Abstract
Background
People living with HIV (PLWH) face a higher risk of cardiovascular disease (CVD), with significant sex differences in outcomes. However, it is unclear whether these differences are driven by distinct risk factor profiles or by a differential impact of shared risk factors. This study aimed to identify factors associated with CVD among PLWH and to specifically assess for effect modification by sex using a large, diverse nationwide database.
Methods
We utilized data from the All of Us Research Program (AoU). The primary outcome was a composite of coronary artery disease or stroke. We used multivariable logistic regression to identify factors associated with CVD in the overall cohort. To assess for effect modification, we introduced interaction terms between sex and key covariates. Subsequently, sex‐stratified analyses were performed to explore these differences.
Results
Among 6464 PLWH (4608 men and 1856 women), women had a higher prevalence of CVD than men (24.8% vs. 21.9%, p = 0.011). We identified significant interaction effects between sex and several key risk factors, including hypertension, unemployment and hyperlipidaemia (p < 0.05 for all). In stratified analyses comparing women to men within these risk groups, the association of hypertension with CVD was substantially stronger in women than in men (adjusted odds ratio [aOR] = 4.928, 95% CI: 2.827–8.586). Similarly, the effects of unemployment and hyperlipidaemia on CVD were more pronounced in women. In fully stratified models, a detectable viral load was a significant risk factor for CVD only among men (aOR = 1.524, 95% CI: 1.130–2.049).
Conclusions
While many traditional and HIV‐specific CVD risk factors are shared between men and women living with HIV (WLWH), our findings reveal that the magnitude of their effect is not uniform. The impact of key risk factors, particularly hypertension, is substantially greater in women, suggesting a heightened vulnerability to these exposures. These findings underscore the critical need for sex‐specific risk assessments and aggressively tailored prevention strategies for PLWH.
Keywords: all of us, CVD, HIV, risk factors, sex differences
INTRODUCTION
With contemporary antiretroviral therapy (ART), people living with HIV (PLWH) are living longer [1] but experiencing a rising burden of cardiovascular diseases (CVDs) [2, 3]. Recent findings from the REPRIEVE trial, the largest global randomized controlled trial for primary CVD prevention among PLWH, underscored the importance of addressing CVD risk in this population [4]. Compared with people without HIV, the relative risks of various CVD manifestations are generally 1.5 to 2‐fold greater for PLWH [5]. The global burden of CVD among PLWH tripled over the past 2 decades and accounted for 2.6 million disability‐adjusted life‐years per year [5]. The increased burden of CVD among PLWH can be driven by traditional cardiometabolic risk factors and persistent immune activation and inflammation [6, 7].
Current U.S. guidelines for CVD prevention in PLWH, such as the 2019 American Heart Association scientific statement, recommend routine screening for traditional risk factors and assessment, while noting that these models do not include HIV‐specific factors and may underestimate risk [8, 9]. The 2024 IDSA Primary Care Guidance for Persons With HIV further reinforces these principles, emphasizing comprehensive cardiovascular risk management as part of routine HIV care [10]. However, while these guidelines emphasize PLWH as a high‐risk group, they do not currently provide sex‐specific recommendations for CVD risk stratification or management. This critical gap underscores the importance of research that can clarify whether risk factors have a differential impact on men versus women, as such evidence is essential for informing more precise and equitable clinical guidance.
Amidst this context, a growing body of evidence indicates that significant sex differences exist. Women living with HIV (WLWH) have a 1.5 to 2‐fold higher relative risk of CVD compared to women without HIV across a range of disease phenotypes, such as myocardial infarction (MI), stroke and heart failure [11]. A large US cohort study by Kaiser Healthcare found that WLWH had a 2.48‐fold higher risk of heart failure, compared to 1.57‐fold for men living with HIV (MLWH) [12]. Another study conducted in Boston, USA, comparing stroke incidence between PLWH and control groups, found that while the association between HIV and stroke was insignificant for men, it was strongly significant for WLWH compared to women without HIV [13].
The reasons for these sex‐based disparities in CVD risk among PLWH are not yet fully elucidated, and several critical knowledge gaps remain. First, it is unclear whether observed disparities are merely a reflection of varying risk factor distributions or if the same risk factors, such as hypertension, exert a differential impact on MLWH and WLWH. Second, despite evidence suggesting that WLWH represent a uniquely vulnerable population, they have remained a relatively understudied group in CVD research [14, 15]. Finally, data are particularly limited regarding these sex differences among diverse and underrepresented populations of PLWH in the United States.
To address these gaps, we aimed to investigate sex differences in CVD events among the PLWH population living in the United States using the nationwide cohort data from the All of Us Research Program (AoU). Specifically, our objectives were to [1] compare the prevalence of CVD and a comprehensive profile of its associated risk factors (demographic, traditional, HIV‐related and lipid biomarkers) between MLWH and WLWH and [2] evaluate whether sex modifies the association between key risk factors and CVD, utilizing both interaction testing and sex‐stratified multivariable models.
METHODS
Data source and study participants
This study used data from the All of Us (AoU) program, which is an ongoing national effort supported by NIH. The AoU data includes a broad, diverse group of the US population, with more than 50% of the participants from racial and ethnic minority groups and more than 80% from populations historically underrepresented in biomedical research (e.g. sexual minorities, geographically disadvantaged locations, individuals with disabilities). The AoU programme collects longitudinal observations of clinical, environmental, lifestyle and genetic data from the participants [16].
Following a similar computational phenotyping approach in a previous study [17], we have identified a cohort of 6464 PLWH (4608 MLWH and 1856 WLWH) aged 18 years and older, using data available up to October 2023. Individuals were excluded from study if they had a CVD diagnosis prior to their HIV/AIDS diagnosis.
Measures
CVD
CVD was defined as a composite endpoint of coronary heart disease (CHD) or stroke (both ischaemic and haemorrhagic) [18]. The identification of CHD was based on a combination of electronic health record (EHR) data and survey responses. For EHR‐based identification, we followed the MidSouth Clinical Data Research Network Coronary Heart Disease Algorithm. This algorithm employs a broad definition that includes not only ischaemic heart diseases but also other related cardiac conditions, such as myocarditis (I51.4), cardiomegaly (I51.7) and Takotsubo syndrome (I51.81) [18]. The complete list of ICD‐9 and ICD‐10 diagnostic and procedure codes used by this algorithm is provided in Table S1. Additionally, CHD was identified by an affirmative response to either of the survey questions: ‘Has a doctor or health care provider ever told you that you had a heart attack?’ or ‘Has a doctor or health care provider ever told you that you have coronary artery/coronary heart disease?’ Stroke cases were identified using previously validated ICD‐9 and ICD‐10 codes (Table S1) or by an affirmative response to the survey question, ‘Has a doctor or health care provider ever told you that you had a stroke?’
Demographic characteristics and traditional risk factors
Demographic characteristics included age, race, ethnicity, insurance status, income level, education, marital status and employment, which all came from the Health Surveys Questionnaires in AoU. We combined self‐reported responses to the past medical history survey and data (ICD codes) in the EHRs to ascertain the presence of prominent vascular risk factors, including hypertension (Observational Medical Outcomes Partnership [OMOP] code 316866), hyperlipidaemia (OMOP code 432867) and type 2 diabetes mellitus (OMOP code 201826). We also extracted self‐reported data from the lifestyle survey to ascertain smoking status (yes or no), alcohol use frequency (never, >2 weekly, >2 monthly) and statin use (yes or no). In addition, we used data from physical measurements to calculate the body mass index (BMI), which was categorized as normal or underweight, overweight and obese with cutoffs of <25, 25–30 and >30, respectively.
HIV related markers and inflammatory biomarkers
We retrieved the most recent CD4 counts and HIV viral load before CVD diagnosis or end of study, whichever occurred first. CD4 counts were categorized into <500 cells/mm3 and ≥500 cells/mm3. HIV viral loads were classified as <200 copies/mL (virally suppressed) and ≥200 copies/mL (virally unsuppressed). For biomarkers, total cholesterol, Low‐Density Lipoprotein Cholesterol (LDL‐C) and High‐Density Lipoprotein Cholesterol (HDL‐C) were collected from blood tests. Total cholesterol was classified into an optimal group (<200 mg/dL) and a high group (≥200 mg/dL). LDL‐C was categorized as an optimal group (<130 mg/dL) and a high group (≥130 mg/dL), while HDL‐C was divided into low group (men: <40 mg/dL; women: <50 mg/dL) and an optimal group (men: ≥40 mg/dL; women: ≥50 mg/dL).
Statistical analysis
In this study, all analyses were performed with R software, and a p value <0.05 was considered statistically significant. All demographic, traditional, HIV‐related markers and inflammatory biomarkers were summarized for the study population using frequencies and percentages. These characteristics were stratified by sex and then by the presence or absence of CVD. The chi‐squared test was used to assess for significant differences between those with and without CVD within each sex stratum. To identify factors independently associated with CVD, we developed a primary multivariable logistic regression model, adjusting for a comprehensive set of covariates, including demographic characteristics, traditional risk factors, HIV‐related markers and inflammatory biomarkers. All results from this model were reported as adjusted odds ratios with their corresponding 95% confidence intervals (CIs).
To investigate if risk factors for CVD differed between MLWH and WLWH, we tested for effect modification by introducing interaction terms between sex and key covariates into the primary model. For covariates that demonstrated a statistically significant interaction, we then performed a stratified analysis to directly compare the odds of CVD between females and males within that specific subgroup (e.g. comparing females with hypertension to males with hypertension). In addition to this, to provide a complete picture of sex‐specific risk profiles, the multivariable models were also run separately for MLWH and all WLWH. Finally, a sensitivity analysis repeated the primary multivariable model on an expanded cohort—including individuals with a CVD diagnosis prior to their HIV diagnosis—to confirm the robustness of the findings.
RESULTS
Participant characteristics
The sample characteristics of MLWH (n = 4608) and WLWH (n = 1856) with CVD compared to those without CVD was shown in Table 1. WLWH exhibited a slightly higher CVD prevalence than MLWH (24.8%, 461/1856 vs. 21.9%, 1008/4608; p = 0.011). For both sexes, compared to individuals without CVD, those with CVD were significantly older, less likely to be employed and had a higher prevalence of diabetes, hypertension and hyperlipidaemia. Furthermore, those with CVD were more likely to have a CD4 count <500 cells/mm3 and low HDL‐C. Additionally, among WLWH, those with CVD were more likely to be of Black or African American race and report a lower income. We also compared the characteristics of the study population with the total population before excluding individuals diagnosed with CVD prior to their HIV diagnosis and found no significant differences, suggesting our exclusion criteria did not introduce significant selection bias (see Table S2).
TABLE 1.
Demographic characteristics of participants stratified by sex and end‐point CVD status.
| Variables | MLWH | WLWH | ||||
|---|---|---|---|---|---|---|
| No CVD (n = 3600) | CVD (n = 1008) | P | No CVD (n = 1395) | CVD (n = 461) | P | |
| Age | ||||||
| 18–44 | 1068 (29.7%) | 139 (13.8%) | <0.001 | 368 (26.4%) | 91 (19.7%) | <0.001 |
| 45–54 | 689 (19.1%) | 258 (25.6%) | 342 (24.5%) | 159 (34.5%) | ||
| 55–64 | 1211 (33.6%) | 404 (40.1%) | 470 (33.7%) | 156 (33.8%) | ||
| 65+ | 632 (17.6%) | 207 (20.5%) | 215 (15.4%) | 55 (11.9%) | ||
| Race | ||||||
| White | 1263 (35.1%) | 369 (36.6%) | 0.610 | 243 (17.4%) | 57 (12.4%) | 0.033 |
| Black or African American | 1437 (39.9%) | 399 (39.6%) | 776 (55.6%) | 278 (60.3%) | ||
| Asian/Other/Unknown | 900 (25.0%) | 240 (23.8%) | 376 (27.0%) | 126 (27.3%) | ||
| Ethnicity | ||||||
| Not Hispanic or Latino | 2842 (78.9%) | 815 (80.9%) | 0.201 | 1083 (77.6%) | 363 (78.7%) | 0.666 |
| Hispanic or Latino | 758 (21.1%) | 193 (19.1%) | 312 (22.4%) | 98 (21.3%) | ||
| Insurance status | ||||||
| Yes | 3196 (88.8%) | 960 (95.2%) | <0.001 | 1231 (88.2%) | 432 (93.7%) | 0.004 |
| No | 289 (8.0%) | 26 (2.6%) | 101 (7.2%) | ‐‐ | ||
| Unknown | 115 (3.2%) | 22 (2.2%) | 63 (4.5%) | ‐‐ | ||
| Income level | ||||||
| <25 k | 1662 (46.2%) | 475 (47.1%) | 0.201 | 803 (57.6%) | 288 (62.5%) | 0.045 |
| 25–50 k | 556 (15.4%) | 161 (16.0%) | 141 (10.1%) | 46 (10.0%) | ||
| >50 k | 782 (21.7%) | 189 (18.8%) | 107 (7.7%) | ‐‐ | ||
| Unknown | 600 (16.7%) | 183 (18.2%) | 344 (24.7%) | 108 (23.4%) | ||
| Education | ||||||
| High school degree or more | 3373 (93.7%) | 946 (93.8%) | 0.789 | 1228 (88.0%) | 417 (90.5%) | 0.111 |
| Less than high school degree | 99 (2.8%) | 30 (3.0%) | 87 (6.2%) | 29 (6.3%) | ||
| Unknown | 128 (3.6%) | 32 (3.2%) | 80 (5.7%) | ‐‐ | ||
| Employment status | ||||||
| Employed | 1413 (39.3%) | 213 (21.1%) | <0.001 | 359 (25.7%) | 85 (18.4%) | 0.002 |
| Not employed | 2187 (60.8%) | 795 (78.9%) | 1036 (74.3%) | 376 (81.6%) | ||
| Marital status | ||||||
| Married/Living with partner | 935 (26.0%) | 273 (27.1%) | <0.001 | 337 (24.2%) | 96 (20.8%) | 0.232 |
| Divorced/Separated/Widowed | 620 (17.2%) | 235 (23.3%) | 476 (34.1%) | 178 (38.6%) | ||
| Never Married | 1909 (53.0%) | 454 (45.0%) | 513 (36.8%) | 161 (34.9%) | ||
| Unknown | 136 (3.8%) | 46 (4.6%) | 69 (4.9%) | 26 (5.6%) | ||
| Viral load | ||||||
| <200 copies/mL | 988 (27.4%) | 229 (22.7%) | 0.010 | 393 (28.2%) | 137 (29.7%) | 0.218 |
| ≥200 copies/mL | 342 (9.5%) | 105 (10.4%) | 202 (14.5%) | 52 (11.3%) | ||
| Unknown | 2270 (63.1%) | 674 (66.9%) | 800 (57.3%) | 272 (59.0%) | ||
| CD4 count | ||||||
| ≥500 cells/mm3 | 920 (25.6%) | 225 (22.3%) | 0.001 | 380 (27.2%) | 124 (26.9%) | <0.001 |
| <500 cells/mm3 | 626 (17.4%) | 224 (22.2%) | 208 (14.9%) | 104 (22.6%) | ||
| Unknown | 2054 (57.1%) | 559 (55.5%) | 807 (57.8%) | 233 (50.5%) | ||
| BMI | ||||||
| Normal or underweight | 1363 (37.9%) | 293 (29.1%) | <0.001 | 342 (24.5%) | 89 (19.3%) | <0.001 |
| Obese | 834 (23.2%) | 214 (21.2%) | 684 (49.0%) | 195 (42.3%) | ||
| Overweight | 1114 (30.9%) | 238 (23.6%) | 299 (21.4%) | 83 (18.0%) | ||
| Unknown | 289 (8.0%) | 263 (26.1%) | 70 (5.0%) | 94 (20.4%) | ||
| Diabetes | ||||||
| No | 3102 (86.2%) | 754 (74.8%) | <0.001 | 1122 (80.4%) | 332 (72.0%) | <0.001 |
| Yes | 498 (13.8%) | 254 (25.2%) | 273 (19.6%) | 129 (28.0%) | ||
| Hypertension | ||||||
| No | 2199 (61.1%) | 325 (32.2%) | <0.001 | 808 (57.9%) | 112 (24.3%) | <0.001 |
| Yes | 1401 (38.9%) | 683 (67.8%) | 587 (42.1%) | 349 (75.7%) | ||
| Hyperlipidaemia | ||||||
| No | 2847 (79.1%) | 521 (51.7%) | <0.001 | 1098 (78.7%) | 260 (56.4%) | <0.001 |
| Yes | 753 (20.9%) | 487 (48.3%) | 297 (21.3%) | 201 (43.6%) | ||
| Smoking history | ||||||
| No | 1350 (37.5%) | 343 (34.0%) | <0.001 | 587 (42.1%) | 198 (43.0%) | 0.429 |
| Yes | 2131 (59.2%) | 649 (64.4%) | 761 (54.6%) | 253 (54.9%) | ||
| Unknown | 119 (3.3%) | ‐‐ | 47 (3.4%) | ‐‐ | ||
| Stain use | ||||||
| No | 2830 (78.6%) | 546 (54.2%) | <0.001 | 1102 (79.0%) | 266 (57.7%) | <0.001 |
| Yes | 770 (21.4%) | 462 (45.8%) | 293 (21.0%) | 201 (43.6%) | ||
| Alcohol frequency | ||||||
| Never/<1 time weekly | 2346 (65.2%) | 683 (67.8%) | 0.235 | 870 (62.4%) | 310 (67.2%) | 0.043 |
| >2 weekly | 785 (21.8%) | 196 (19.4%) | 200 (14.3%) | 46 (10.0%) | ||
| Unknown | 469 (13.0%) | 129 (12.8%) | 325 (23.3%) | 105 (22.8%) | ||
| Total cholesterol group | ||||||
| Optimal | 1423 (39.5%) | 458 (45.4%) | <0.001 | 532 (38.1%) | 195 (42.3%) | 0.016 |
| High | 378 (10.5%) | 133 (13.2%) | 202 (14.5%) | 82 (17.8%) | ||
| Unknown | 1799 (50.0%) | 417 (41.4%) | 661 (47.4%) | 184 (39.9%) | ||
| LDL‐C group | ||||||
| Optimal | 1530 (42.5%) | 496 (49.2%) | <0.001 | 607 (43.5%) | 227 (49.2%) | 0.030 |
| High | 218 (6.1%) | 63 (6.3%) | 105 (7.5%) | 41 (8.9%) | ||
| Unknown | 1852 (51.4%) | 449 (44.5%) | 683 (49.0%) | 193 (41.9%) | ||
| HDL‐C group | ||||||
| Optimal | 1199 (33.3%) | 336 (33.3%) | <0.001 | 400 (28.7%) | 121 (26.2%) | <0.001 |
| Low | 588 (16.3%) | 254 (25.2%) | 330 (23.7%) | 154 (33.4%) | ||
| Unknown | 1813 (50.4%) | 418 (41.5%) | 665 (47.7%) | 186 (40.3%) | ||
Note: Cells with counts less than 20 are masked (‐‐) to protect confidentiality.
Abbreviations: BMI, body mass index; CVD, cardiovascular disease; HDL‐C, high‐density lipoprotein cholesterol; LDL‐C, low‐density lipoprotein cholesterol.
Factors associated with CVD
The results from the multivariable logistic regression analysis identifying factors associated with CVD were presented in Table 2. After adjusting for all covariates, being unemployed was associated with more than double the odds of having CVD (aOR = 2.152, 95% CI: 1.810–2.567). The 45–54 age group also showed increased odds compared to the 18–44 reference group (aOR = 1.551, 95% CI: 1.267–1.900). Conversely, having no insurance was associated with significantly lower odds of having a CVD diagnosis (aOR = 0.407, 95% CI: 0.285–0.567). Beyond these factors, hypertension (aOR = 2.355, 95% CI: 2.031–2.733), hyperlipidaemia (aOR = 1.766, 95% CI: 1.500–2.078) and low HDL‐C (aOR = 1.444, 95% CI: 1.209–1.724) remained strongly associated with higher odds of CVD. Finally, among HIV‐specific markers, a CD4 count <500 cells/mm3 was associated with significantly increased odds of CVD (aOR = 1.627, 95% CI: 1.334–1.984).
TABLE 2.
Multivariable logistic regression results for factors associated with cardiovascular disease among People living with HIV.
| Variables | aOR | 95% CI | p |
|---|---|---|---|
| Sex (Male) | |||
| Female | 1.070 | (0.918, 1.247) | 0.383 |
| Age group (18–44 years) | |||
| 45–54 | 1.551 | (1.267, 1.900) | <0.001 |
| 55–64 | 1.050 | (0.863, 1.280) | 0.627 |
| 65+ | 0.732 | (0.577, 0.928) | 0.010 |
| Race (White) | |||
| Black or African American | 1.015 | (0.855, 1.205) | 0.869 |
| Asian/Other/Unknown | 1.165 | (0.910, 1.491) | 0.225 |
| Ethnicity (Not Hispanic or Latino) | |||
| Hispanic or Latino | 0.840 | (0.657, 1.074) | 0.164 |
| Insurance Status (Yes) | |||
| No | 0.407 | (0.285, 0.567) | <0.001 |
| Unknown | 0.643 | (0.416, 0.967) | 0.040 |
| Income level (<25 k) | |||
| 25–50 k | 1.245 | (1.010, 1.532) | 0.039 |
| >50 k | 1.027 | (0.813, 1.294) | 0.825 |
| Unknown | 1.202 | (1.007, 1.432) | 0.041 |
| Education (High school degree or more) | |||
| Less than high school degree | 0.945 | (0.671, 1.315) | 0.741 |
| Unknown | 0.679 | (0.469, 0.966) | 0.035 |
| Employment status (Employed) | |||
| Unemployed | 2.152 | (1.810, 2.567) | <0.001 |
| Marital status (Married/Living with partner) | |||
| Divorced/Separated/Widowed | 1.144 | (0.947, 1.382) | 0.163 |
| Never Married | 0.888 | (0.750, 1.052) | 0.167 |
| Unknown | 1.262 | (0.902, 1.753) | 0.169 |
| BMI (Normal or underweight) | |||
| Overweight | 0.863 | (0.721, 1.032) | 0.107 |
| Obese | 0.800 | (0.668, 0.957) | 0.015 |
| Unknown | 5.479 | (4.437, 6.776) | <0.001 |
| Hypertension (No) | |||
| Yes | 2.355 | (2.031, 2.733) | <0.001 |
| Diabetes (No) | |||
| Yes | 1.042 | (0.884, 1.226) | 0.623 |
| Smoking history (No) | |||
| Yes | 1.002 | (0.872, 1.153) | 0.975 |
| Unknown | 0.464 | (0.285, 0.728) | 0.001 |
| Alcohol frequency (Never/less than 1 time weekly) | |||
| >2 weekly | 0.835 | (0.699, 0.995) | 0.045 |
| Unknown | 0.964 | (0.797, 1.163) | 0.700 |
| Statin use (No) | |||
| Yes | 1.543 | (1.314, 1.811) | <0.001 |
| Hyperlipidaemia (No) | |||
| Yes | 1.766 | (1.500, 2.078) | <0.001 |
| CD4 count (≥500 cells/mm3) | |||
| <500 cells/mm3 | 1.627 | (1.334, 1.984) | <0.001 |
| Unknown | 0.933 | (0.754, 1.156) | 0.527 |
| Viral load (<200 copies/mL) | |||
| ≥200 copies/mL | 1.205 | (0.948, 1.527) | 0.126 |
| Unknown | 1.551 | (1.274, 1.890) | <0.001 |
| Total cholesterol (Optimal) | |||
| High group | 1.255 | (0.978, 1.605) | 0.072 |
| Unknown | 1.167 | (0.832, 1.638) | 0.372 |
| LDL‐C (Optimal) | |||
| High group | 0.761 | (0.552, 1.045) | 0.093 |
| Unknown | 1.326 | (0.875, 1.999) | 0.180 |
| HDL‐C (Optimal) | |||
| Low group | 1.444 | (1.209, 1.724) | <0.001 |
| Unknown | 0.611 | (0.391, 0.958) | 0.031 |
Abbreviations: BMI, body mass index; CVD, cardiovascular disease; HDL‐C, high‐density lipoprotein cholesterol; LDL‐C, low‐density lipoprotein cholesterol.
Interaction effects of sex on CVD risk
We further explored potential interaction effects between sex and key covariates on CVD, with results presented in Table 3. We detected significant interaction effects between sex and age, employment status, viral load level, hypertension and hyperlipidaemia (all p < 0.05). To better understand these differences, we performed stratified analyses comparing the odds of CVD in females versus males within these specific subgroups (Figure 1). The modifying effect of sex was most pronounced for hypertension; among participants with hypertension, females had nearly five times the odds of having CVD compared to males (aOR = 4.928, 95% CI: 2.827–8.586). Similarly, among unemployed individuals, females had 91% higher odds of CVD than their male counterparts (aOR = 1.914, 95% CI: 1.172–3.128). A significant sex difference was also observed among participants with hyperlipidaemia, where females had approximately double the odds of CVD compared to males (aOR = 1.976, 95% CI: 1.081–3.607). No significant differences in the odds of CVD between females and males were observed in the 55–64 and 65+ age groups or among those with VL ≥200 copies/mL (Figure 1).
TABLE 3.
Multivariable logistic regression results for interaction effects between sex and key covariates associated with cardiovascular disease among People living with HIV.
| Interactions | aOR | 95% CI | p |
|---|---|---|---|
| Sex × Age Group | |||
| Male × Age 18–44 | Ref | ||
| Female × Age 45–54 | 0.676 | (0.442, 1.033) | 0.070 |
| Female × Age 55–64 | 0.523 | (0.343, 0.796) | 0.003 |
| Female × Age 65+ | 0.399 | (0.236, 0.670) | 0.001 |
| Sex × Employment Status | |||
| Male × Employed | Ref | ||
| Female × Unemployed | 0.663 | (0.467, 0.944) | 0.022 |
| Sex × Viral Load Level | |||
| Male × VL < 200 copies/mL | Ref | ||
| Female × VL ≥200 copies/mL | 0.485 | (0.292, 0.799) | 0.005 |
| Female × VL Unknown | 0.813 | (0.537, 1.232) | 0.328 |
| Sex × CD4 Count | |||
| Male × CD4 ≥ 500 cells/mm3 | Ref | ||
| Female × CD4 < 500 cells/mm3 | 1.133 | (0.740, 1.736) | 0.565 |
| Female × CD4 Unknown | 0.760 | (0.498, 1.161) | 0.204 |
| Sex × Hypertension | |||
| Male × No Hypertension | Ref | ||
| Female × Hypertension | 1.705 | (1.236, 2.361) | 0.001 |
| Sex × Statin Use | |||
| Male × No Statin Use | Ref | ||
| Female × Statin Use | 0.896 | (0.634, 1.265) | 0.533 |
| Sex × Hyperlipidaemia | |||
| Male × No Hyperlipidaemia | Ref | ||
| Female × Hyperlipidaemia | 0.684 | (0.485, 0.962) | 0.029 |
Note: All covariates included in Table 2 were retained in this multivariable model. This table presents only the interaction effects between sex and selected variables.
FIGURE 1.
Forest plot of interaction effects between sex and key covariates on CVD. All covariates included in Table 2 were retained in this multivariable model. This figure presents only the significant interaction effects from Table 3.
Sex‐Stratified Multivariable Analysis
In sex‐stratified multivariable models (Table S3), we identified both shared and unique risk factors for CVD. For both men and women, unemployed, CD4 count <500 cells/mm3, hyperlipidaemia and statin use were significantly associated with higher odds of CVD. Hypertension was a strong risk factor for both sexes, with a more pronounced effect observed in women (aOR = 3.627, 95% CI: 2.721–4.868) than in men (aOR = 2.020, 95% CI: 1.695–2.410). Key sex‐specific differences were also noted; a detectable viral load (≥200 copies/mL) was a significant risk factor only for men (aOR = 1.524, 95% CI: 1.130–2.049), whereas older age was unexpectedly associated with lower odds of CVD only among women (55–64 years: aOR = 0.653, 95% CI: 0.455–0.936; 65+ years: aOR = 0.364, 95% CI: 0.227–0.580).
Sensitivity analysis
To assess the robustness of our findings, we conducted a sensitivity analysis that included all participants with CVD, regardless of whether the diagnosis occurred before or after their HIV diagnosis (Table S4). The results of this analysis were highly consistent with our primary model from Table 2. All key factors identified in the main analysis—including unemployment, hypertension, hyperlipidaemia and a low CD4 count—remained statistically significant, with odds ratios of a similar direction and magnitude. This consistency indicates that our findings are robust and not substantially influenced by the exclusion of individuals with pre‐existing CVD.
DISCUSSION
Evidence increasingly highlights sex‐specific cardiovascular risks, and the 2021 Lancet Commission on CVD in Women [19] noted that women's long‐standing underrepresentation in clinical research has created major knowledge gaps, contributing to persistent underdiagnosis and undertreatment. Using a large and diverse nationwide cohort, this study identified a critical distinction in CVD risk among PLWH: although many risk factors are shared between MLWH and WLWH, their effects are not uniform. The primary finding is that the effect of key risk factors—notably hypertension, unemployment, and hyperlipidaemia—was substantially greater in WLWH, revealing a unique cardiovascular vulnerability. These findings highlight the need to consider sex as a fundamental effect modifier in CVD risk assessment among PLWH.
In this study, we observed that CVD prevalence among WLWH was 24.8%, higher than the 21.9% observed in MLWH. Our findings were consistent with previous studies. For example, a study reported unexpectedly high rates of CVD among WLWH living in high‐income countries [20]. Additionally, another study examining CHD among PLWH in Brazil revealed that, over a 10‐year follow‐up period, women were three times more likely than men to develop CHD [21]. Moreover, they found no protective effect of female sex on CVD after adjusting for multiple risk factors such as smoking, hypertension and low HDL‐C [22]. This reminded us that WLWH face a distinct disadvantage regarding CVD.
Our study also elucidates the underlying source of CVD risk disparity between MLWH and WLWH. We demonstrated that the difference arises not merely from risk factor distribution but more crucially from differences in their effect magnitude. After multivariable adjustment, the associations of hypertension and hyperlipidaemia with CVD were approximately five‐fold and two‐fold stronger, respectively, in WLWH compared with MLWH, which is consistent with evidence from general populations that women may be more susceptible to vascular damage under elevated blood pressure or dyslipidaemia stress [23, 24, 25]. Coupled with HIV‐associated chronic inflammation [26], such sex‐specific vascular sensitivity may help explain why the same risk exposures translate to higher CVD risk in women, though direct mechanistic studies in PLWH remain sparse. Moreover, unemployment was associated with increased CVD risk in both sexes but exerted a stronger effect in women. This pattern aligns with broader evidence that socioeconomic disadvantage disproportionately harms women's cardiovascular health through pathways involving financial strain, reduced access to care, psychosocial stress and adverse health behaviours [27, 28, 29].
We also identified several additional sex‐specific associations. Detectable viral load was a significant CVD risk factor only among men, possibly reflecting sex differences in immune activation and viral pathophysiology [30, 31]. Interestingly, while prior studies have reported strong correlations between HDL‐C levels and CVD risk in PLWH [32, 33], our study complemented existing evidence and revealed a sex difference in this relationship: low HDL‐C was significantly associated with increased CVD risk in MLWH but not in WLWH. This sex difference may partly reflect the physiological effects of oestrogen, which raises HDL‐C levels and enhances HDL function, thereby contributing to the relative cardioprotection observed in women [34, 35]. In contrast, HIV infection and higher viral load have been linked to reduced HDL‐C and impaired cholesterol efflux, which may more strongly affect men, who generally have lower baseline HDL‐C [36, 37, 38]. Older age was unexpectedly associated with lower odds of CVD among women, which may reflect survivor bias or complex interactions among ageing, hormonal changes and cardiovascular risk not captured in cross‐sectional analyses [14, 39]. Finally, higher CD4 counts were associated with a lower risk of CVD in both sexes, consistent with evidence that low CD4 counts promote chronic immune activation and inflammation, contributing to atherogenesis through endothelial dysfunction and a prothrombotic state [40].
Our analysis also revealed several seemingly paradoxical findings that warrant cautious interpretation. The apparent ‘protective’ effect of no insurance is unlikely to represent a true biological benefit but rather reflects detection bias—individuals without stable access to care are less likely to undergo diagnostic evaluations, resulting in under‐ascertainment of CVD in EHR [41, 42]. Similarly, the observed association between statin use and higher odds of CVD is a classic example of confounding by indication, as statins are prescribed primarily to individuals already at elevated cardiovascular risk [43, 44]. The inverse association between obesity and CVD may represent the ‘obesity paradox’ [45, 46], as reported in other chronic disease populations, in which weight loss may result from the underlying CVD itself. Therefore, this disease‐related weight loss or other residual confounding, rather than obesity itself, may account for the observed higher risk among those with normal BMI. wherein residual confounding or disease‐related weight loss may make normal BMI appear riskier in cross‐sectional analyses. Likewise, the lack of association between smoking or alcohol use and CVD in our cohort should be interpreted cautiously [47, 48, 49, 50, 51], as it may reflect differences in exposure classification, self‐report bias or the underrepresentation of heavy users in our sample.
Our study has some limitations. First, the cross‐sectional design limits the establishment of any causal relationships from the data. Second, some key variables (e.g. viral load, CD4 count) had a significant amount of missing data, which may affect the robustness of our analysis. And there is a large number of missing data on biomarkers, which limited the inclusion of more inflammatory biomarkers in our analysis, such as D‐dimer and C‐reactive protein. Third, reliance on some self‐reported diagnoses may have introduced misclassification bias. Fifth, some variables, such as cholesterol levels, were based on a single measurement, which may not fully reflect changes over time. In addition, data on menopause status were not used, limiting our ability to explore its potential impact on CVD risk among WLWH. Finally, the imbalance of sample size between MLWH (n = 4608) and WLWH (n = 1856) may introduce bias and affect the power of comparison.
CONCLUSION
In this large, diverse nationwide cohort of PLWH, we identified significant sex differences in the prevalence and determinants of CVD. Although some traditional and HIV‐related risk factors were shared between men and women, their magnitudes of association with CVD were not uniform. Hypertension, hyperlipidaemia and unemployment exerted stronger effects in women, highlighting a distinct cardiovascular vulnerability in this population. Additional findings—including the role of viral load, CD4 count, HDL‐C and behavioural and socioeconomic determinants—underscore the complex interplay between biological and structural factors in shaping CVD risk among PLWH. These results emphasize the importance of incorporating sex‐specific risk assessment and targeted prevention strategies into HIV care to reduce cardiovascular disparities and improve long‐term outcomes.
AUTHOR CONTRIBUTIONS
HZ and XY contributed to organizing the study and drafting the manuscript. HX was responsible for data collection and analysis. QL organized the study results. FS, XL and SW provided valuable input on the study design and offered constructive feedback on the manuscript.
Supporting information
Table S1. Codes used to ascertain outcomes in electronic health record.
ACKNOWLEDGEMENTS
The research reported in this publication was supported by the Augusta SCORE pilot project (38920‐6; XY). The content is solely the responsibility of the authors and does not necessarily represent the official views of the funders. Funders had no role in the design of the study, collection, analysis and interpretation of the data.
DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.
REFERENCES
- 1. Michaels SH, Clark R, Kissinger P. Declining morbidity and mortality among patients with advanced human immunodeficiency virus infection. N Engl J Med. 1998;339(6):405‐406. [DOI] [PubMed] [Google Scholar]
- 2. Feinstein MJ, Bahiru E, Achenbach C, et al. Patterns of cardiovascular mortality for HIV‐infected adults in the United States: 1999 to 2013. Am J Cardiol. 2016;117(2):214‐220. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Palella FJ, Baker RK, Moorman AC, et al. Mortality in the highly active antiretroviral therapy era: changing causes of death and disease in the HIV outpatient study. J Acquir Immune Defic Syndr. 2006;43(1):27‐34. [DOI] [PubMed] [Google Scholar]
- 4. Grinspoon SK, Fitch KV, Zanni MV, et al. Pitavastatin to prevent cardiovascular disease in HIV infection. N Engl J Med. 2023;389(8):687‐699. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Shah ASV, Stelzle D, Lee KK, et al. Global burden of atherosclerotic cardiovascular disease in people living with HIV: systematic review and meta‐analysis. Circulation. 2018;138(11):1100‐1112. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Hunt PW, Lee SA, Siedner MJ. Immunologic biomarkers, morbidity, and mortality in treated HIV infection. J Infect Dis. 2016;214(Suppl 2):S44‐S50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Losina E, Hyle EP, Borre ED, et al. Projecting 10‐year, 20‐year, and lifetime risks of cardiovascular disease in persons living with human immunodeficiency virus in the United States. Clin Infect Dis. 2017;65(8):1266‐1271. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Feinstein MJ, Hsue PY, Benjamin LA, et al. Characteristics, prevention, and Management of Cardiovascular Disease in people living with HIV: a scientific statement from the American Heart Association. Circulation. 2019;140(2):e98‐e124. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Triant VA, Lyass A, Hurley LB, et al. Cardiovascular risk estimation is suboptimal in people with HIV. J Am Heart Assoc. 2024;13(10):e029228. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Horberg M, Thompson M, Agwu A, et al. Primary Care Guidance for Providers of Care for Persons With Human Immunodeficiency Virus: 2024 Update by the HIV medicine Association of the Infectious Diseases Society of America. Clin Infect Dis. 2024. doi: 10.1093/cid/ciae479 [DOI] [PubMed] [Google Scholar]
- 11. Alonso A, Barnes AE, Guest JL, Shah A, Shao IY, Marconi V. HIV infection and incidence of cardiovascular diseases: an analysis of a large healthcare database. J Am Heart Assoc. 2019;8(14):e012241. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Go AS, Reynolds K, Avula HR, et al. Human immunodeficiency virus infection and variation in heart failure risk by age, sex, and ethnicity: the HIV HEART study. Mayo Clin Proc. 2022;97(3):465‐479. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Chow FC, Regan S, Feske S, Meigs JB, Grinspoon SK, Triant VA. Comparison of ischemic stroke incidence in HIV‐infected and non‐HIV‐infected patients in a US health care system. J Acquir Immune Defic Syndr. 2012;60(4):351‐358. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Solomon D, Sabin CA, Mallon PWG, Winston A, Tariq S. Cardiovascular disease in women living with HIV: a narrative review. Maturitas. 2018;108:58‐70. [DOI] [PubMed] [Google Scholar]
- 15. Barr E, Marshall LJ, Collins LF, et al. Centring the health of women across the HIV research continuum. Lancet HIV. 2024;11(3):e186‐e194. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Murray J. The ‘all of us’ research program. N Engl J Med. 2019;381(19):1884. [DOI] [PubMed] [Google Scholar]
- 17. Yang X, Zhang J, Cai R, et al. Computational phenotyping with the all of us research program: identifying underrepresented people with HIV or at risk of HIV. JAMIA Open. 2023;6(3):ooad071. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Acosta JN, Leasure AC, Both CP, et al. Cardiovascular health disparities in racial and other underrepresented groups: initial results from the all of us research program. J Am Heart Assoc. 2021;10(17):e021724. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Vogel B, Acevedo M, Appelman Y, et al. The lancet women and cardiovascular disease commission: reducing the global burden by 2030. Lancet (London, England). 2021;397(10292):2385‐2438. [DOI] [PubMed] [Google Scholar]
- 20. Grinspoon SK, Zanni MV, Triant VA, et al. Performance of the pooled cohort equations and D:a:D risk scores among individuals with HIV in a global cardiovascular disease prevention trial: a cohort study leveraging data from REPRIEVE. Lancet HIV. 2025;12(2):e118‐e129. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Fuchs SC, Alencastro PR, Ikeda ML, et al. Risk of coronary heart disease among HIV‐infected patients: a multicenter study in Brazil. The Scientific World J. 2013;2013:163418. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Grinspoon SK, Watanabe M, Ribaudo HJ, et al. Factors affecting risk of CVD events in a global CVD prevention cohort of people with HIV. Clin Infect Dis. 2025. doi: 10.1093/cid/ciaf210 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Nguyen AH, Hurwitz M, Sullivan SA, Saad A, Kennedy JLW, Sharma G. Update on sex specific risk factors in cardiovascular disease. Front Cardiovasc Med. 2024;11:1352675. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Vriend EMC, Galenkamp H, van Valkengoed IGM, van den Born BH. Sex disparities in hypertension prevalence, blood pressure trajectories and the effects of anti‐hypertensive treatment. Blood Press. 2024;33(1):2365705. [DOI] [PubMed] [Google Scholar]
- 25. Ohldieck AE, Midtbø H, Kringeland E, Aune A, Gerdts E. Stronger association between blood pressure and arterial stiffness in older women than men: the Hordaland health study. High Blood Press Cardiovasc Prev. 2025;32(3):353‐361. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Krishnan S, Scully EP. Untangling the sources of HIV‐related cardiovascular risk. Clin Infect Dis. 2025. doi: 10.1093/cid/ciaf211 [DOI] [PubMed] [Google Scholar]
- 27. Bann D, Wright L, Hughes A, Chaturvedi N. Socioeconomic inequalities in cardiovascular disease: a causal perspective. Nat Rev Cardiol. 2024;21(4):238‐249. [DOI] [PubMed] [Google Scholar]
- 28. Kaur G, Masket D, Reddy T, et al. Socioeconomic disparities in Women's cardiovascular health in the United States and Canada. Can J Cardiol. 2024;40(6):1056‐1068. [DOI] [PubMed] [Google Scholar]
- 29. Sanchez Rico M, Plessz M, Airagnes G, et al. Cardiovascular burden and unemployment: a retrospective study in a large population‐based French cohort. PLoS One. 2023;18(7):e0288747. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Hsue PY. Mechanisms of cardiovascular disease in the setting of HIV infection. Can J Cardiol. 2019;35(3):238‐248. [DOI] [PubMed] [Google Scholar]
- 31. Streeck H, Maestri A, Habermann D, et al. Dissecting drivers of immune activation in chronic HIV‐1 infection. EBioMedicine. 2022;83:104182. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Cotter AG, Satchell CS, O'Halloran JA, Feeney ER, Sabin CA, Mallon PW. High‐density lipoprotein levels and 10‐year cardiovascular risk in HIV‐1‐infected patients. AIDS (London, England). 2011;25(6):867‐869. [DOI] [PubMed] [Google Scholar]
- 33. Whelton SP, Deal JA, Zikusoka M, et al. Associations between lipids and subclinical coronary atherosclerosis. Aids. 2019;33(6):1053‐1061. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Herrington DM, Howard TD, Hawkins GA, et al. Estrogen‐receptor polymorphisms and effects of estrogen replacement on high‐density lipoprotein cholesterol in women with coronary disease. N Engl J Med. 2002;346(13):967‐974. [DOI] [PubMed] [Google Scholar]
- 35. Beazer JD, Freeman DJ. Estradiol and HDL function in women ‐ a Partnership for Life. J Clin Endocrinol Metab. 2022;107(5):e2192‐e2194. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Lo J, Rosenberg ES, Fitzgerald ML, et al. High‐density lipoprotein‐mediated cholesterol efflux capacity is improved by treatment with antiretroviral therapy in acute human immunodeficiency virus infection. Open Forum Infect Dis. 2014;1(3):ofu108. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Tort O, Escribà T, Egaña‐Gorroño L, et al. Cholesterol efflux responds to viral load and CD4 counts in HIV+ patients and is dampened in HIV exposed. J Lipid Res. 2018;59(11):2108‐2115. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Lembas A, Załęski A, Peller M, Mikuła T, Wiercińska‐Drapało A. Human immunodeficiency virus as a risk factor for cardiovascular disease. Cardiovasc Toxicol. 2024;24(1):1‐14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Ji H, Kwan AC, Chen MT, et al. Sex differences in myocardial and vascular aging. Circ Res. 2022;130(4):566‐577. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Zanni MV, Grinspoon SK. HIV‐specific immune dysregulation and atherosclerosis. Curr HIV/AIDS Rep. 2012;9(3):200‐205. [DOI] [PubMed] [Google Scholar]
- 41. Al‐Sahab B, Leviton A, Loddenkemper T, Paneth N, Zhang B. Biases in electronic health records data for generating real‐world evidence: an overview. J Healthc Inform Res. 2024;8(1):121‐139. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42. Boyd AD, Gonzalez‐Guarda R, Lawrence K, et al. Equity and bias in electronic health records data. Contemp Clin Trials. 2023;130:107238. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Sendor R, Stürmer T. Core concepts in pharmacoepidemiology: confounding by indication and the role of active comparators. Pharmacoepidemiol Drug Saf. 2022;31(3):261‐269. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44. Johansen ME, Green LA, Sen A, Kircher S, Richardson CR. Cardiovascular risk and statin use in the United States. Ann Fam Med. 2014;12(3):215‐223. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45. Simati S, Kokkinos A, Dalamaga M, Argyrakopoulou G. Obesity paradox: fact or fiction? Curr Obes Rep. 2023;12(2):75‐85. [DOI] [PubMed] [Google Scholar]
- 46. Aggrawal K, Gupta V, Singh B, et al. Exploring the obesity parADOX: a multisystem review. Am J Med Sci. 2025;370(4):315‐320. [DOI] [PubMed] [Google Scholar]
- 47. Silverberg MJ, Levine TM, Lea AN, et al. Cardiovascular disease risk factor control in people with and without HIV. Clin Infect Dis. 2024;78(5):1264‐1271. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48. Kelly SG, Plankey M, Post WS, et al. Associations between tobacco, alcohol, and drug use with coronary artery plaque among HIV‐infected and uninfected men in the multicenter AIDS cohort study. PLoS One. 2016;11(1):e0147822. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49. Lifson AR, Lando HA. Smoking and HIV: prevalence, health risks, and cessation strategies. Curr HIV/AIDS Rep. 2012;9(3):223‐230. [DOI] [PubMed] [Google Scholar]
- 50. Lifson AR, Neuhaus J, Arribas JR, van den Berg‐Wolf M, Labriola AM, Read TR. Smoking‐related health risks among persons with HIV in the strategies for Management of Antiretroviral Therapy clinical trial. Am J Public Health. 2010;100(10):1896‐1903. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51. Grand M, Bia D, Diaz A. Cardiovascular risk assessment in people living with HIV: a systematic review and meta‐analysis of real‐life data. Curr HIV Res. 2020;18(1):5‐18. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1. Codes used to ascertain outcomes in electronic health record.
Data Availability Statement
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.