Co-occurrence of endometriosis and uterine fibroids: a systematic review and meta-analysis

Summary

Background

Endometriosis and uterine fibroids are estrogen-dependent gynecological disorders with an increasing burden to women’s health worldwide. Despite overlapping symptoms and long-term consequences, the magnitude of their co-occurrence remains unclear.

Methods

We systematically searched PubMed, Embase, and Scopus up to July 21st, 2025, for studies reporting data on fibroids occurrence in patients with and without endometriosis. Effect estimates (adjusted or, if unavailable, crude) were pooled using random-effects models. Age and body mass index (BMI) were explored in uni- and multivariate meta-regressions. Subgroup analyses considered study design, data source, geographic location, parity, age and study quality. Sensitivity analysis included only asymptomatic controls without endometriosis. Study quality and certainty of evidence were assessed using the Newcastle–Ottawa Scale and Grading of Recommendations Assessment, Development, and Evaluation guidelines, respectively. The protocol was registered on PROSPERO (CRD42024614711).

Findings

4351 articles were identified, of which 23 met eligibility criteria, including 3,109,231 patients (118,716 with endometriosis and 2,990,515 controls). Endometriosis was associated with higher odds of fibroids (pooled OR 2.91; 95% CI, 1.78–4.75). BMI influenced estimates (p < 0.01). Between-group differences in data source (p < 0.01) and study quality (p = 0.03) were observed. Parity showed a trend effect: OR 1.56 (95% CI: 0.53–4.59) for studies with >50% nulliparous women, 4.52 (95% CI: 2.53–8.10) for those with >50% multiparous, 5.23 (95% CI, 2.09–13.04) for studies with >95% multiparous women. In sensitivity analysis with only asymptomatic controls (n = 3,103,146), the OR increased to 7.01 (95% CI: 4.72–10.41).

Interpretation

This first meta-analysis on the co-occurrence of endometriosis and fibroids shows that women with endometriosis have a three-to tenfold higher odds of fibroids compared with controls. Clinical awareness of this association is crucial to optimize both patient management and short- and long-term women’s health outcomes.

Funding

This study was funded by the Italian Ministry of Health–Current research IRCCS.

Research in context.

Evidence before this study

Endometriosis and uterine fibroids are common estrogen-dependent gynecological disorders that substantially affect women’s health worldwide. Despite overlapping symptoms and potential long-term consequences, the extent of their co-occurrence remains unclear, and no prior meta-analysis had addressed this association. We systematically searched PubMed, Embase, and Scopus from inception to July 21st, 2025, for studies reporting fibroid occurrence in women with and without endometriosis, using controlled vocabulary and free-text terms for both conditions. Effect estimates were pooled with random-effects models, and meta-regression explored the influence of age and body mass index. Prespecified subgroup analyses accounted for study design, data source, geographic location, parity, age, and study quality. Study quality and certainty of evidence were assessed using the Newcastle–Ottawa Scale and the GRADE framework. The protocol was registered on PROSPERO (CRD42024614711).

Added value of this study

This first meta-analysis on endometriosis and fibroids, encompassing 23 studies and over three million women, found nearly threefold higher odds of fibroids in those with endometriosis. The association was consistent across sensitivity analyses, stronger in high-quality studies, in populations with more multiparous women, and when compared with asymptomatic controls from the general population. Body mass index was positively associated with effect size, while age had minimal influence except in analyses restricted to asymptomatic controls. Despite high heterogeneity and low certainty of evidence, the consistent direction and magnitude across diverse settings strengthen the epidemiological link between these conditions.

Implications of all the available evidence

Women with endometriosis face an increased risk of fibroids compared to controls without endometriosis. Given their overlapping symptoms and the possibility that co-occurrence may reduce treatment effectiveness for endometriosis, heightened clinical awareness is essential to improve diagnostic accuracy, optimize management, and enhance patient satisfaction.

Introduction

Endometriosis is a chronic, estrogen-dependent disease affecting 5–10% of women of reproductive age¹ and rising to nearly 30% among women with infertility.²

Years of research on endometriosis increasingly highlight that fully understanding its pathogenesis, symptoms, and long-term outcomes requires accounting for concomitant factors and comorbidities that significantly shape individual prognosis.³ Notably, endometriosis frequently coexists with other uterine pathologies, including adenomyosis, fibroids, and obstructive Müllerian anomalies.^4,5 Uterine comorbidities may be associated with impaired uterine function, which in turn is linked to endometriosis through retrograde menstruation and translocation of endometrial tissue into the pelvic cavity.^6,7 Moreover, uterine disorders have been independently associated with infertility,⁸ and their coexistence could negatively impact the reproductive prognosis of women with endometriosis. Crucially, uterine comorbidities are associated with menstrual symptoms and changes in menstrual patterns, often leading to heavy menstrual bleeding,⁹ a frequent clinical presentation in endometriosis. According to the International Federation of Gynecology and Obstetrics (FIGO) classification, structural causes of abnormal uterine bleeding, summarized by the PALM acronym, include polyps, adenomyosis, and fibroids.¹⁰ Therefore, as abnormal menstrual symptoms are common in endometriosis, evaluating uterine disorders is critically important.

Among uterine diseases, uterine fibroids (also known as leiomyomas) are the most prevalent benign uterine tumors, affecting up to 25% of women and representing a major contributor to gynecological morbidity.¹¹ Their occurrence peaks between ages 35 and 50, with reported increases of 67% and 79% over the past 30 years,¹² likely reflecting improved ascertainment through non-invasive diagnostic methods.

Emerging evidence suggests that women with fibroids have a significantly higher occurrence of gynecological and systemic comorbidities, particularly endometriosis and, in some cases, infertility.¹³ However, the reported association between fibroids and endometriosis remains highly heterogeneous, with recent evidence suggesting that their concurrent occurrence may be underestimated.¹⁴ Clarifying the true extent of this association is crucial for improving diagnosis, optimizing treatment decisions and tailoring patient counselling, especially in cases with non-specific or overlapping menstrual symptoms, infertility, or poor response to first-line treatments.

This meta-analysis aims to provide a comprehensive quantitative assessment of the association between uterine fibroids and endometriosis. By determining whether the presence of endometriosis significantly increases the likelihood of uterine fibroids, we seek to refine an integrated clinical approach that could enhance early detection of co-occurring disease and optimize long-term health outcomes for millions of endometriosis-affected women worldwide.

Methods

This systematic review and meta-analysis was conducted in accordance with the Preferred Reporting Item for Systematic Reviews and Meta-analysis (PRISMA) and the Meta-analysis Of Observational Studies in Epidemiology (MOOSE) guidelines.^15,16 The study protocol was prospectively registered on the publicly accessible database PROSPERO¹⁷ under the registration ID CRD42024614711.

Search strategy and information sources

Two reviewers (A.F. and L.S.) searched the literature up to July 21st, 2025, using PubMed, Embase, and Scopus databases. Disagreements were resolved by consultation with a third reviewer (N.S.). The search algorithm incorporated terms relevant to endometriosis and uterine fibroids. The full search strategy is provided in the Supplementary Material 1.

The literature search was conducted without restrictions on publication year or geographic location. Only peer-reviewed, full-text manuscripts written in English were assessed for eligibility. Bibliographies of relevant studies were reviewed to identify any relevant publication not retrieved in the initial search.

Eligibility criteria and study selection

Studies were included if they provided adjusted or crude estimates of the occurrence of uterine fibroids in patients with and without endometriosis, or if such estimates could be calculated from the reported data.

Exclusion criteria were: i) descriptive studies (case reports, case series) or those without original data (reviews, abstracts, editorials, comments); ii) in vitro or animal studies; iii) studies in which all patients with endometriosis had fibroids, precluding the estimation of true co-occurrence rates and meaningful comparisons with controls; iv) studies where fibroid occurrence could not be distinguished from other uterine conditions (e.g., adenomyosis).

Data extraction

The following data were collected and tabulated: i) first author and year of publication; ii) country, study design, and data source; iii) sample size, and number of cases and controls; iv) characteristics of the study population, categorized as: a) women undergoing surgery for gynecological conditions, b) women presenting with gynecological symptoms, or c) women attending outpatient clinics for routine check-ups without gynecological symptoms or concerns; v) number of women with uterine fibroids in both case and control groups; vi) baseline characteristics of the study population, including age, body mass index (BMI), parity, infertility status, and other available variables.

Quality assessment and certainty of the evidence

The quality of the included studies was assessed by three reviewers (A.F., C.F. and B.M.) using the Newcastle–Ottawa scale (NOS),¹⁸ and any disagreements were resolved by consulting a third reviewer (N.S.). Specifically, the quality of cohort and case-control studies was assessed according to McPheeters et al.,¹⁹ while cross-sectional studies were evaluated according to Blanchard et al.²⁰

The certainty of the evidence was graded according to the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) guidelines²¹ as high, moderate, low, or very low.

Data analysis

Adjusted odds ratios (adjORs) with corresponding 95% confidence intervals (CIs) were extracted whenever available. If adjORs were not reported, raw data on the occurrence of uterine fibroids in women with and without endometriosis were extracted from the original studies and organized into 2 × 2 contingency tables, from which crude odds ratios (ORs) with 95% CIs were calculated. The standard errors (SEs) of both adjORs and ORs were computed accordingly. All effect estimates and SEs were subsequently pooled using a random-effects (RE) meta-analysis model, with between-study variance (τ²) estimated via the restricted maximum likelihood (REML) method.

Between-study heterogeneity was assessed using the Q test²² and the I² statistic,²³ which quantifies the proportion of total variation in effect estimates attributable to heterogeneity rather than sampling error. Heterogeneity was considered low when the Q test p-value was greater than 0.05 and the I² was below 30%.²⁴

The potential impact of publication bias and small-study effects on pooled estimates was assessed through visual inspection of funnel plot asymmetry and quantified using Egger’s linear regression test²⁵ and Begg’s adjusted rank correlation test²⁶ when at least 10 studies were available.

A meta-regression analysis was performed using mean age and mean BMI reported in each individual study as covariates, both in univariate and multivariate models. When means were available only for subgroups, the overall study mean was estimated using the method described by Arian et al.²⁷ When only the median and range were available, the mean was estimated following the approach by Wan et al.²⁸

All statistical analyses were conducted using R (version 4.5.1; R Core Team, 2025,²⁹ employing the packages meta and metafor. The metabin() function was used to compute ORs and SEs from raw data, and metagen() was used to calculate pooled effect estimates. Publication bias was assessed using the metabias(), rma(), and regtest() functions. Meta-regression analyses were carried out with the metareg() command.

Sensitivity and subgroups analyses

The main analysis included all eligible studies. A secondary analysis was restricted to studies enrolling asymptomatic controls without endometriosis, representative of the general population. For both the main and secondary analyses, the same set of sensitivity and subgroup analyses was applied.

As sensitivity analyses, a leave-one-out meta-analysis was conducted to assess the robustness of the pooled estimates. Furthermore, since age is a well-established confounder in epidemiological associations in women’s health,³⁰ and not all studies reported estimates adjusted for age, we conducted two additional sensitivity analyses to account for potential age-related bias: (i) the first excluded studies where there was a significant age difference at baseline between the endometriosis and control groups (‘studies without significantly age difference at baseline’); (ii) the second was more restrictive and included only studies that specifically reported no significant age difference between groups at baseline, or that provided estimates adjusted for age (‘age-matched studies’). Meta-regressions by mean age and BMI were also conducted in these more homogeneous and less biased comparison groups to further explore their effect on the association.

Predefined subgroup analyses were also conducted to explore the potential influence of methodological and clinical confounders on the pooled estimates. Differences between subgroups were assessed using the Q test, with statistical significance defined as p < 0.05. Studies were categorized according to the following criteria: (1) study design: case-control, cohort, or cross-sectional; (2) data source: hospital records (clinical extracted from the medical records of one or a few hospitals, or limited multicenter studies), large-scale databases (data from national surveys, large prospective cohorts, population biobanks, or national clinical/administrative registries), or interviews (primary data collected via participant questionnaires or interviews, with clinical confirmation from hospital records or subsequent diagnoses); (3) geographic location: Asia, Asia Minor, Australia, Europe or North America; (4) parity distribution: studies were classified as mostly nulliparous if >50% of participants were nulliparous, or mostly multiparous if >50% of participants were multiparous; (5) age: studies were classified as mean age ≤35 years if the mean age of the population was ≤35 years or if >50% of participants were aged ≤ 35; or as mean age >35 years if the mean age was >35 years or if >50% of participants were over 35; and (6) study quality: good, fair, or poor (based on NOS score).

Estimates of population parity were based on the reported prevalence of nulliparous and parous women in each study. In studies exclusively enrolling infertile women, or only women with at least one prior delivery, nulliparity and parity were assumed to be 100% and 0%, respectively. When only the median number of births was reported, a median of 0 was interpreted as indicating that ≥50% of participants were nulliparous, while a median ≥1 was interpreted as <50% nulliparous. For multiparity, a stepwise subgrouping strategy was applied: first, studies were classified as mostly multiparous if >50% of participants were multiparous; then, a more restrictive subgroup was defined as those with >95% multiparous participants, to allow for effect modification analysis and to improve internal validity through greater population homogeneity. The >95% threshold was used to reflect practical dominance of the exposure. Studies without parity data were excluded from parity-based analyses.

For age, groups were based on the mean age reported or derived from study data. The cut-off of 35 years was chosen based on clinical rationale (given the increased prevalence of gynecological conditions such as fibroids after age 35³¹) as well as statistical considerations based on the average age across included populations. Studies that did not report or allow estimation of mean age or age distribution were excluded from age-based analyses.

Role of funding source

This study was funded by Italian Ministry of Health—Current research IRCCS. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Results

The systematic search identified 4351 articles, of which 4347 from database and 4 from citation searching. 617 duplicated records were excluded. The remaining 3730 were screened, and 23 studies32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54 that met the inclusion criteria were finally included in the analysis (PRISMA flow chart provided in Fig. 1). Main characteristics of included studies are provided in Table 1.

Fig. 1.

Flow diagram showing the search and selection of studies.

Table 1.

Main characteristics of included studies.

First author (year)	Country	Study design	Data Source: (1) hospital record; (2) database; (3) interview	Characteristics of the study population: (1) women undergoing surgery for gynecological disease; (2) women addressed for gynecological symptoms; (3) women accessing outpatient clinics for routine check-ups, with no gynecological concerns	Average age in study population: mean (SD)	Average BMI in study population: mean (SD)	Fertility and parity prevalence in study population: (1) infertile women (%); (2) nulliparous women (%)	Population size: (1) overall population; (2) endometriosis; (3) fibroids
Fraser I.S. (1990)32	Australia	cohort	(1)	(1) + (2)	34.2 (6.6)	–	(1) 21%; (2) 39%.	(1) 182; (2) 45; (3) 37.
Carter J.E. (1994)33	USA	cohort	(1)	(1) + (2)	39.3 (14.1)	–	(1) –; (2)–.	(1) 141; (2) 113; (3) 60.
Matorras R. (1996)34,c	Spain	case-control	(1)	(1)	29.5 (3.5)	–	(1) 100%; (2) –.	(1) 348; (2) 174; (3) 67.
Hemmings R. (2004)35,c	Canada	cohort	(1)	(1)	37.3 (6.4)	–	(1) –; (2) –.	(1) 2776; (2) 896; (3) 696.
Lee D.W. (2009)36,b	USA	cohort	(2)	(3)	42.1 (8.1)	–	(1) –; (2) –.	(1) 1,293,398; (2) 10,855; (3) 13,263.
Weuve J. (2010)37	USA	cross-sectional	(3)	(3)	38.7 (8.5)	–	(1) –; (2) 27%.	(1) 1227; (2) 87; (3) 151.
Tanmahasamut P. (2014)38,d	Thailand	cross-sectional	(1)	(1)	39.4 (17.4)	23.6 (4.6)	(1) –; (2) <50%.	(1) 331; (2) 101; (3) 243.
Walch K. (2014)39,d	Austria	case-control	(1)	(2)	34.7 (7.8)	24.3 (3.7)	(1) –; (2) –.	(1) 102; (2) 44; (3) 20.
Lin W.-C. (2016)40	Taiwan	cohort	(2)	(3)	34.6 (9.6)	–	(1) 1%; (2) –.	(1) 79,512; (2) 2009; (3) 1141.
Al-Jefout M. (2017)41	Jordan	cross-sectional	(3)	(3)	<30 yrs: 85%a ≥30 yrs: 15%	–	(1) 4%; (2) –.	(1) 1772; (2) 45; (3) 20.
Al-Jefout M. (2018)42,d	United Arab Emirates	cross-sectional	(3)	(3)	<30 yrs: 94%a ≥30 yrs: 6%	–	(1) 1%; (2) –.	(1) 3572; (2) 55; (3) 30.
Dixon-Suen S.C. (2019)43,d	Australia	cohort	(2)	(2)	–	–	(1) –; (2) 44.4%.	(1) 837,942; (2) 48,870; (3) 47,407.
Wu B.-J. (2021)44,d	China	cross-sectional	(3)	(3)	39.5 (6)	–	(1) –; (2) < 5.5% (Parity 0 or 1).	(1) 2200; (2) 29; (3) 440.
Farland L.V. (2022)45	USA	cohort	(2)	(3)	33.1 (4.1)	–	(1) –; (2) 0%.	(1) 91,825; (2) 1560; (3) 4212.
Bean E. (2022)46,b	UK	case-control	(1)	(2)	33.3 (8.8)	29.7 (10.1)	(1) –; (2) 57%.	(1) 1341; (2) 66; (3) 197.
Menzhinskaya I.V. (2023)47,c	Russian Federation	case-control	(1)	(1)	30.4 (4.6)	–	(1) 43%; (2) –.	(1) 101; (2) 74; (3) 25.
Yuk J.-S. (2023)48	Korea	cohort	(2)	(3)	36 (3)	–	(1) –; (2) 80%.	(1) 630,523; (2) 39,349; (3) 190,583.
Xholli A. (2024)49,d	Italy	cross-sectional	(1)	(2)	35.8 (7.3)	22.9 (4.7)	(1) –; (2) –.	(1) 287; (2) 157; (3) 55.
Holdsworth-Carson S.J. (2024)50	Australia	cohort	(1)	(1)	29.5 (7.6)	25.5 (8.4)	(1) –; (2) 80%.	(1) 255; (2) 184; (3) 35.
Russo C. (2024)51	Italy	cohort	(1)	(2)	35.5 (10.7)	22.9 (4.9)	(1) –; (2) 60%.	(1) 221; (2) 70; (3) 48.
Duan Y. (2025)52,d	USA	cross-sectional	(2)	(3)	35.8 (10.1)	28.7 (7.1)	(1) –; (2) –.	(1) 2638; (2) 178; (3) 319.
Tekle H. (2025)53,d	USA/Puerto Rico	cohort	(2)	(3)	54.9 (9)	27.3 (6)	(1) –; (2) 18%.	(1) 41,641; (2) 4499; (3) 11,202.
McGrath I.M. (2025)54	UK/Estonia	cohort	(2)	(3)	–	–	(1) –; (2) –.	(1) 116,896; (2) 9256; (3) 12,070.

SD: standard deviation; BMI: body mass index.

Note: The data presented in the table refer to subgroups of the study population used in the analysis. When available, crude data on the proportion of infertile or nulliparous women, either in the overall cohort or within subgroups, were extracted or calculated. In studies including only infertile women or only women with at least one delivery, infertility and nulliparity were assumed to be 100% and 0%, respectively. In studies reporting only the median parity, a median ≥1 was interpreted as <50% nulliparous, while a median of 0 was interpreted as ≥50% nulliparous.

Mean and SD for age and BMI were either directly extracted or estimated as follows: when reported only for subgroups, when reported only for subgroups, values were pooled and calculated according to Arian et al.²⁷; when reported as median and range, mean and SD were estimated following the method by Wan et al.²⁸

^aAl-Jefout M. (2017) and Al-Jefout M. (2018) reported age only as the percentage of patients younger or older than 30 years. As it was not possible to estimate the mean and SD, these studies were excluded from the meta-regression but included in the age subgroup analysis.

^bReported adjusted odds ratios for the presence of fibroids in patients with vs. without endometriosis.

^cReported no significant baseline differences or included age-matched populations for endometriosis and non-endometriosis groups.

^dReported significant baseline differences between endometriosis and non-endometriosis groups.

According to quality assessment, 16 out of 23 were evaluated as good quality,36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46^{,48,49,51,52,54} 4 as fair quality,^35,47,50,53 and 3 as poor quality.32, 33, 34 Full quality assessment details of the included studies are provided in Supplementary Table S1.

A summary of all meta-analyses for the primary and secondary analyses is provided in Table 2. Results of the univariate and multivariate meta-regressions for the primary and secondary analysis are presented in Table 3.

Table 2.

Summary of all meta-analyses results.

	Number of studies	Population size	OR [95% CI]	I2	τ2	Q-testa	Begg’s testa	Egger’s testa	Q-test for subgroup differencesa
1. Primary analysis: endometriosis vs. controls	23	3,109,231	2.91 [1.78–4.75]	99.5%	1.33	0.00	0.09	0.19
Studies without significant age difference at baseline	15	2,220,518	2.81 [1.62–4.86]	99.4%	1.07	0.00	0.20	0.06
Age-matched studies	5	1,297,964	2.98 [0.95–9.32]	99.6%	1.58	<0.01	n.a.	n.a.
a. Type of study	23	3,109,231							0.17
Cohort	12	3,095,312	3.12 [1.62–6.04]	99.7%	1.30	0.00	0.95	<0.01
Case-control	4	1892	1.43 [0.73–2.81]	66.1%	0.30	0.03	n.a.	n.a.
Cross-sectional	7	12,027	3.79 [1.28–11.19]	94.5%	1.96	<0.01	n.a.	n.a.
b. Data source	23	3,109,231							<0.01
Hospital records	11	6085	1.02 [0.69–1.51]	72.8%	0.30	<0.01	0.53	0.84
Databases	8	3,094,4375	6.26 [3.92–9.98]	99.8%	0.45	0.00	n.a.	n.a.
Interviews	4	8771	9.85 [4.64–20.87]	67.1%	0.36	0.03	n.a.	n.a.
c. Geographic location	23	3,109,231							0.08
Asia	4	712,566	4.30 [1.06–17.48]	98.0%	1.98	<0.01	n.a.	n.a.
Asia Minor	2	5344	10.06 [2.62–38.60]	58.5%	0.57	0.12	n.a.	n.a.
Australia	3	838,379	3.35 [0.96–11.66]	94.7%	1.10	<0.01	n.a.	n.a.
Europe	7	119,296	1.25 [0.56–2.80]	96.3%	1.03	<0.01	n.a.	n.a.
North America	7	1,433,646	3.74 [1.71–8.22]	99.8%	1.09	0.00	n.a.	n.a.
d. Parity	16	1,651,131							0.09
Mostly nulliparous (>50%)	5	632,688	1.56 [0.53–4.59]	97.4%	1.41	<0.01	n.a.	n.a.
Mostly multiparous (>50%)	11	1,060,305	4.52 [2.53–8.10]	99.5%	0.84	0.00	0.16	0.70
Almost all multiparous (>95%)	6	179,229	5.23 [2.09–13.04]	91.1%	1.14	<0.01	n.a.	n.a.
e. Age	21	2,154,393							0.94
Mean age ≤ 35 years	10	179,010	2.68 [1.52–4.71]	87.6%	0.66	<0.01	1	0.94
Mean age > 35 years	11	1,975,383	2.58 [1.09–6.11]	99.7%	2.06	0.00	0.88	0.22
f. Study quality	23	3,109,231							0.03
Good quality	16	3,063,787	3.75 [1.96–7.16]	99.3%	1.64	0.00	0.30	0.23
Fair quality	4	44,773	1.89 [1.18–3.03]	94.6%	0.15	<0.01	n.a.	n.a.
Poor quality	3	671	1.19 [0.68–2.09]	48.6%	0.12	0.14	n.a.	n.a.
2. Sensitivity analysis: endometriosis vs. asymptomatic controls	12	3,103,146	7.01 [4.72–10.41]	99.7%	0.43	0.00	0.30	0.52
Studies without significant age difference at baseline	7	2,215,153	6.64 [4.08–10.82]	99.7%	0.39	<0.01	n.a.	n.a.
Age-matched studies	1	1,293,398	22.22 [21.09–23.41]	n.a.	n.a.	n.a.	n.a.	n.a.
a. Type of study	12	3,103,146							0.61
Cohort	7	3,091,737	6.50 [3.83–11.02]	99.8%	0.51	0.00	n.a.	n.a.
Case-control	0	0	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.
Cross-sectional	5	11,409	8.03 [4.27–15.12]	70.29	0.35	0.01	n.a.	n.a.
b. Data source	12	3,103,146							0.32
Hospital records	0	0	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.
Databases	8	3,094,375	6.26 [3.92–9.98]	99.8%	0.45	0.00	n.a.	n.a.
Interviews	4	8771	9.85 [4.64–20.87]	67.1%	0.36	0.03	n.a.	n.a.
c. Geographic location	12	3,103,146							<0.01
Asia	3	203,793	7.56 [3.05–19.18]	96.7%	0.59	<0.01	n.a.	n.a.
Asia Minor	2	5344	10.06 [2.62–38.60]	58.5%	0.57	0.12	n.a.	n.a.
Australia	1	837,942	10.39 [10.16–10.62]	n.a.	n.a.	n.a.	n.a.	n.a.
Europe	1	116,896	5.43 [5.17–5.7]	n.a.	n.a.	n.a.	n.a.	n.a.
North America	5	1,430,729	6.03 [2.98–12.2]	99.8%	0.63	0.00	n.a.	n.a.
d. Parity	9	1,690,214							0.31
Mostly nulliparous (>50%)	1	630,523	8.16 [7.97–8.36]	n.a.	n.a.	n.a.	n.a.	n.a.
Mostly multiparous (>50%)	8	1,059,691	6.31 [3.84–10.37]	99.6%	0.42	0.00	n.a.	n.a.
Almost all multiparous (>95%)	5	178,881	7.27 [3.52–15.02]	83.1%	0.52	<0.01	n.a.	n.a.
e. Age	10	2,148,308							0.52
Mean age ≤ 35 years	4	176,681	5.67 [2.93–10.99]	79.8%	0.33	<0.01	n.a.	n.a.
Mean age > 35 years	6	1,971,627	7.71 [3.95–15.04]	99.8%	0.65	0.00	n.a.	n.a.
f. Study quality	12	3,103,146							<0.01
Good quality	11	3,061,505	7.72 [5.26–11.33]	99.5%	0.36	0.00	0.21	0.72
Fair quality	1	4,1641	2.68 [2.52–2.86]	n.a.	n.a.	n.a.	n.a.	n.a.
Poor quality	0	0	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.

OR: odds ratio; CI: confidence interval; n.a.: not applicable.

Note: Not unbalanced age studies excluded studies with significant baseline age differences; age-matched studies included only studies with no significant baseline age difference or age-adjusted estimates. Studies were classified as mostly nulliparous or mostly multiparous if >50% of participants were nulliparous or multiparous, respectively. A >95% threshold defined the almost all multiparous group.

Age groups were defined based on the mean age reported or calculated from individual study data, with studies categorized as mean age ≤ 35 years when the reported or estimated mean age of the population was 35 years or younger and as mean age > 35 years when the reported or estimated mean age of the population was older than 35 years.

^ap-value of the corresponding statistical test.

Table 3.

Summary of univariate and multivariate meta-regression results.

	Univariate analysis						Multivariate analysis
	Number of studies	Age’s coefficient [95% CI]	p-value	Number of studies	BMI’s coefficient [95% CI]	p-value	Number of studies	Age’s coefficient [95% CI]	p-value	BMI’s coefficient [95% CI]	p-value
1. Primary analysis: endometriosis vs. controls	19	0.04 [−0.05 to 0.14]	0.38	8	0.34 [0.23–0.45]	<0.01	8	0.02 [−0.02 to 0.05]	0.36	0.33 [0.21 to 0.45]	<0.01
Studies without significant age difference at baseline	13	0.12 [−0.04 to 0.27]	0.15	3	0.28 [0.07–0.50]	0.01	3	n.a.	n.a.	n.a.	n.a.
Age matched studies	5	0.18 [−0.01 to 0.37]	0.06	1	n.a.	n.a.	1	n.a.	n.a.	n.a.	n.a.
2. Sensitivity analysis: endometriosis vs. asymptomatic controls	8	−0.02 [−0.10 to 0.07]	0.71	2	n.a.	n.a.	2	n.a.	n.a.	n.a.	n.a.
Studies without significant age difference at baseline	5	0.17 [0.05–0.28]	<0.01	0	n.a.	n.a.	0	n.a.	n.a.	n.a.	n.a.
Age matched studies	1	n.a.	n.a.	0	n.a.	n.a.	0	n.a.	n.a.	n.a.	n.a.

Note: Univariate analysis was performed using either mean age or mean BMI as a covariate. Multivariate analysis was performed including both mean age and mean BMI as covariates.

CI: confidence interval; BMI: body mass index; n.a.: not applicable.

Primary analysis: occurrence of uterine fibroids in endometriosis vs. controls

Studies overview

A total of 23 studies were included, encompassing 3,109,231women (118,716 with endometriosis and 2,990,515 controls). Twelve were cohort studies (3,095,312 women),^{32,33,35,36,40,43,45,48,50,51,53,54} 4 case-control (1892 women),^34,39,46,47 and 7 were cross-sectional studies (12,027 women).^{37,38,41,42,44,49,52} Eleven studies were based on hospital records (6085 women),32, 33, 34, 35^{,38,39,46,47,}49, 50, 51 8 on databases (3,094,4375 women),^{36,40,43,45,48,}52, 53, 54 and 4 by interviews (8771 women).^37,41,42,44 The geographic location of included studies included a variety of countries, specifically: Asia (4 studies, 712,566 women),^38,40,44,48 Asia Minor (2 studies, 5344 women),^41,42 Australia (3 studies, 838,379 women),^32,43,50 Europe (7 studies, 119,296 women),^{34,39,46,47,49,51,54} and North America (7 studies, 1,433,646women).^33,35, 36, 37^,45,52,53 Sixteen studies reported sufficient data to allow subgroup analysis by parity.^32,34,37,38,40, 41, 42, 43, 44, 45, 46, 47, 48^,50,51,53 Among these, 4 included populations in which more than 50% of women were nulliparous (n = 632,688).^{34,46,48,50,51} The remaining 11 studies included populations where more than 50% had at least one previous pregnancy (1,060,305 women),^32,37,38,40, 41, 42, 43, 44, 45^,47,53 including six studies in which more than 95% of participants were multiparous (179,229 women).40, 41, 42^,44,45 All but two studies reported age data or sufficient information to categorize the study population based on mean age.32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42^,44, 45, 46, 47, 48, 49, 50, 51, 52, 53 Specifically, 10 studies included populations with a reported or estimated mean age of the population ≤35 years (179,010 women),^32,34,39, 40, 41, 42^,45, 46, 47^,50 while 11 studies included populations with a mean age >35 years (1,975,383 women).^33,35, 36, 37, 38^,44,48,49,51, 52, 53

Main results and certainty of the evidence

The pooled OR of uterine fibroids in women with endometriosis compared to those without was 2.91 (95% CI: 1.78–4.75; I² = 99.5%; Fig. 2), with no evidence of significant publication based on Egger’s test (p = 0.19) and Begg’s test (p = 0.09). The certainty of evidence was judged as low according to the GRADE framework (Supplementary Table S2).

Fig. 2.

Forest plot of all included studies.

Leave-one-out sensitivity analysis confirmed the robustness of this association (Supplementary Fig. S1). In sensitivity analyses restricted to studies without significant age imbalance at baseline (2,220,518 patients: 64,783 with endometriosis and 2,155,735 controls),32, 33, 34, 35, 36, 37^,40,41,45, 46, 47, 48^,50,51,54 the association remained consistent. A further restriction to studies matched for age or reporting age-adjusted estimates (1,297,964 patients: 12,065 with endometriosis and 1,285,899 controls)34, 35, 36^,46,47 yielded similar pooled results, although with confidence intervals including the null (Supplementary Figs. S2–S4).

Meta-regression analyses showed no significant effect of age on the association between endometriosis and fibroids in either the primary or sensitivity analyses. Conversely, BMI was consistently associated with effect estimates, both in univariate and multivariate models (coefficient at multivariate analysis = 0.33; 95% CI: 0.21–0.45; p < 0.01; Table 3).

Subgroup analyses

Subgroup analyses are summarized in Table 2. Corresponding forest and funnel plots are presented in Supplementary Figs. S5–S17.

Statistically significant differences between subgroups were observed for data source (p < 0.01) and study quality (p = 0.03). In particular, studies based on hospital records showed the lowest association between endometriosis and fibroids (OR 1.02; 95% CI: 0.69–1.51), whereas large-scale database studies reported substantially higher estimates (OR 6.26; 95% CI: 3.92–9.98), followed by interview-based studies (OR 9.85; 95% CI: 4.64–20.87).

Regarding study quality, good-quality studies yielded the strongest association (OR 3.75; 95% CI: 1.96–7.16), followed by fair-quality studies (OR 1.89; 95% CI: 1.18–3.03) and poor-quality studies, which showed a lower and non-significant estimate (OR 1.19; 95% CI: 0.68–2.09).

In contrast, no statistically significant differences were observed between subgroups defined by study design (p = 0.17) and geographic region (p = 0.08). Nevertheless, differences in effect sizes were noted across these categories. The association between endometriosis and fibroids was higher in cross-sectional studies (OR 3.79; 95% CI:1.28–11.19) and cohort studies (OR 3.12; 95% CI: 1.62–6.04), compared to case-control designs (OR 1.43; 95% CI: 0.73–2.81). Geographically, studies from Asia Minor showed the highest estimate (OR 10.06; 95% CI: 2.62–38.60), followed by Asia (OR 4.30; 95% CI: 1.06–17.48), North America (OR 3.74; 95% CI: 1.71–8.22), and Australia (OR 3.35; 95% CI: 0.96–11.66), while European studies reported the lowest estimate (OR 1.25; 95% CI: 0.56–2.80).

A trend toward progressively stronger associations was observed with increasing prevalence of multiparity, although between-group differences did not reach statistical significance (p = 0.09). Compared to studies with mostly nulliparous populations (OR 1.56; 95% CI: 0.53–4.59), the association was stronger in those where more than 50% of women were multiparous (OR 4.52; 95% CI: 2.53–8.10), and reached its highest value in studies where over 95% of participants were multiparous (OR 5.23; 95% CI: 2.09–13.04).

Finally, stratification by age showed similar estimates in populations with a mean age ≤35 years (OR 2.68; 95% CI: 1.52–4.71) and >35 years (OR 2.58; 95% CI: 1.09–6.11; p = p = 0.94).

Sensitivity analysis: occurrence of uterine fibroids in endometriosis vs. asymptomatic controls

Studies overview

A total of 12 studies compared women with endometriosis to asymptomatic controls,^36,37,40, 41, 42, 43, 44, 45^,48,52, 53, 54 including a total of 3,103,146 women. The main eligibility criteria and the characteristics of the study populations for the included studies are provided in Supplementary Table S3.

Of the 12 included studies, seven were cohort studies (3,091,737women)^{36,40,43,45,48,53,54} and five were cross-sectional studies (11,409 women).^{37,41,42,44,52} Eight studies derived data from databases (3,094,375 women) and four from interviews (8771 women); none relied on hospital records, reflecting the outpatient setting of the populations included in this sensitivity analysis. Three studies were conducted in Asia (203,793 women),^40,44,48 two in Asia Minor (5344 women),^41,42 one in Australia (837,942 women),⁴³ one in Europe (116,896 women)⁵⁴ and five in North America (1,430,729 women).^{36,37,45,52,53} One study included mostly nulliparous women (630,523 women),⁴⁸ while eight included mostly multiparous women (1,059,691women).^37,40, 41, 42, 43, 44, 45^,53 Of these, five studies included populations in which more than 95% of women were multiparous (n = 178,881).40, 41, 42^,44,45

Main results and certainty of the evidence

In the analysis restricted to studies comparing women with endometriosis to asymptomatic controls, the pooled odds ratio for the co-occurrence of uterine fibroids was markedly higher than in the primary analysis (OR 7.01; 95% CI: 4.72–10.41; I² = 99.7%) (Fig. 3). The certainty of evidence was judged as low according to the GRADE framework (Supplementary Table S2).

Fig. 3.

Forest plot of studies included in the sensitivity analysis of asymptomatic controls.

Leave-one-out sensitivity analysis confirmed the robustness of this association (Supplementary Fig. S18). In the analysis excluding studies with significant baseline age differences between groups (2,215,153 patients: 63,161 with endometriosis and 2,151,992 controls),^{36,37,40,41,45,48,54} the pooled estimate remained consistent (OR 6.64; 95% CI: 4.08–10.82; I² = 99.7%) (Supplementary Figs. S19 and S20). Meta-regression based on mean age within this subset revealed a statistically significant positive association between age and effect size (coefficient = 0.17; 95% CI: 0.05–0.28; p < 0.01).

Subgroup analyses

Subgroup analyses are summarized in Table 2. Corresponding forest and funnel plots are presented in Supplementary Figs. S21–S33.

Significant differences between subgroups were observed for geographic region (p < 0.01) and study quality (p < 0.01). The highest estimates were observed in studies from Asia Minor (OR 10.06; 95% CI: 2.62–38.60), Australia (OR 6.03; 95% CI: 2.98–12.2), and Asia (OR 7.56; 95% CI: 3.05–19.18). Good-quality studies showed the strongest association between endometriosis and fibroids (OR 7.72; 95% CI: 5.26–11.33) compared to fair-quality studies (OR 2.68; 95% CI: 2.52–2.86).

Subgroup differences were not statistically significant for study design (p = 0.61), data source (p = 0.32), parity (p = 0.31), or age (p = 0.52). However, consistently elevated estimates were observed across all the subgroups, including both cohort and cross-sectional designs, different parity categories, and populations with younger and older mean ages, suggesting robustness of the association across diverse contexts.

Discussion

This systematic review and meta-analysis provide evidence supporting an epidemiological association between endometriosis and uterine fibroids.

Across a total population of 3,109,231 women (118,716 with endometriosis and 2,990,515 controls), the pooled OR for the presence of fibroids in women with endometriosis was 2.91 (95% CI: 1.78–4.75) This association remained robust in sensitivity analyses, including those restricted to studies with age-balanced groups or age-adjusted estimates. Meta-regression found no significant effect of age, while BMI showed a consistent positive association with effect size (p < 0.01). Subgroup analyses revealed significantly stronger associations in high-quality studies and a trend toward higher estimates in studies with increasing proportions of multiparous women. Although the trend did not reach statistical significance, the association was weakest in studies with mostly nulliparous women and progressively increased in those with >50% and >95% multiparous participants.

A sensitivity analysis restricted to studies comparing women with endometriosis to asymptomatic controls showed a markedly stronger association (OR = 7.01; 95% CI: 4.72–10.41). This result was consistent across subgroup analyses and stronger in high-quality studies. In meta-regression analysis age significantly modified the association (p < 0.01).

Despite the consistent direction of the association across analyses, statistical heterogeneity remained high, and the overall certainty of the evidence was rated as low, likely due to residual clinical and methodological variability across studies. As a result, while the observed association appears robust, its magnitude may not be fully generalizable across all populations and settings.

The finding that uterine fibroids occur 3 to 10 times more frequently in women with endometriosis compared to controls is supported by epidemiological, biological, and clinical considerations.

First, endometriosis and uterine fibroids share common pathogenic mechanisms, particularly involving hormonal influences. Hyperestrogenism plays a critical role in the pathogenesis of endometriosis, both in the establishment and maintenance of lesions.⁹ Similarly, uterine fibroids are hormonally responsive, with growth stimulated by estrogen exposure.¹¹ Several risk factors, including early menarche, nulliparity, and hormonal dysregulation secondary to obesity and metabolic alterations, are common to both conditions.55, 56, 57, 58 Furthermore, both endometriosis and uterine fibroids commonly regress after menopause or during prolonged therapies inducing ovulation suppression and hypoestrogenism.^59,60

Supportive evidence also comes from genetic studies. Genome-wide association studies (GWAS) on endometriosis have demonstrated significant genetic correlations between endometriosis and traits indicative of increased exposure to menstruation and hormones, such as excessive or irregular menstruation, shorter menstrual cycles, earlier menarche, and uterine fibroids.⁶¹ Additional genetic correlation analyses confirmed these findings, identifying genome-wide correlations for 22 comorbid traits. Among these, uterine fibroids and heavy menstrual bleeding showed the highest number of shared genetic variants with endometriosis (11 variants with ≥90% probability).⁶²

Clinically, symptoms arising from uterine pathology overlap significantly between endometriosis and uterine fibroids. Our sensitivity analysis showed a stronger association when women with endometriosis were compared to asymptomatic controls from the general population undergoing routine check-ups. This likely reflects the impact of control group selection on the observed association. When controls are asymptomatic and at lower baseline risk for fibroids, the contrast with endometriosis cases is more pronounced, resulting in a stronger association. In contrast, studies using hospital-based controls often include women evaluated for symptoms such as infertility or heavy menstrual bleeding—conditions that are independently associated with fibroids, and in some cases, also with endometriosis. This clinical overlap may reduce the contrast between cases and controls, potentially attenuating the observed effect estimates. Notably, the prevalence of endometriosis in hospital-based studies was markedly elevated, averaging approximately 32%, which is not reflective of its actual prevalence in the general population.¹

This issue has both methodological and clinical implications. In general outpatient settings, fibroids are more likely to go undetected compared to specialist centers that manage endometriosis, where patients typically present with complex or severe symptoms requiring thorough diagnostic work-up. In such settings, recognizing the potential co-occurrence of fibroids and endometriosis is essential to ensure accurate evaluation and appropriate clinical management.

Although fibroids are asymptomatic in some patients, approximately 30% of women experience significant symptoms, including severe pelvic pain, dysmenorrhea, non-cyclic pelvic pain, infertility, abnormal uterine bleeding, heavy menstrual bleeding, fatigue, dyspareunia, and bladder or bowel dysfunction.⁶³ Notably, symptoms of fibroids may overlap with those of endometriosis or adenomyosis, highlighting the importance for clinicians to remain vigilant about possible co-occurrence. For instance, in the setting of infertility, concurrent adenomyosis and fibroids have been identified in up to 20% of infertile women with endometriosis.⁶⁴ Therefore, in cases of endometriosis-associated infertility, the presence of coexisting fibroids may influence treatment decisions and patient prognosis.⁶⁵ Similarly, the presence of uterine fibroids in women with adenomyosis-associated abnormal uterine bleeding should inform treatment choices within a shared decision-making framework, balancing short- and long-term therapeutic goals and reproductive desires.⁶⁶ Furthermore, women presenting with significant pain despite relatively small fibroids may have concomitant endometriosis and adenomyosis, a co-occurrence clinicians should be mindful of when evaluating and managing patients.⁶⁷

To the best of our knowledge, this is the first meta-analysis providing a quantitative estimate of the burden of uterine fibroids in women with endometriosis. A major strength of this study is the large sample size (over 3 million patients), enabling robust estimates of effect. Additionally, we performed a sensitivity analysis restricted to studies where controls without endometriosis were asymptomatic women representative of an outpatient setting. This methodological approach aimed to reduce misclassification bias (due to undiagnosed endometriosis in symptomatic controls) and to minimize potential outcome detection bias, as fibroids might be overdiagnosed in women undergoing evaluation or surgery for gynecological symptoms. Indeed, the retrospective design and specific characteristics of healthcare systems in many studies included in our meta-analysis may introduce selection bias, especially considering that most of these studies were conducted in referral centres predominantly visited by patients experiencing uterine diseases and pelvic pain.⁶⁸ However, we acknowledge that these control groups may still include undiagnosed cases or women with subclinical gynecological conditions, and thus may not represent truly disease-free populations.

However, several limitations must be considered. First, significant heterogeneity was observed across studies (I² > 99% in most analyses), reflecting variability in study populations, diagnostic criteria, and study designs. Although we attempted to address this heterogeneity through subgroup analyses, residual confounding remains possible. The main bias to be acknowledged is ascertainment bias, arising either from detection bias—potentially leading to underestimation of fibroid prevalence when comparing endometriosis cases with asymptomatic controls from the general population—or from referral bias, which may affect hospital-based sampling and lead to overrepresentation of fibroids among controls. Such biases may have led to systematic distortions in disease prevalence and associations, thereby contributing to the heterogeneity of the estimates. Notably, the prevalence of endometriosis was 3.8% in both the main analysis and the sensitivity analysis restricted to asymptomatic controls—likely reflective of the true prevalence in the general population—whereas it was approximately ten times higher in hospital-based datasets, likely contributing to the lack of association observed in this subset. In this context, a higher prevalence of endometriosis within a study population—typically indicative of a hospital-based or highly selected sample—may reduce the contrast between exposed and unexposed groups, thus attenuating the observed association with fibroids due to selection bias and population homogeneity. Finally, the design and reporting methods of the original studies limited our ability to explore critical confounders such as race or ethnicity, menstrual history, fertility status, hormonal treatments, and symptoms at clinical presentation (e.g., AUB). Consequently, the possibility of a weaker or even stronger association in specific populations with potentially lower heterogeneity cannot be excluded. Moreover, hysterectomy status was not consistently reported and could not be accounted for, possibly leading to underestimation or misclassification of fibroid diagnosis in some studies, despite the established reliability of non-invasive diagnostic methods. Another limitation is that our meta-analysis combined data from cohort, case-control, and cross-sectional studies, introducing methodological heterogeneity due to differences in study design and confounding adjustment. Moreover, although adjusted ORs or crude estimates calculated from raw data provided a standardized and comparable measure of association across studies,⁶⁹ they are not measures of frequency and may substantially overestimate risk, especially in the context of common outcomes. These factors may further limit the generalizability and interpretability of the effect estimates. Lastly, the certainty of evidence from our analyses was rated low according to GRADE criteria, indicating the need for further high-quality studies to confirm these associations.

Endometriosis and uterine fibroids are among the most prevalent gynecological disorders, both of which can significantly impact women’s quality of life and reproductive health. This meta-analysis suggests a co-occurrence between endometriosis and uterine fibroids, possibly reflecting shared pathophysiological mechanisms, particularly those related to estrogen signaling.

Given the significant overlap of core symptoms, such as pelvic pain and infertility, recognizing and addressing the presence of fibroids in women with endometriosis is essential. Treating one condition without identifying the other may result in inadequate symptom relief, persisting infertility, or the need for additional medical or surgical interventions.⁶⁹

Although malignant transformation is rare for both conditions, they have been linked to increased mortality risk from gynecological cancers, potentially mediated by shared risk factors such as infertility and nulliparity.^70,71 Moreover, the coexistence of endometriosis and fibroids may increase long-term risk of premature mortality, particularly by contributing to greater susceptibility to cardiovascular disease in later life.⁷²

While further research is necessary to fully elucidate the biological foundations underlying the coexistence of endometriosis and uterine fibroids, clinical awareness of this association is critical for optimizing patient management and improving both short- and long-term health outcomes.

Contributors

E.S., P.V., P.Ve., N.S., M.C. and A.F. designed this study and N.S., E.S., P.V., and P.Ve. supervised this study. A.F., L.S., C.F., B.M., R.B., and M.C. selected studies for the inclusion and evaluation of quality. N.S. and A.F. collected the data. A.F. and L.S. accessed and verified the underlying data. Statistical analysis was performed by A.F. Interpretations of the results were made by N.S., E.S., and P.V. The first draft of the manuscript was written by A.F., N.S., and M.C. The revision process and approbation of the final version of the manuscript were made by E.S., P.V., P.Ve., N.S., M.C. and A.F.

Data sharing statement

All related data have been presented within the manuscript. The dataset supporting the conclusions of this article is available from the authors on request.

Declaration of interests

P.V. is the Co-Editor in Chief of Journal of Endometriosis and Uterine Disorders. P.Ve. has received royalties from Wolters Kluwer for chapters on endometriosis management in the clinical decision support resource UpToDate; and maintains both a public and private gynaecological practice. E.S. reports grants from Ferring, IBSA and Theramex, grants and personal fees from Ibsa and Gedeon-Richter, outside the submitted work. All the other authors do not have any conflict of interest to declare.