Association Between Uterine Fibroids and Risk of Atherosclerotic Cardiovascular Disease

Abstract

Background

Uterine fibroids and atherosclerotic cardiovascular disease (ASCVD) share biological pathways, yet whether risk of ASCVD is different among those with fibroids compared with those without remains unexplored in large US cohorts with longitudinal data. This study assessed the association between uterine fibroids and risk of incident ASCVD.

Methods

A US population‐based cohort study was done using Optum’s de‐identified Clinformatics Data Mart Database (2000–2022). Follow‐up continued until an ASCVD event, disenrollment, incident fibroid diagnosis in controls, or June 30, 2022. Individuals with fibroids were exact age‐matched (1:5) to individuals without fibroids with an annual gynecologic claim. Incident ASCVD, a composite of coronary artery disease, cerebrovascular disease, and peripheral artery disease, was evaluated, including individual events (eg, myocardial infarction and ischemic stroke).

Results

Among 450 177 individuals with fibroids and 2 250 885 controls (mean age: 41 years, SD 6.3), the 1‐year and 10‐year cumulative incidence (95% CI) of ASCVD was 0.74% (0.71–0.77) and 5.42% (5.18–5.67) for the fibroid group versus 0.30% (0.29–0.31) and 3.00% (2.90–3.11) for controls. Adjusted analyses showed an increased ASCVD risk in the fibroid group (1‐year risk ratio: 2.47 [95% CI, 2.32–2.61]; 1‐year risk difference, 0.41% [95% CI, 0.40–0.47]; 10‐year risk ratio, 1.81 [95% CI, 1.66–1.96]; 10‐year risk difference, 2.40% [95% CI, 2.07, 2.77]. The increased risk was consistent for all individual components of ASCVD. Results were consistent across race and ethnicity and age subgroup analyses and sensitivity analyses addressing measurement error.

Conclusions

Uterine fibroids are associated with sustained increased ASCVD risks up to 10 years postdiagnosis, supporting targeted ASCVD prevention in this population.

Nonstandard Abbreviations and Acronyms

HCPCS

Healthcare Common Procedure Coding System

RD

risk difference

Research Perspective.

What Is New?

• This large US population‐based cohort study of >2.7 million individuals demonstrates that uterine fibroids are independently associated with a significantly increased risk of atherosclerotic cardiovascular disease, with risk ratios of 2.47 at 1 year and 1.81 at 10 years postdiagnosis compared with individuals without fibroids.

• The cardiovascular risk elevation persists across all individual atherosclerotic cardiovascular disease components (coronary artery disease, cerebrovascular disease, and peripheral artery disease) and remains consistent across different racial, ethnic, and age subgroups, suggesting a robust association.

• The study provides the first comprehensive longitudinal evidence from a large US cohort that uterine fibroids may serve as an important marker for identifying women at elevated cardiovascular risk.

What Question Should Be Addressed Next?

• Future mechanistic studies should investigate the specific biological pathways linking uterine fibroids to atherosclerotic cardiovascular disease development, particularly examining shared inflammatory, hormonal, and vascular remodeling processes that may explain this association.

• Research should evaluate whether fibroid‐specific characteristics (size, number, location, and treatment modality) modify cardiovascular risk and whether incorporating fibroid status into existing cardiovascular risk prediction models improves risk stratification accuracy.

Uterine fibroids (leiomyomas) are common benign gynecological tumors, ¹ impacting >26 million premenopausal individuals in the United States, disproportionately affecting Black individuals. ² , ³ Symptoms include heavy menstrual bleeding, dysmenorrhea, and bulk symptoms (eg, pelvic pressure, urinary frequency). ⁴ Despite their prevalence and impact on quality of life, long‐term health implications remain poorly understood. ⁵

Though the cause remains unclear, fibroids share pathogenic features with atherosclerotic cardiovascular disease (ASCVD). ⁶ , ⁷ Both originate from monoclonal smooth muscle cells, and fibroid growth involves smooth muscle proliferation, fibrosis, and calcification, processes linked to vascular remodeling disorders. ⁷ , ⁸ , ⁹ , ¹⁰ ASCVD, involving plaque accumulation in arterial walls leading to blood flow obstruction, also involves these processes. Fibroids can also trigger excessive inflammatory responses, including cytokines (eg, interleukin‐6, tumor necrosis factor‐α) and chemokines, within the uterine environment. ⁶ , ⁷ These proteins may enter systemic circulation, possibly causing chronic systemic inflammation and oxidative stress, ⁷ , ⁸ , ⁹ , ¹⁰ elevating ASCVD risk. Despite this multifaceted relationship, whether fibroids are associated with future ASCVD remains uncertain.

Studies on fibroids and ASCVD risk factors are sparse and inconclusive, often limited by sample sizes, ¹¹ , ¹² ranging from 104 to 505 individuals with fibroids, and under‐representation of racial and ethnic minority groups most affected. ¹³ Furthermore, cross‐sectional designs hinder establishment of temporality or assessment of uterine fibroids as an ASCVD risk factor. ¹⁴ , ¹⁵ These gaps highlight the need for studies utilizing large, diverse data sets with extended follow‐up. To address these data gaps, we leveraged large‐scale administrative claims data to evaluate the association of uterine fibroids and risk of ASCVD events among US individuals.

METHODS

This cohort study used Optum’s de‐identified Clinformatics Data Mart Database (Optum CDM) (2000–2022), a large US health care database covering inpatient, outpatient, and pharmacy claims. Because the data were fully de‐identified, ethics review and informed consent were not required under the Common Rule. The study followed the STROBE (strengthening the reporting of observational studies in epidemiology) reporting guideline. ¹⁶ This study was approved by the University of Pennsylvania Institutional Review Board and granted waiver of informed consent because only deidentified claims data were used. Data use agreements were in place. Data cannot be directly shared by the authors due to a user data share agreement but could be requested directly from Optum CDM.

Study Population and Exposure Definition

Individuals were eligible if they were registered as “female” in Optum CDM. Individuals were classified as having uterine fibroids requiring (1) >1 inpatient or outpatient claims for fibroids ≥1 day apart, or (2) 1 fibroid claim preceded by transvaginal ultrasonography or pelvic magnetic resonance imaging claim within 30 days (Table S1), ¹⁷ , ¹⁸ , ¹⁹ identified using International Classification of Diseases, Ninth and Tenth Revisions (ICD‐9; ICD‐10), Current Procedural Terminology (CPT), or Healthcare Common Procedure Coding System (HCPCS) codes. Eligible individuals were 18 to 50 years old to focus on premenopausal women during the primary age range when uterine fibroids are clinically relevant and hormonally active, because fibroids typically regress after menopause due to declining estrogen levels. ¹⁷ , ²⁰ This definition of uterine fibroids has been applied in other analyses. ¹⁷ , ¹⁸ , ¹⁹

Individuals with fibroids were exact age‐matched to 5 controls with no prior fibroid claim but with a claim for a general gynecological examination within 30 days of the matched fibroid date, using risk‐set sampling to avoid selection bias; specifically, matched controls were eligible to be re‐recruited as exposed if later diagnosed with fibroids. Baseline was defined as the date of the first fibroid claim or the matched annual gynecologic examination.

Inclusion criteria required ≥183 days of continuous enrollment in the health plan (45‐day gap in coverage allowed) before baseline, no history of a hysterectomy, myomectomy, ASCVD, or menopause, identified using ICD‐9/10 or CPT/HCPCS codes.

Outcome Ascertainment and Follow‐Up Period

The primary outcome was first‐time major ASCVD diagnosis or event, defined by the American Heart Association, ²¹ including coronary artery disease (including stable ischemic heart disease and acute events such as myocardial infarction), cerebrovascular disease (including acute events of transient ischemic attack and ischemic stroke), and peripheral artery disease, ascertained using ICD‐9/10 or CPT/HCPCS codes. Secondary outcomes examined individual ASCVD components of coronary artery disease, cerebrovascular disease, and peripheral artery disease. Acute events were also examined: angina, myocardial infarction, transient ischemic attack, and ischemic stroke. Claims were required to be inpatient or >1 outpatient claim. Event date was the date of the first claim. Individuals were followed from baseline until the first occurrence of event date, end of follow‐up in the data set, disenrollment in Optum CDM, or new fibroid diagnosis for matched controls.

Covariate Ascertainment

Covariates were ascertained using ICD‐9/10, CPT, HCPCS, and National Drug Codes codes (Table S1). A directed acyclic graph guided covariate selection to control for confounding and reduce potential selection bias and overadjustment (Figure S1). Baseline values of sociodemographic (age, race, and ethnicity), cardiovascular risk factors (smoking, obesity, obstructive sleep apnea, type 2 diabetes, hyperlipidemia, hypertension), mental health (depression, anxiety), reproductive and obstetric health (parity, history of hypertensive disorder of pregnancy, history of gestational diabetes, polycystic ovary syndrome, infertility), cancer, health care utilization, and oral contraceptive use were hypothesized as either common causes or risk factors of uterine fibroids and ASCVD or proxies for unmeasured common causes. Incident values of these variables were identified for time‐varying weights explained below.

Age was calculated from birth and cohort entry date. Race and ethnicity were extracted from the Optum’s de‐identified Clinformatics Data Mart Database Socioeconomic (SES) view, which uses a proprietary algorithm to classify race and ethnicity. ²² Other variables were identified by the presence or absence of ICD‐9/10 or CPT/HCPCS codes (Table S1). Diabetes, hyperlipidemia, hypertension, depression, anxiety, infertility, and oral contraceptive use were also identified by medication use, classified using National Drug Codes or American Hospital Formulary Service Pharmacologic‐Therapeutic Classifications, including anti‐diabetic medications, statins, anti‐hypertensive medications, antidepressants, anxiolytics, gonadotropins and clomiphene, and combined hormonal oral contraceptives or oral progestin, respectively.

Statistical Analysis

Baseline differences between groups were compared using counts, proportions, means, and SDs. Distributions of continuous variables were examined using histograms. Three sets of inverse probability weights accounted for (1) baseline confounding, (2) selection bias due to disenrollment in Optum CDM updated at each month of follow‐up, and (3) selection bias induced by censoring control group individuals later diagnosed with uterine fibroids updated at each month of follow‐up. Expanded details on weights are provided in Appendix S1. In brief, inverse probability weights for baseline confounding creates a cohort where treatment groups have similar distributions of measured covariates. Inverse probability weights for selection bias reweights the remaining population to represent those who were lost to follow‐up or censored, thereby reducing bias that could occur if participants who leave the study differ systematically from those who remain. Standardized mean differences statistically compare groups before and after weighting, with standardized mean differences <0.1 indicating sufficient balance.

For each outcome, a pooled logistic regression model, $\begin{array}{c}\text{logit Pr}\left(D_{t+1}=1|D_t=0,A\right)={\theta }_{0t}+{\theta }_{1}A+{\theta }_{2}A*h\left(t\right)\hfill \end{array}$ was fit, where $D_t$ was an indicator for developing the outcome during time $t$, $A$ was an indicator for exposure (1: fibroids; 0: no fibroids), ${\theta }_{0t}$ was a time‐varying intercept, and $h\left(t\right)$ was follow‐up time, modeled with linear and quadratic terms. Since the monthly probability of an event is small, the parameters of a pooled logistic model closely approximate the parameters of a Cox proportional hazards model. ²³ This approach was used to allow the relationship between the exposure and outcome to vary with time, produce standardized cumulative incidence curves, and estimate risks ratios (RR) and risk differences (RD). The model’s predicted values estimated 1‐, 3‐, 5‐, and 10‐year risks of each outcome, RR, and RD. Adjusted models were weighted by the cumulative product of the inverse probability weights accounting for baseline confounding, time‐varying selection bias due to disenrollment in Optum CDM, and informative censoring due to new uterine fibroid diagnosis among matched controls. Weights were stabilized and truncated at the 99^th percentile. 95% CIs were estimated using robust standard errors. ²⁴

There were no missing clinical data because variables were defined as the presence or absence of codes; details on this assumption are discussed below. Race and ethnicity ascertained via Optum CDM included an “Unknown” category; since this variable itself is imputed, this category was maintained.

Subgroup Analyses

Given racial disparities in both diagnoses of uterine fibroids and morbidity and mortality due to ASCVD, stratified analyses by race and ethnicity (White, Black, Hispanic, Asian) were conducted to examine whether the fibroid‐ASCVD association is consistent across racial groups or whether social determinants of health might modify this relationship, and to ensure our findings are generalizable across diverse populations. The cohort was restricted for each racial and ethnic group and matching, weighting, and analyses were repeated within the subsample.

The incidence of both uterine fibroids and ASCVD are directly influenced by age. To examine whether effect measure modification exists by age at fibroid diagnosis, we conducted subgroup analyses stratifying the cohort into age <40 years and age ≥40 years. Similarly, the cohort was restricted for each age subgroup and matching, weighting, and analyses were repeated within the subsample.

Sensitivity Analyses

We conducted additional analyses to account for potential errors in how uterine fibroids and obesity were recorded in the claims data, since these conditions may be underdocumented or misclassified in administrative records. ⁶ , ²¹ , ²⁵ , ²⁶ In brief, we applied probabilistic bias analyses to quantify the direction and magnitude of bias due to measurement errors in our exposure (uterine fibroids) and key confounder (obesity). These methods adjust our observed results based on how accurately we can identify these conditions in the data, accounting for both missed cases and false positives. Additional details are provided in Appendix S2.

Furthermore, subclinical ASCVD may be unmeasurable in claims data, but can lead to uterine fibroids, ²⁷ , ²⁸ , ²⁹ because risk factors have been associated with development. This could be concerning since ASCVD may predate the diagnosis of uterine fibroids. To evaluate the potential for confounding by ASCVD risk factors, we repeated the primary analysis, excluding individuals who had any of the following at baseline: diabetes, hypertension, hyperlipidemia, obstructive sleep apnea, obesity, or prior prescriptions for anti‐diabetic medications, anti‐hypertensive medications, or statins. We also conducted a sensitivity analysis where we included additional covariates, specifically binary indicators of renal insufficiency, HIV status, rheumatoid arthritis, and use of antiplatelet or anticoagulant medications.

To assess unmeasured confounding, we calculated E‐values for each estimate of the primary outcome of interest. E‐values are defined as the minimum strength of association, on the RR scale, that an unmeasured confounder would need to have with both uterine fibroids and ASCVD to fully explain away a specific exposure‐outcome association, conditional on the measured covariates. ³⁰ , ³¹

Analyses were performed in SAS (SAS Institute) and R (RStudio, Inc.).

Patient and Public Involvement

No patients participated in the initial design and implementation of the study because all data were fully deidentified. To comply with data use agreements, investigators were prohibited from reidentifying and contacting patients included in the study data set to share research results. Public dissemination will occur through conference presentations, press releases, and plain language summaries shared on social media platforms.

RESULTS

Baseline Characteristics

Overall, 450 177 individuals with uterine fibroids were matched to 2 250 885 controls (Figure 1), with an average age at baseline of 40.93 years (SD 6.27). Average follow‐up time was 5.18 years in the fibroid group (SD 5.0) and 4.02 years (SD 4.6) in the control group (Table 1). Before weighting, the proportion of Black individuals and the prevalence of ASCVD risk factors were higher in the fibroid group. After weighting, the standardized mean differences between groups were <0.1 for demographic and clinical variables. Additional characteristics are displayed in Table S2.

Figure 1. Flowchart of the cohort selection process, Optum Clinformatics DataMart (2000–2022).

ASCVD indicates atherosclerotic cardiovascular disease. *Exclusions are not mutually exclusive.

Table 1.

Baseline Characteristics of Individuals With Uterine Fibroids and Age‐Matched Controls Before and After Weighting, N=2 701 062, Optum (2000–2022)

	Before IPTW			After IPTW*
	Overall	No uterine fibroid group	Uterine fibroid group	No uterine fibroid group	Uterine fibroid group	SMD
Total	2 701 062	2 250 885	450 177
Follow‐up, y, mean (SD)	4.22 (4.7)	4.02 (4.6)	5.18 (5.0)	4.02 (4.6)	5.18 (5.0)	0.240
Age, mean (SD)	40.93 (6.3)	40.93 (6.3)	40.94 (6.3)	40.93 (6.3)	40.95 (6.3)	0.003
Race and ethnicity						0.004
Asian	131 278 (4.9)	105 877 (4.7)	25 401 (5.6)	109 474.9 (4.9)	21 915.9 (4.9)
Black	284 153 (10.5)	190 504 (8.5)	93 649 (20.8)	237 297.9 (10.5)	47 614.5 (10.6)
Hispanic	288 818 (10.7)	230 490 (10.2)	58 328 (13.0)	24 0849.3 (10.7)	48246.5 (10.7)
White	1 640 407 (60.7)	1 410 855 (62.7)	229 552 (51.0)	1 366 701.1 (60.7)	27 1926.2 (60.5)
Missing	356 406 (13.2)	313 159 (13.9)	43 247 (9.6)	297 042.8 (13.2)	59 777.3 (13.3)
Cardiovascular disease risk factors
Obesity	106 844 (4.0)	86 394 (3.8)	20 450 (4.5)	89 484.4 (4.0)	18 718.7 (4.2)	0.010
Obstructive sleep apnea	13 900 (0.5)	10 836 (0.5)	3064 (0.7)	11 441.9 (0.5)	2586.4 (0.6)	0.009
Diabetes	235 321 (8.7)	186 397 (8.3)	48 924 (10.9)	196 696.1 (8.7)	39920.0 (8.9)	0.005
Hyperlipidemia	162 395 (6.0)	129 510 (5.8)	32 885 (7.3)	135 642.4 (6.0)	27 448.8 (6.1)	0.003
Hypertension	192 385 (7.1)	156 056 (6.9)	36 329 (8.1)	160 987.5 (7.2)	33 356.4 (7.4)	0.01
Smoking	59 891 (2.2)	50 427 (2.2)	9464 (2.1)	50 041.3 (2.2)	10 269.6 (2.3)	0.004
Mental health
Depression	492 551 (18.2)	391 594 (17.4)	100 957 (22.4)	411 235.2 (18.3)	83 140.7 (18.5)	0.006
Anxiety	169 579 (6.3)	136 652 (6.1)	32 927 (7.3)	141795.0 (6.3)	29016.1 (6.5)	0.006
Reproductive and obstetric health
Parous	228 796 (8.5)	190 669 (8.5)	38 127 (8.5)	19 0973.6 (8.5)	38 597.8 (8.6)	0.004
History of hypertensive disorder of pregnancy	20 269 (0.8)	16 124 (0.7)	4145 (0.9)	16 957.7 (0.8)	3501.8 (0.8)	0.003
History of gestational diabetes	25 489 (0.9)	20 902 (0.9)	4587 (1.0)	21 332.5 (0.9)	4426.6 (1.0)	0.004
PCOS	170 255 (6.3)	118 218 (5.3)	52 037 (11.6)	142 534.0 (6.3)	28 875.0 (6.4)	0.004
Infertility	7299 (0.3)	5252 (0.2)	2047 (0.5)	5663.5 (0.3)	1760.2 (0.4)	0.025
Use of oral contraceptive pills	621 068 (23.0)	507 917 (22.6)	113 151 (25.1)	516 990.7 (23.0)	101220.6 (22.5)	0.011
Healthcare utilization
Prior annual physical examination	501 683 (18.6)	424 713 (18.9)	76 970 (17.1)	418 440.5 (18.6)	84538.7 (18.8)	0.006

Data are presented as N (%), unless otherwise specified. IPTW indicates inverse probability of treatment weights; PCOS, polycystic ovary syndrome, and SMD, standardized mean differences.

^*IPTWs for exposure to fibroids were calculated using models adjusting for race and ethnicity, age (linear+quadratic terms) at entry and baseline values for smoking, obesity, diabetes or medication use, hypertension or medication use, hyperlipidemia or statin use, parity, history of hypertensive disorder of pregnancy, history of gestational diabetes, polycystic ovary syndrome, infertility or medication use, oral contraceptive use, depression or medication use, anxiety or medication use, cancer, and recent annual physical examination.

ASCVD Outcomes

There were 6.45 ASCVD events per 1000 person‐years in the uterine fibroid group, compared with 2.99 events per 1000 person‐years in the matched controls. Adjusted risk of ASCVD and individual components by group are presented in Table 2 (Table S3 displays unadjusted results). The 1‐ and 10‐year risks of ASCVD were 0.74% (95% CI, 0.71–0.77) and 5.42% (95% CI, 5.18–5.67) in the fibroid group and 0.30% (95% CI, 0.29–0.31) and 3.00% (95% CI, 2.90–3.11) in the matched controls. Individuals with uterine fibroids had a 0.41 percentage point (aRD, 95% CI, 0.40–0.47) higher adjusted risk and 2.47‐times (aRR, 95% CI, 2.32–2.61) the risk of experiencing an ASCVD event within 1 year compared with matched controls. RDs, measured in percentage points, increased over time (3‐year aRD, 1.05 [95% CI, 1.00–1.15]; 5‐year aRD, 1.55 [95% CI, 1.44–1.70]; 10‐year aRD, 2.40 [95% CI, 2.07–2.77]); while the RRs decreased over time (3‐year aRR, 2.32 [95% CI, 2.19–2.44]); 5‐year aRR, 2.18 [95% CI, 2.05–2.30]; 10‐year aRR,1.81 [95% CI, 1.66–1.96]). Figure 2A displays cumulative incidence curves of ASCVD by group.

Table 2.

Adjusted risks of ASCVD and Individual Components by Uterine Fibroid Status, Optum Clinformatics DataMart (2000–2022)

	Events/Person‐y: uterine fibroids	aRisk % (95% CI): Uterine fibroids	aRD%* (95% CI)	aRR* (95% CI)
	Events/person‐y: no uterine fibroids	aRisk % (95% CI): No uterine fibroids
Outcome: ASCVD
1 y	3244/426 900	0.74 (0.71–0.77)	0.41 (0.40–0.47)	2.47 (2.32–2.61)
	5601/1 934 900	0.30 (0.29–0.31)
3 y	6649/997 131	1.90 (1.84–1.96)	1.05 (1.00–1.15)	2.32 (2.19–2.44)
	11 508/4 237 714	0.82 (0.80–0.84)
5 y	9054/1 392 552	2.91 (2.81–3.01)	1.55 (1.44–1.70)	2.18 (2.05–2.30)
	15 671/5 702 540	1.34 (1.31–1.37)
10 y	12 663/1 962 872	5.42 (5.18–5.67)	2.40 (2.07–2.77)	1.81 (1.66–1.96)
	23 020/7 703 405	3.00 (2.90–3.11)
Outcome: coronary artery disease
1 y	1352/427 960	0.32 (0.31–0.34)	0.15 (0.14–0.19)	2.22 (2.02–2.42)
	2556/1 936 403	0.14 (0.14–0.14)
3 y	2720/1 002 890	0.84 (0.81–0.88)	0.38 (0.34–0.44)	2.03 (1.86–2.19)
	5350/4 244 272	0.40 (0.38–0.41)
5 y	3765/1 404 340	1.33 (1.28–1.39)	0.55 (0.47–0.64)	1.87 (1.71–2.04)
	7390/5 715 206	0.67 (0.65–0.69)
10 y	5592/1 989 079	2.91 (2.78–3.05)	0.94 (0.69–1.22)	1.61 (1.42–1.82)
	11 396/7 731 394	1.68 (1.63–1.74)
Outcome: angina
1 y	832/428 245	0.20 (0.19–0.21)	0.12 (0.11–0.14)	2.79 (2.48–3.14)
	1325/1 936 932	0.07 (0.07–0.07)
3 y	1840/1 004 025	0.52 (0.49–0.55)	0.32 (0.29–0.37)	2.73 (2.46–3.03)
	2691/4 246 660	0.19 (0.18–0.20)
5 y	2519/1 406 260	0.81 (0.76–0.86)	0.50 (0.44–0.57)	2.64 (2.36–296)
	3602/5 719 836	0.31 (0.29–0.32)
10 y	3533/1 992 924	1.44 (1.32–1.57)	0.80 (0.64–0.98)	2.28 (1.93–2.67)
	5083/7 742 590	0.63 (0.59–c0.68)
Outcome: myocardial infarction
1 y	351/428521	0.07 (0.07–0.08)	0.03 (0.03–0.05)	1.94 (1.62–2.32)
	714/1 937 274	0.04 (0.04–0.04)
3 y	693/1005738	0.20 (0.18– 0.21)	0.08 (0.06–0.11)	1.77 (1.51–2.06)
	1536/4 248 180	0.11 (0.10– 0.12)
5 y	976/1410011	0.31 (0.28– 0.34)	0.12 (0.08–0.17)	1.65 (1.39–1.94)
	2123/5 722 791	0.19 (0.17– 0.20)
10 y	1419/2 001 960	0.63 (0.55– 0.72)	0.21 (0.08–0.34)	1.50 (1.18–1.89)
	3142/7 749 182	0.42 (0.38– 0.47)
Outcome: cerebrovascular disease
1 y	1176/428 102	0.27 (0.25– 0.28)	0.14 (0.13–0.17)	2.32 (2.11–2.56)
	2107/1 936 620	0.11 (0.11– 0.12)
3 y	2478/1 003 280	0.71 (0.67– 0.75)	0.38 (0.35–0.44)	2.24 (2.05–2.44)
	4407/4 245 346	0.32 (0.31–,0.33)
5 y	3428/1 404 922	1.12 (1.06–,1.18)	0.59 (0.52–0.68)	2.14 (1.94–2.34)
	6110/5 717 389	0.53 (0.51– 0.55)
10 y	4956/1 990 432	2.20 (2.05– 2.37)	0.95 (0.73–1.19)	1.77 (1.55–2.01)
	9249/7 736 600	1.25 (1.18– 1.32)
Outcome: ischemic stroke
1 y	455/428466	0.10 (0.09– 0.11)	0.05 (0.04–0.07)	2.17 (1.85–2.53)
	845/1 937 210	0.05 (0.04– 0.05)
3 y	988/1 005 403	0.28 (0.26– 0.30)	0.14 (0.11–0.17)	2.08 (1.81–2.38)
	1818/4 247 938	0.13 (0.13– 0.14)
5 y	1346/1 409 318	0.44 (0.40– 0.48)	0.21 (0.17–0.27)	1.98 (1.70–2.29)
	2507/5 722 441	0.22 (0.21– 0.24)
10 y	1908/2 000 549	0.83 (0.74– 0.94)	0.33 (0.19–0.47)	1.66 (1.35–2.03)
	3737/7 748 347	0.50 (0.46– 0.55)
Outcome: transient ischemic attack
1 year	679/428 380	0.15 (0.14– 0.17)	0.09 (0.08–0.11)	2.51 (2.20–2.87)
	1137/1 937 055	0.06 (0.06– 0.07)
3 y	1438/1 004 680	0.42 (0.39– 0.45)	0.24 (0.21–0.29)	2.50 (2.22–2.81)
	2312/4 247 285	0.17 (0.16– 0.18)
5 y	1980/1 407 669	0.66 (0.61– 0.70)	0.38 (0.33–0.45)	2.43 (2.14–2.76)
	3147/5 721 125	0.27 (0.26– 0.29)
10 y	2743/1 996 590	1.20 (1.09– 1.33)	0.61 (0.46–0.79)	2.07 (1.72–2.47)
	4531/7 745 438	0.58 (0.54– 0.63)
Outcome: peripheral artery disease
1 y	262/428 562	0.06 (0.06– 0.07)	0.04 (0.03–0.05)	2.52 (2.04–3.09)
	483/1 937 347	0.03 (0.02– 0.03)
3 y	594/1 005 838	0.17 (0.15– 0.19)	0.09 (0.07–0.12)	2.33 (1.95–2.76)
	1012/4 248 532	0.07 (0.07– 0.08)
5 y	824/1 410 141	0.27 (0.24– 0.30)	0.14 (0.11–0.19)	2.17 (1.79–2.61)
	1456/5 723 413	0.12 (0.11– 0.13)
10 y	1223/2 002 167	0.58 (0.50– 0.68)	0.27 (0.15–0.41)	1.90 (1.45–2.50)
	2257/7 750 342	0.31 (0.27– 0.34)

aRD indicates adjusted risk difference; aRisk, adjusted risk; aRR, adjusted risk ratio; and ASCVD, atherosclerotic cardiovascular disease.

^*Adjusted models were weighted for exposure to fibroids by race and ethnicity, age (linear+quadratic terms) at entry and baseline values for smoking, obesity, diabetes or medication use, hypertension or medication use, hyperlipidemia or statin use, parity, history of hypertensive disorder of pregnancy, history of gestational diabetes, polycystic ovary syndrome, infertility or medication use, oral contraceptive use, depression or medication use, anxiety or medication use, cancer, and recent annual physical examination. Time‐varying weights were used to adjust for disenrollment and fibroid diagnosis in control group using the same baseline variables and monthly updates of these variables.

Figure 2. Cumulative Incidence curves for ASCVD and subcomponents by uterine fibroid status, Optum Clinformatics DataMart (2000‐2022).

Models were weighted for exposure to fibroids by race and ethnicity, age (linear+quadratic terms) at entry and baseline values for smoking, obesity, diabetes or medication use, hypertension or medication use, hyperlipidemia or statin use, parity, history of hypertensive disorder of pregnancy, history of gestational diabetes, polycystic ovary syndrome, infertility or medication use, oral contraceptive use, depression or medication use, anxiety or medication use, cancer, and recent annual physical examination. Time‐varying weights were used to adjust for disenrollment and fibroid diagnosis in the control group using the same baseline variables and monthly updates of these variables. ASCVD indicates atherosclerotic cardiovascular disease.

Individuals with uterine fibroids had increased risk of individual ASCVD components compared with matched controls. RDs were largest for coronary artery disease, the most common ASCVD component (1‐year aRD: 0.15% [95% CI, 0.14–0.19]); 10‐year aRD,, 0.94 [95% CI, 0.69–1.22]). RRs were highest for peripheral artery disease, the least common ASCVD component (1‐year aRR, 2.52 [95% CI, 2.04–3.09]); 10‐year aRR, 1.90 [95% CI, 1.45, 2.50]).

Subgroup Analyses

The increased risk of ASCVD associated with uterine fibroids remained consistent across race‐stratified analyses (Table S4). Comparing those with uterine fibroids to matched controls, the aRR of ASCVD was lowest among Black individuals (1‐year: aRR, 2.21 [95% CI, 1.93–2.51]), though the aRD was largest. Risk was greatest, though the most imprecise, among Asian individuals (1‐year: aRR, 3.12 [95% CI, 2.19–4.39]) at all time points except for 10 years when the RR was highest among Hispanic individuals (10‐year: aRR, 2.16 [95% CI, 1.74–2.68]).

When stratified by baseline age, the association between uterine fibroids and ASCVD risk was stronger among women aged <40 years compared with those aged ≥40 years across all follow‐up periods (Table S5). At 10 years, the aRD was 4.14% ([95% CI, 3.64%–4.69%]) for women <40 years versus 1.66% ([95% CI, 1.55%, 1.76%]) for women ≥40 years, with corresponding aRR of 3.51 ([95% CI, 2.90–4.20]) and 1.58 ([95% CI, 1.43–1.73]), respectively.

Sensitivity Analyses

Adjusting for exposure misclassification of uterine fibroids yielded a bias‐adjusted estimate (1‐year aRR, 3.27) further from the null than the observed estimate (1‐year aRR, 2.47) (Figure 3). When adjusting for confounder misclassification, the estimate was closer to the null (1‐year aRR, 1.80). When accounting for both exposure and confounder misclassification, the 1‐year aRR estimate was 1.89, which was attenuated from the observed estimate (1‐year aRR, 2.47). Overall, results suggest that the bias caused by misclassification was slightly away from the null. This was consistent for the 3, 5, and 10‐year RRs.

Figure 3. Probabilistic bias analyses using 100 000 simulations were conducted for misclassification of the exposure and obesity; 2.5th percentile and 97.5th percentile estimates are reported from the simulations.

Adjusted models were weighted for exposure to fibroids by race and ethnicity, age (linear+quadratic terms) at entry and baseline values for smoking, obesity, diabetes or medication use, hypertension or medication use, hyperlipidemia or statin use, parity, history of hypertensive disorder of pregnancy, history of gestational diabetes, polycystic ovary syndrome, infertility or medication use, oral contraceptive use, depression or medication use, anxiety or medication use, cancer, and recent annual physical examination. Time‐varying weights were used to adjust for disenrollment and fibroid diagnosis in the control group using the same baseline variables and monthly updates of these variables. Confounder misclassification assumed distributions: 𝑠𝑒𝑛𝑠𝑖𝑡𝑖𝑣𝑖𝑡𝑦~(α=24.61, 𝛽=11.85); 𝑠𝑝𝑒𝑐𝑖𝑓𝑖𝑐𝑖𝑡𝑦~𝑏𝑒𝑡𝑎(α=1217.87, 𝛽=42.86) Exposure misclassification assumed distributions: 𝑠𝑒𝑛𝑠𝑖𝑡𝑖𝑣𝑖𝑡𝑦~𝑏𝑒𝑡𝑎(α=12.10, 𝛽=48.41), 𝑠𝑝𝑒𝑐𝑖𝑓𝑖𝑐𝑖𝑡𝑦~𝑏𝑒𝑡𝑎(α=239.52, 𝛽=1.69).

Excluding individuals with cardiovascular risk factors at baseline did not meaningfully change estimates for the primary ASCVD composite outcome nor the individual components (Table S6). Results did not change appreciably when considering additional covariates (Table S7).

For the primary outcome of ASCVD, E‐values ranged from 3.02 to 4.69, indicating that an unmeasured confounder would need to increase the risk of both uterine fibroids and ASCVD by 3.02 to 4.69‐fold to fully explain away the observed associations (Table S8).

DISCUSSION

In this US‐based cohort study leveraging administrative claims data, individuals with uterine fibroids had elevated ASCVD risk compared with matched controls. Results remained consistent after weighting for confounding and selection biases and conducting sensitivity analyses addressing measurement error.

Previous studies found similar associations but were limited by cross‐sectional design and statistical power. Brewster et al., using the National Health and Nutrition Examination Survey of 5522 individuals, found increased odds of cardiovascular disease (defined as ischemic heart disease, heart failure, or stroke) among those with self‐reported fibroids. ¹⁴ Similarly, 2 cross‐sectional studies in the Netherlands and Suriname reported higher ASCVD rates among women with fibroids, though small sample sizes (N=1342 and 728) likely limited statistical significance, ¹² , ¹³ and reliance on self‐report limits the precision of these findings. Laughlin‐Tommaso et al. analyzed the Coronary Artery Risk Development in Young Adults (CARDIA) study among 972 women aged 35–49 years. ¹¹ While ASCVD risk factors such as coronary artery calcification were higher among individuals with ultrasound‐confirmed fibroids after 1 year, these differences were not statistically significant after adjusting for confounders, including blood pressure and body mass index, likely due to limited power for rare outcomes. Although blood pressure and body mass index were not available in the current study, we adjusted for ASCVD risk factors using both diagnoses and medications, though these represent more severe conditions that resulted in medical visits and may not fully capture the risk that would be detected through routine body mass index and blood pressure measurements. The increased ASCVD risk associated with fibroids remained even after accounting for these factors.

Overcoming these limitations, our study used real‐world data from 450 177 individuals in the United States, providing sufficient sample size to investigate uterine fibroids and ASCVD. Using Optum CDM, cases were age‐ and date‐matched to controls, enabling adjustment for confounders such as medication use and reproductive health factors. Importantly, our results reflect associations among women with fibroids severe enough to warrant medical attention and health care encounters, rather than all fibroids including asymptomatic cases. Among this large population, the 1‐ and 10‐year risks of ASCVD among individuals with uterine fibroids were 2.47 and 1.81‐times, respectively, the risks of the matched controls. We used causal inference methods to enhance our findings. Inverse probability weighting was used to ensure sufficient balance and conditional exchangeability on selected, measured characteristics across the groups. We assumed consistency by requiring a well‐defined exposure group using ICD 9/10 codes incorporating claims for uterine fibroids and associated diagnostic procedures. The positivity assumption was empirically examined by distributions of included individuals within each level of the measured values of each variable in the data, as well as sufficient overlap after weighting. Though these assumptions are untestable, ³² they were considered through careful study design and diagnostic checks. Additional weighting methods addressed potential selection bias, and probabilistic bias analyses accounted for measurement error. This comprehensive data set and rigorous methodology offer stronger evidence of the association between uterine fibroids and ASCVD.

While the definitive causal mechanisms linking fibroids and ASCVD require further investigation, the shared pathophysiologic processes involving smooth muscle proliferation, fibrosis, calcification, and systemic inflammation provide a plausible biologic foundation for the observed association. ⁷ Fibroids may overproduce inflammatory mediators, ⁶ , ⁷ which can enter systemic circulation, promoting chronic inflammation, endothelial dysfunction, oxidative stress, and pro‐atherosclerotic pathways, ultimately increasing ASCVD risk. ⁷ , ⁸ , ⁹ , ¹⁰ The ASCVD risk estimator, published by the American College of Cardiology, is commonly used in clinical practice to assess a patient’s 10‐year ASCVD risk, with a 7.5% threshold guiding a risk discussion about prevention therapies. ³³ , ³⁴ The calculator factors in age, sex, race, blood pressure, cholesterol, diabetes, smoking status, and antihypertensive, statin, and aspirin therapy. While previous studies have tested whether female‐specific risk factors, such as preeclampsia (a known ASCVD risk factor) improve prediction models, results suggest they do not. ³⁵ Nonetheless, understanding the cardiovascular implications of conditions such as preeclampsia, polycystic ovary syndrome, and, as suggested by this study, uterine fibroids, is essential for personalized counseling and shared decision‐making between patients and providers using sex‐specific risk factors. However, the relationship between fibroids and ASCVD needs to be validated in other populations and data sets before considering it as a risk‐enhancing factor for ASCVD. Additionally, future research should examine the impact of fibroid removal procedures, such as hysterectomy, myomectomy, on cardiovascular outcomes to determine whether surgical intervention modifies the association between fibroids and ASCVD risk.

Limitations

First, unmeasured confounding cannot be ruled out despite adjusting for key clinical and demographic factors. Parity is limited by pregnancies during coverage, and subclinical ASCVD undetectable in claims could contribute to fibroid development through inflammation, creating spurious associations. Limited laboratory data precluded accounting for lipid levels, hemoglobin A1C, or inflammatory markers, ⁷ though sensitivity analyses excluding individuals with baseline cardiovascular risk factors showed consistent results. Furthermore, E‐values calculated ranged from 3.02 to 4.69, though unmeasured confounding and other limitations may still influence the findings. Similarly, residual selection bias or surveillance bias may persist if unmeasured factors related to health care‐seeking behavior, access to care, or underlying health conditions influenced both the likelihood of fibroid diagnosis and subsequent ASCVD risk in ways not captured by our measured covariates. Second, claims data susceptibility to measurement errors was addressed through probabilistic bias analyses for underreporting of fibroids and obesity, which slightly attenuated but still strongly supported increased ASCVD risk. While we required 6 months of pre‐baseline data, some individuals may have had prior fibroids, and retrospective data may have billing delays, though these are expected to be minimal. Third, the commercially insured US population limits generalizability to those with public or no insurance. This represents an important limitation because health care access, utilization patterns, and outcomes may differ significantly across insurance types and socioeconomic strata. Fourth, while we lacked mortality data, low death rates (12.9–17.9/100 000) among similar‐aged female individuals (35–54 years) minimize concerns about bias from death as a competing event. ³⁶ Fifth, fibroid pathology details (location, size, and number) were unavailable, with 78.8% having unspecified location. We addressed this by examining symptom‐based differences, with consistent results across classification timeframes despite possible misclassification. Finally, we did not adjust for surgical fibroid removal to avoid selection bias, because treatment represents a downstream effect.

CONCLUSIONS

In this US‐based cohort study using claims data, individuals with uterine fibroids had a higher ASCVD risk compared with those without fibroids. Results did not meaningfully change after adjusting for confounding, selection bias, and measurement error. Findings highlight the need for targeted ASCVD prevention among individuals with fibroids.

Sources of Funding

This work was supported through Patient‐Centered Outcomes Research Institute (PCORI) BPS‐2022C3‐30268 and National Heart, Lung, and Blood Institute (NHLBI) F31HL182358‐01.

Disclosures

None.

Disclaimer

All statements in this report, including its findings and conclusions, are solely those of the authors and do not necessarily represent the views of the Patient‐Centered Outcomes Research Institute (PCORI), its Board of Governors or Methodology Committee.

Supporting information

Appendices S1–S2

Tables S1–S8

Figure S1