared to cisgender women (Table 3) (HR [95% CI]: 1.17 [1.08-1.27]) and lower HFpEF risk compared to cisgender men (Table 3) HR [95% CI]: 0.93 [0.88-0.99]). However, the direct effect of gender minority status on incident HFpEF vs cisgender women was fully attenuated to nonsignificance, and although TGD veterans as a group possessed lower HFpEF risk than cisgender men, only a transmasculine identity was associated with lower HFpEF risk.
Table 3.
Association Between Gender Identity and HFpEF and HFrEF
| Gender Identity | Incident Events/N | Incidence Rate [95% CI]a | Model 1 (Total Effect)b |
Model 2 (Direct Effect)c |
||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Versus Cisgender Women |
Versus Cisgender Men |
Versus Cisgender Women |
Versus Cisgender Men |
|||||||
| HR [95% CI] | P Value | HR [95% CI] | P Value | HR [95% CI] | P Value | HR [95% CI] | P Value | |||
| HFpEF | ||||||||||
| Cisgender man | 36,307/917,466 | 4.74 [4.69-4.79] | -- | -- | 1.00 | -- | -- | -- | 1.00 | -- |
| Cisgender woman | 2,242/144,300 | 1.81 [1.74-1.89] | 1.00 | -- | -- | -- | 1.00 | -- | -- | -- |
| TGD | 1,164/42,157 | 3.10 [2.92-3.28] | 1.22 [1.13-1.31] | <0.001 | 0.93 [0.88-0.99] | 0.014 | 1.01 [0.94-1.09] | 0.797 | 0.94 [0.89-1.00] | 0.043 |
| Trans masculine | 211/7,129 | 3.23 [2.81-3.69] | 1.19 [1.04-1.38] | 0.015 | 0.94 [0.82-1.07] | 0.357 | 0.92 [0.80-1.07] | 0.296 | 0.85 [0.74-0.98] | 0.022 |
| Trans feminine | 183/7,991 | 2.55 [2.20-2.95] | 1.25 [1.07-1.45] | 0.004 | 0.88 [0.76-1.02] | 0.084 | 1.04 [0.89-1.21] | 0.595 | 0.89 [0.77-1.03] | 0.112 |
| Unclassified TGD | 770/27,037 | 3.22 [3.00-3.46] | 1.22 [1.12-1.32] | <0.001 | 0.94 [0.87-1.01] | 0.088 | 1.03 [0.94-1.12] | 0.554 | 0.98 [0.91-1.06] | 0.630 |
| HFrEF | ||||||||||
| Cisgender man | 37,240/917,466 | 4.86 [4.81-4.91] | -- | -- | 1.00 | -- | -- | -- | 1.00 | -- |
| Cisgender woman | 1,318/144,300 | 1.07 [1.01-1.13] | 1.00 | -- | -- | -- | 1.00 | -- | -- | -- |
| TGD | 1,171/42,157 | 3.12 [2.94-3.30] | 2.18 [2.01-2.36] | <0.001 | 0.87 [0.82-0.92] | <0.001 | 1.77 [1.63-1.93] | <0.001 | 0.90 [0.85-0.95] | <0.001 |
| Trans masculine | 181/7,129 | 2.77 [2.38-3.21] | 1.87 [1.59-2.19] | <0.001 | 0.76 [0.66-0.88] | 0.003 | 1.51 [1.28-1.77] | <0.001 | 0.76 [0.66-0.88] | <0.001 |
| Trans feminine | 161/7,991 | 2.25 [1.91-2.62] | 1.99 [1.69-2.35] | <0.001 | 0.74 [0.63-0.86] | 0.001 | 1.67 [1.42-1.98] | <0.001 | 0.77 [0.66-0.90] | 0.001 |
| Unclassified TGD | 829/27,037 | 3.47 [3.24-3.71] | 2.30 [2.11-2.51] | <0.001 | 0.93 [0.87-1.00] | 0.049 | 1.87 [1.70-2.06] | <0.001 | 0.96 [0.90-1.03] | 0.310 |
HFpEF = heart failure with preserved ejection fraction; HFrEF = heart failure with reduced ejection fraction; other abbreviation as in Table 1.
Calculated as number of events over total person-years and reported per 1,000 person-years. CI calculated using the Poisson distribution.
Adjusted for age, race, Hispanic ethnicity, and sexual minority identity.
Additionally adjusted for BMI, diabetes, hypertension, lipids, eGFR, substance use, anxiety, depression, HIV, homelessness, ADI, VA enrollment priority group, and state-level tally of anti-TGD policies in veterans’ state of residence.
In assessing HFpEF with EF decline, gender minority status was associated with greater risk compared to cisgender women (Table 4) (HRdirect [95% CI]: 1.15 [1.02-1.30]) but not compared to cisgender men (Table 4) (HRdirect [95% CI]: 0.99 [0.90-1.08]). In assessing HFimpEF, gender minority status was associated with lower risk vs cisgender women (Table 4) (HRdirect [95% CI]: 0.65 [0.58-0.74]) but not vs cisgender men (Table 4) (HRdirect [95% CI]: 0.98 [0.90-1.07]). The association between TGD subgroups and HFpEF with EF decline and HFimpEF was not explored due to a paucity of events in some subgroups. The percentage of each gender identity among TGD individuals stratified by HF phenotype is summarized in Supplemental Table 6. Results are displayed in Central Illustration.
Table 4.
Association of Gender Identity With HFpEF With EF Decline and HFimpEF
| Gender Identity | Incident Events/N | Incidence Rate [95% CI]a | Model 1 (Total Effect)b |
Model 2 (Direct Effect)c |
||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Versus Cisgender Women |
Versus Cisgender Men |
Versus Cisgender Women |
Versus Cisgender Men |
|||||||
| HR [95% CI] | P Value | HR [95% CI] | P Value | HR [95% CI] | P Value | HR [95% CI] | P Value | |||
| HFpEF with EF decline | ||||||||||
| Cisgender man | 15,548/32,439 | 163.31 [160.75-165.90] | -- | -- | 1.00 | -- | -- | -- | 1.00 | -- |
| Cisgender woman | 842/2,034 | 130.02 [121.38-139.10] | 1.00 | -- | -- | -- | 1.00 | -- | -- | -- |
| TGD | 485/1,059 | 152.32 [139.06-166.49] | 1.14 [1.02-1.28] | 0.021 | 0.97 [0.88-1.06] | 0.473 | 1.15 [1.02-1.30] | 0.023 | 0.99 [0.90-1.08] | 0.749 |
| HFimpEF | ||||||||||
| Cisgender man | 15,768/34,685 | 161.31 [158.80-163.85] | -- | -- | 1.00 | -- | -- | -- | 1.00 | -- |
| Cisgender woman | 780/1,244 | 262.55 [244.45-281.64] | 1.00 | -- | -- | -- | 1.00 | -- | -- | -- |
| TGD | 516/1,120 | 161.52 [147.88-176.08] | 0.64 [0.57-0.72] | <0.001 | 0.98 [0.90-1.07] | 0.647 | 0.65 [0.58-0.74] | <0.001 | 0.98 [0.90-1.07] | 0.579 |
EF = ejection fraction; HFimpEF = heart failure with improved ejection fraction; other abbreviations as in Tables 1 and 3.
Calculated as number of events over total person-years and reported per 1,000 person-years. CI calculated using the Poisson distribution.
Adjusted for age, race, Hispanic ethnicity, and sexual minority identity.
Additionally adjusted for BMI, diabetes, hypertension, lipids, eGFR, substance use, anxiety, depression, HIV, homelessness, ADI, VA enrollment priority group, and state-level anti-TGD policies in veterans’ state of residence.
Central Illustration.
Heart Failure Incidence, Phenotype, and Trajectory in Gender-Diverse Veterans
Central illustration summarizing associations between gender identity and heart failure (HF) in a national Veterans Health Administration cohort. Transgender and gender diverse (TGD) individuals exhibited incident HF risk intermediate between cisgender women and men. Differences in HF phenotype were observed, with patterns in HFrEF and HFpEF varying by gender identity. TGD veterans demonstrated higher risk of HFpEF with ejection fraction decline and lower likelihood of HF with improved ejection fraction compared with cisgender women. These findings highlight the influence of gender identity on HF epidemiology and support incorporation of gender identity into cardiovascular risk stratification and prevention strategies. EF = ejection fraction; HFpEF = heart failure with preserved ejection fraction; HFrEF = heart failure with reduced ejection fraction.
Sensitivity analyses: association between self-identified gender and HF
Among 145,474 veterans free of prevalent HF who self-identified their gender identity, there were 14,212 incident HF events, only 14 of which occurred among TGD veterans (Supplemental Table 7). In regression models, self-identified TGD veterans’ HF risk was lower than that of self-identified cisgender men but did not significantly differ from that of cisgender women (Supplemental Table 7). However, because of the low number of events among self-identified TGD veterans, these results should be interpreted with caution.
Discussion
In this national cohort of veterans receiving care in the VHA, gender identity was associated with differences in the incidence and phenotype of HF. Prior studies have suggested that TGD populations experience higher risk of atherosclerotic CVD relative to cisgender peers and may experience differences in incident HF.10,12 However, differences in HF phenotype and trajectories among TGD populations have not been well characterized. The present study extends prior work by evaluating HF phenotypes and trajectories in a large national cohort.
The present study leveraged a large national EHR data set to identify more than 42,000 TGD adults, contributing to a growing body of cardiovascular outcomes research among SGM populations despite historical limitations in SOGI data capture.8, 9, 10, 11,42 Importantly, policies affecting transgender military service continue to evolve, with implications for both access to care and the ascertainment of gender identity within VHA administrative data.43
In this study, incident HF rates among TGD veterans were broadly consistent with estimates reported in administrative claims analyses of commercially insured TGD populations, and higher than contemporary commercially-insured, nonolder adult cohorts.12,44,45 Differences across studies likely reflect variation in population characteristics, HF definitions, and underlying cardiovascular risk profiles.
Important differences were observed across gender identities and HF phenotypes. HF risk patterns among TGD veterans did not align consistently with either sex assigned at birth or affirmed gender alone. Transfeminine veterans experienced intermediate overall HF risk between cisgender women and cisgender men, whereas transmasculine veterans experienced overall HF risk similar to cisgender women. These findings may reflect differences in cumulative sex hormone exposure, including residual androgenic exposure (eg from natal puberty and variable suppression on femininizing GAHT) and acquired estrogen exposure or differences in body composition and cardiometabolic burden.
When stratified by HF phenotype, rates of incident HFrEF among TGD veterans were intermediate between those observed among cisgender men and cisgender women. This pattern may reflect the known association between androgen exposure and adverse ventricular remodeling.46, 47, 48 Associations between gender identity and HFpEF were less consistent, although transmasculine veterans consistently experienced lower HFpEF risk than cisgender men, mirroring findings observed in prior claims-based analyses. The lower HFpEF risk among transmasculine veterans may reflect lower prevalence of obesity, atrial fibrillation, or other HFpEF-associated comorbidities, although residual confounding by unmeasured cardiometabolic factors cannot be excluded. The consistency of these findings with prior administrative claims analyses supports the need for further investigation into mechanisms underlying gender identity-associated differences in HF risk.12
In contrast to prior claims-based analyses, we did not observe higher HFpEF risk among transfeminine veterans in fully adjusted models. Similarly, differences between transmasculine veterans and cisgender women were attenuated after sequential adjustment. Because many cardiovascular risk factors included in these models may function as mediators rather than traditional confounders of the relationship between gender identity and HF risk, adjustment for these variables may attenuate associations between gender identity and HF risk and limits causal interpretation of fully adjusted models.
Collectively, these observations suggest that HF risk in TGD populations reflects complex interactions among hormonal exposure, cardiometabolic burden, minority stress exposure, social vulnerability, and health care access. These findings may reflect differing contributions of androgen and estrogen exposure, body composition, vascular remodeling, inflammation, and cumulative cardiometabolic burden. Estrogen signaling may influence myocardial fibrosis and diastolic function, whereas androgen exposure may contribute to adverse ventricular remodeling and cardiometabolic profiles.15,46 These sex hormone-mediated pathways are particularly relevant to TGD populations, in whom the type, timing, and duration of hormonal exposure may differ substantially from cisgender peers. For transfeminine people, exogenous estrogen combined with antiandrogen therapy may modify HFpEF risk through effects on myocardial stiffness, adiposity, and volume homeostasis, whereas residual androgen exposure from incomplete suppression may sustain some degree of HFrEF risk. Among transmasculine people, exogenous testosterone may promote left ventricular hypertrophy and adverse cardiometabolic changes, although shorter cumulative exposure relative to cisgender men may attenuate these effects. GAHT may therefore plausibly modify HF phenotype risk in a subgroup-specific manner, although hormone exposure could not be adequately characterized in the present data set.
Chronic psychosocial stress and discrimination may additionally contribute to cardiovascular risk through inflammatory, autonomic, endothelial, and neurohormonal pathways increasingly recognized as contributors to HF development.6,7,36, 37, 38 The higher prevalence of homelessness, anxiety, depression, HIV, smoking, and alcohol use disorder observed among TGD veterans in this cohort likely reflects broader structural inequities that may contribute to HF risk across the life course. These exposures may differentially affect HF phenotype; for example, HIV-associated cardiomyopathy and substance use may contribute more directly to HFrEF.
These interpretive considerations are further complicated by the large unclassified TGD subgroup, which comprises a heterogeneous population including veterans with incomplete SOGI documentation, nonbinary identities, or insufficient longitudinal data to assign transfeminine or transmasculine classification. This heterogeneity may limit direct comparisons with other gender identity groups and attenuate subgroup-specific associations. However, inclusion of this subgroup is important; it represents the largest proportion of TGD individuals identifiable using NLP-based methods and its composition likely reflects real-world diversity in gender identity that is inadequately captured by current EHR infrastructure. An estimated 20% of transgender men and women are not on GAHT,49 so those who validly self-identified as a “man” or “woman” who did not have records indicating receipt of GAHT would not be categorized as having self-identified as gender minority, further deflating PPV. These limitations underscore the need for systematic collection of self-reported SOGI data in large cardiovascular data sets.
The present study also examined trajectories of left ventricular EF and the transition between HF phenotypes across gender identities. Among veterans with incident HFpEF, decline in EF was common; similarly, among veterans with baseline HFrEF, subsequent HFimpEF was also commonly observed. Differences in HF trajectories between TGD and cisgender veterans further suggest potential heterogeneity in myocardial recovery and remodeling across gender identities. Lower rates of HFimpEF among TGD veterans compared with cisgender women may reflect differences in myocardial reserve, treatment access or adherence, competing comorbidity burden, or HF etiology.16 Disparities in guideline-directed medical therapy access or adherence among TGD veterans, potentially driven by health care avoidance, discrimination, or competing psychosocial needs, may also contribute to lower recovery rates and warrant further investigation. Conversely, relatively high rates of transition from HFpEF to HFrEF may suggest differing remodeling pathways across gender-diverse populations, although mechanistic investigation is needed.50
Collectively, these findings suggest that gender identity may be relevant to HF risk stratification. Our findings add to growing evidence that HF risk differs across gender identities and that the epidemiology of HF is incompletely characterized without inclusion of TGD populations. These results support the need for inclusive, population-level cardiovascular research to clarify mechanisms of HF risk, improve risk prediction, and advance equitable cardiovascular care.
Study Limitations
This study has several limitations. First, this observational analysis cannot establish causal relationships between gender identity and incident HF. Second, HF outcomes were identified using administrative codes, which may result in misclassification of HF phenotype. Third, gender identity was inferred using NLP and administrative algorithms rather than self-report. Although the algorithm demonstrated high sensitivity and specificity, the overall PPV for TGD classification was modest (31.1%), reflecting the algorithm's combined prediction of both gender modality and gender identity rather than modality alone. Importantly, the large unclassified TGD subgroup reflects both algorithmic limitations and the inherent incompleteness of SOGI documentation within routine clinical care; many veterans had documentation suggestive of gender diversity without sufficient structured or longitudinal information to confidently assign a binary transfeminine or transmasculine classification. Accordingly, the unclassified TGD group should not be interpreted as a biologically or clinically homogeneous population. Fourth, incomplete data regarding GAHT exposure and duration limited our ability to evaluate the contribution of hormonal therapy to HF risk and may have introduced residual confounding. Because hormonal exposure likely mediates part of the relationship between gender identity and HF, this represents an important gap in the present analysis. Fifth, sensitivity analyses restricted to veterans with self-reported gender identity were limited by small sample size and low event counts. Finally, follow-up duration was modest in a relatively young cohort, mortality was not modeled as a competing risk, and findings from this veteran population may not be generalizable to all TGD populations.
Conclusions
In this national cohort of veterans receiving care in the VHA, gender identity was associated with differences in the incidence and phenotype of HF. These findings extend current understanding of HF epidemiology and underscore the importance of including TGD populations in cardiovascular research to improve risk characterization and advance equitable cardiovascular care.
Funding support and author disclosures
Dr Streed was partially supported through the American Heart Association career development grant (20CDA35320148), National Heart, Lung, and Blood Institute career development grant (K01HL151902), Doris Duke Charitable Foundation (Grant #2022061). Dr Streed received consultation fees from L’Oreal, the Texas Health Institute, and the Research Institute for Gender Therapeutics unrelated to this work. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
Acknowledgments
The authors humbly thank transgender and gender diverse people who graciously participate in research to improve community well-being. The authors also acknowledge veterans who are transgender and gender diverse for their service. The authors thank T. Elizabeth Workman, PhD, John R. O’Leary MA, and Melissa Skanderson, MSW for assistance with this manuscript.
Footnotes
The statements in this article are solely the responsibility of the authors and do not necessarily represent the views of the AHA, DDCF, NIH, or author employers.
The authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’ institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information, visit the Author Center.
Appendix
For an expanded Methods section as well as supplemental tables and figures, please see the online version of this paper.
Supplementary material
References
- 1.Heidenreich P.A., Albert N.M., Allen L.A., et al. Forecasting the impact of heart failure in the United States: a policy statement from the American Heart Association. Circ Heart Fail. 2013;6:606–619. doi: 10.1161/HHF.0b013e318291329a. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Martin S.S., Aday A.W., Almarzooq Z.I., et al. 2024 heart disease and stroke statistics: a report of US and global data from the American Heart Association. Circulation. 2024;149:e347–e913. doi: 10.1161/CIR.0000000000001209. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Sayed A., Abramov D., Fonarow G.C., et al. Reversals in the Decline of Heart Failure Mortality in the US, 1999 to 2021. JAMA Cardiol. 2024:585–589. doi: 10.1001/jamacardio.2024.0615. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Herman J.L., Flores A.R., O’Neill K.K. The Williams Institute; UCLA School of Law: 2025. How Many Adults and Youth Identify as Transgender in the United States? [Google Scholar]
- 5.Jones J. 2026. LGBTQ+ Identification Holds at 9% in U.S.: Gallup. [Google Scholar]
- 6.Caceres B.A., Streed C.G., Jr., Corliss H.L., et al. Assessing and addressing cardiovascular health in LGBTQ adults: a scientific statement from the American Heart Association. Circulation. 2020;142:e321–e332. doi: 10.1161/CIR.0000000000000914. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Streed C.G., Jr., Beach L.B., Caceres B.A., et al. Assessing and addressing cardiovascular health in people who are transgender and gender diverse: a scientific statement from the American Heart Association. Circulation. 2021;144:e136–e148. doi: 10.1161/CIR.0000000000001003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Deb B., Porter K., van Cleeff A., Reardon L.C., Cook S. Emphasizing sexual orientation and gender identity data capture for improved cardiovascular care of the LGBTQ+ population. JAMA Cardiol. 2024;9:295–302. doi: 10.1001/jamacardio.2023.5267. [DOI] [PubMed] [Google Scholar]
- 9.Porter K.F., Deb B., Katyukha A., Punnanithinont N., Fradley M.G., Cook S.C. Reporting sex and gender differences in cardiovascular research. US Cardiol. 2024;18 doi: 10.15420/usc.2024.18. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Streed C.G., Jr., Duncan M.S., Heier K.R., et al. Incident atherosclerotic cardiovascular disease among veterans by gender identity: a cohort study. J Gen Intern Med. 2025;40:3149–3158. doi: 10.1007/s11606-025-09701-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Streed C.G., Jr., Duncan M.S., Heier K.R., et al. Prevalent atherosclerotic cardiovascular disease among veterans by sexual orientation. J Am Heart Assoc. 2024;13 doi: 10.1161/JAHA.124.036898. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Everitt I.K., Quinn E.K., Duncan M.S., et al. Prevalence and incidence of heart failure phenotypes among transgender and gender-diverse adults in the United States. J Am Heart Assoc. 2026 doi: 10.1161/JAHA.125.046269. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Willemars M.M.A., Nabben M., Verdonschot J.A.J., Hoes M.F. Evaluation of the interaction of sex hormones and cardiovascular function and health. Curr Heart Fail Rep. 2022;19:200–212. doi: 10.1007/s11897-022-00555-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Ebong I.A., Appiah D., Mauricio R., et al. Sex hormones and heart failure risk. JACC: Adv. 2025;4 doi: 10.1016/j.jacadv.2025.101650. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Sabbatini A.R., Kararigas G. Menopause-related estrogen decrease and the pathogenesis of HFpEF: JACC review topic of the week. J Am Coll Cardiol. 2020;75:1074–1082. doi: 10.1016/j.jacc.2019.12.049. [DOI] [PubMed] [Google Scholar]
- 16.Aimo A., Vergaro G., Castiglione V., et al. Effect of sex on reverse remodeling in chronic systolic heart failure. JACC Heart Fail. 2017;5:735–742. doi: 10.1016/j.jchf.2017.07.011. [DOI] [PubMed] [Google Scholar]
- 17.Assari S. Veterans and risk of heart disease in the United States: a cohort with 20 years of follow up. Int J Prev Med. 2014;5:703–709. [PMC free article] [PubMed] [Google Scholar]
- 18.Walker L.E., Poltavskiy E., Janak J.C., Beyer C.A., Stewart I.J., Howard J.T. US Military Service and Racial/Ethnic differences in cardiovascular disease: an analysis of the 2011-2016 behavioral risk factor surveillance system. Ethn Dis. 2019;29:451–462. doi: 10.18865/ed.29.3.451. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Mattocks K.M., Kauth M.R., Sandfort T., Matza A.R., Sullivan J.C., Shipherd J.C. Understanding health-care needs of sexual and gender minority veterans: how targeted research and policy can improve health. LGBT Health. 2014;1:50–57. doi: 10.1089/lgbt.2013.0003. [DOI] [PubMed] [Google Scholar]
- 20.Ashley F., Brightly-Brown S., Rider G.N. Beyond the trans/cis binary: introducing new terms will enrich gender research. Nature. 2024;630:293–295. doi: 10.1038/d41586-024-01719-9. [DOI] [PubMed] [Google Scholar]
- 21.Workman T.E., Goulet J.L., Brandt C., et al. IEEE International Conference on Big Data (Big Data); 2020. A Prototype Application to Identify LGBT Patients in Clinical Notes; pp. 4270–4275. [Google Scholar]
- 22.Streed C.G., King D., Grasso C., et al. Validation of an administrative algorithm for transgender and gender diverse persons against self-report data in electronic health records. J Am Med Inform Assoc. 2023;30:1047–1055. doi: 10.1093/jamia/ocad039. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Jasuja G.K., de Groot A., Quinn E.K., et al. Beyond gender identity disorder diagnoses codes: an examination of additional methods to identify transgender individuals in administrative databases. Med Care. 2020;58:903–911. doi: 10.1097/MLR.0000000000001362. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Wolfe H.L., Reisman J.I., Yoon S.S., et al. Validating data-driven methods for identifying transgender individuals in the Veterans Health administration of the US Department of Veterans affairs. Am J Epidemiol. 2021;190:1928–1934. doi: 10.1093/aje/kwab102. [DOI] [PubMed] [Google Scholar]
- 25.Haskell S.G., Brandt C., Burg M., et al. Incident cardiovascular risk factors among men and women veterans after return from deployment. Med Care. 2017;55:948–955. doi: 10.1097/MLR.0000000000000801. [DOI] [PubMed] [Google Scholar]
- 26.Equality Maps Snapshot: LGBTQ Equality By State . Movement Advancement Project; 2025. [Google Scholar]
- 27.Delgado C., Baweja M., Crews D.C., et al. A unifying approach for GFR estimation: recommendations of the NKF-ASN task force on reassessing the inclusion of race in diagnosing kidney disease. Am J Kidney Dis. 2022;79:268–288.e1. doi: 10.1053/j.ajkd.2021.08.003. [DOI] [PubMed] [Google Scholar]
- 28.James P.A., Oparil S., Carter B.L., et al. 2014 evidence-based guideline for the management of high blood pressure in adults: report from the panel members appointed to the Eighth Joint National Committee (JNC 8) JAMA. 2014;311:507–520. doi: 10.1001/jama.2013.284427. [DOI] [PubMed] [Google Scholar]
- 29.American Diabetes Association Professional Practice C. 2 Diagnosis and classification of diabetes: standards of care in Diabetes-2024. Diabetes Care. 2024;47:S20–S42. doi: 10.2337/dc24-S002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Wyse J.J., McGinnis K.A., Edelman E.J., et al. Twelve-month retention in opioid agonist treatment for opioid use disorder among patients with and without HIV. AIDS Behav. 2022;26:975–985. doi: 10.1007/s10461-021-03452-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.McGinnis K.A., Skanderson M., Edelman E.J., et al. Impact of behavioral and medication treatment for alcohol use disorder on changes in HIV-related outcomes among patients with HIV: a longitudinal analysis. Drug Alcohol Depend. 2020;217 doi: 10.1016/j.drugalcdep.2020.108272. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Singh G.K. Area deprivation and widening inequalities in US mortality, 1969-1998. Am J Public Health. 2003;93:1137–1143. doi: 10.2105/ajph.93.7.1137. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Kind A.J., Jencks S., Brock J., et al. Neighborhood socioeconomic disadvantage and 30-day rehospitalization: a retrospective cohort study. Ann Intern Med. 2014;161:765–774. doi: 10.7326/M13-2946. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Little R.J.A., Rubin D.B. Third edition. Wiley; Hoboken, NJ: 2020. Statistical analysis with missing data. [Google Scholar]
- 35.Rubin D.B. Wiley-Interscience; Hoboken, N.J.: 2004. Multiple imputation for nonresponse in surveys. [Google Scholar]
- 36.Brooks V.R. Lexington Books; Lexington, Mass: 1981. Minority stress and lesbian women. [Google Scholar]
- 37.Frost D.M., Meyer I.H. Minority stress theory: application, critique, and continued relevance. Curr Opin Psychol. 2023;51 doi: 10.1016/j.copsyc.2023.101579. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Meyer I.H. Prejudice, social stress, and mental health in lesbian, gay, and bisexual populations: conceptual issues and research evidence. Psychol Bull. 2003;129:674–697. doi: 10.1037/0033-2909.129.5.674. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Caceres B.A., Streed C.G., Jr., Corliss H.L., et al. Assessing and addressing cardiovascular health in LGBTQ adults: a scientific statement from the American Heart Association. Circulation. 2020;142 doi: 10.1161/CIR.0000000000000914. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Newhouse J.P., McClellan M. Econometrics in outcomes research: the use of instrumental variables. Annu Rev Public Health. 1998;19:17–34. doi: 10.1146/annurev.publhealth.19.1.17. [DOI] [PubMed] [Google Scholar]
- 41.R Core Team . R Foundation for Statistical Computing; Vienna, Austria: 2025. R: A Language and Environment for Statistical Computing. [Google Scholar]
- 42.Beach L.B., Streed C.G., Jr. Beyond sex-based medicine in cardiovascular health research and practice. Circulation. 2023;147:529–531. doi: 10.1161/CIRCULATIONAHA.122.063156. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Department of Veterans Affairs; Veterans Health Administration . U.S. Department of Veterans Affairs; Washington, DC: 2025. Rescission of VHA Directive 1341(4), Providing Health Care for Transgender and Intersex Veterans. [Google Scholar]
- 44.Gerber Y., Weston S.A., Redfield M.M., et al. A contemporary appraisal of the heart failure epidemic in Olmsted County, Minnesota, 2000 to 2010. JAMA Intern Med. 2015;175:996–1004. doi: 10.1001/jamainternmed.2015.0924. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Goyal A., Norton C.R., Thomas T.N., et al. Predictors of incident heart failure in a large insured population: a one million person-year follow-up study. Circ Heart Fail. 2010;3:698–705. doi: 10.1161/CIRCHEARTFAILURE.110.938175. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Santos J.D., Oliveira-Neto J.T., Tostes R.C. The cardiovascular subtleties of testosterone on gender-affirming hormone therapy. Am J Physiol Heart Circ Physiol. 2023;325:H30–H53. doi: 10.1152/ajpheart.00015.2023. [DOI] [PubMed] [Google Scholar]
- 47.Murphy C.N., Delles C., Davies E., Connelly P.J. Cardiovascular disease in transgender individuals. Atherosclerosis. 2023;384 doi: 10.1016/j.atherosclerosis.2023.117282. [DOI] [PubMed] [Google Scholar]
- 48.Moreira Allgayer R.M.C., Borba G.D.S., Moraes R.S., Ramos R.B., Spritzer P.M. The effect of gender-affirming hormone therapy on the risk of subclinical atherosclerosis in the transgender population: a systematic review. Endocr Pract. 2023;29:498–507. doi: 10.1016/j.eprac.2022.12.017. [DOI] [PubMed] [Google Scholar]
- 49.Castelijn J.M., Huisman B., Steensma T.D., et al. Prevalence of gender-affirming hormone therapy in non-binary and genderqueer individuals, a systematic review and meta-analysis. Endocrine. 2025;90:420–426. doi: 10.1007/s12020-025-04381-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Paulus W.J., Tschope C. A novel paradigm for heart failure with preserved ejection fraction: comorbidities drive myocardial dysfunction and remodeling through coronary microvascular endothelial inflammation. J Am Coll Cardiol. 2013;62:263–271. doi: 10.1016/j.jacc.2013.02.092. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.