Epidemiology, Risk Factors, Diagnosis, and Comorbidities of Endometriosis: An Umbrella Review

This umbrella review was performed in accordance with the PRIOR 2023 methodological guidance [ 16 ] and the Cochrane Handbook for Systematic Reviews of Interventions (version 6.4) [ 20 ]. The protocol was prospectively registered in the PROSPERO international registry (CRD420261378862), and reporting followed the PRISMA-OvR 2021 checklist [ 17 ] ( Supplementary File S1 ). Systematic reviews, with or without quantitative meta-analysis, were eligible if they (i) addressed any clinical aspect of endometriosis—epidemiology, pathogenesis, diagnosis, pharmacological or surgical treatment, fertility, quality of life, or comorbidities; (ii) included women with confirmed or clinically suspected endometriosis of any age, including adolescents; and (iii) were published between January 2016 and March 2026 (foundational epidemiological reviews were considered without time restriction). Narrative reviews without a systematic search, duplicate reviews from the same author group based on identical samples, and reviews based exclusively on animal models or cell lines without clinical data were excluded [ 21 ]. Eligibility was not restricted by the country of origin of the primary studies. Reviews published in English or Russian were eligible for inclusion; one Russian-language review meeting all other criteria was identified and included. Reviews published in other languages were excluded due to translation resource constraints, which is acknowledged as a limitation. Systematic searches were conducted in four electronic databases: PubMed, Embase, Cochrane Library, and Scopus. Search strategies were developed in consultation with a medical librarian and adapted to the syntax of each database. The core logical structure was as follows: (endometriosis [topic]) AND (systematic review OR meta-analysis OR overview of reviews OR umbrella review [publication type or topic]). The full search string applied to PubMed (run on 15 March 2026) was as follows: ((“Endometriosis”[MeSH] OR “endometriosis”[Title/Abstract] OR “endometrioma”[Title/Abstract] OR “deep infiltrating endometriosis”[Title/Abstract]) AND (“Systematic Review”[Publication Type] OR “Meta-Analysis”[Publication Type] OR “systematic review”[Title/Abstract] OR “meta-analysis”[Title/Abstract] OR “overview of reviews”[Title/Abstract] OR “umbrella review”[Title/Abstract])) AND (“2016/01/01”[PDAT]: “2026/03/31”[PDAT]). Analogous strategies were developed using Emtree terms (Embase), CENTRAL filters (Cochrane Library), and TITLE-ABS-KEY operators (Scopus). Database-level language filters were applied consistently with the eligibility criteria stated above (English and Russian), and document types were limited to systematic reviews, meta-analyses, and overviews. Complete database-specific search strings, dates of execution, and the number of records retrieved per database are provided in Supplementary File S2 . Title/abstract screening, full-text eligibility assessment, data extraction, and AMSTAR-2 quality appraisal were performed independently and in duplicate by two reviewers (initials), with disagreements resolved by discussion and, where necessary, adjudication by a third senior reviewer (initials). Inter-rater agreement was quantified using Cohen’s kappa coefficient: κ = 0.84 (95% CI: 0.79–0.89) for title/abstract screening, κ = 0.88 (95% CI: 0.82–0.94) for full-text eligibility, and κ = 0.81 (95% CI: 0.75–0.87) for AMSTAR-2 domain ratings—with all values indicating substantial to almost perfect agreement. The methodological quality of each included review was appraised using AMSTAR-2, a validated 16-domain instrument [ 18 ]. The following seven domains are designated as critical: (i) protocol registration prior to review commencement; (ii) comprehensiveness of the search strategy; (iii) risk-of-bias assessment of primary studies; (iv) appropriateness of meta-analytic methods; (v) consideration of risk of bias when interpreting results; (vi) assessment of publication bias; and (vii) reporting of funding sources [ 18 ]. Overall confidence in each review was rated as high, moderate, low, or critically low, in accordance with the published AMSTAR-2 decision rules. The overlap of primary studies across included systematic reviews and meta-analyses was assessed using the Corrected Covered Area (CCA) method. For each clinical domain, a citation matrix was constructed (rows = unique primary studies; columns = included reviews), and CCA was calculated as CCA = (N − r)/(r × c − r), where N denotes the total number of inclusions, r denotes the number of unique primary studies, and c denotes the number of reviews. CCA values were interpreted as showing slight (0–5%), moderate (6–10%), high (11–15%), or very high (>15%) overlap; where high or very high overlap was identified, only the most recent and methodologically strongest review was retained for quantitative synthesis to avoid double-counting; full overlap matrices and domain-specific CCA values are reported in Supplementary File S3 . For meta-analyses providing quantitative data, the strength of each association was classified following the scheme of Fusar-Poli and Radua (2018), adapted for umbrella reviews [ 19 ]. The evidence classification scheme is presented in Table 1 .

Endometriosis is a chronic inflammatory disease in which tissue histologically and functionally similar to the endometrium is found outside the uterine cavity [ 1 , 2 ]. As an estrogen-dependent condition, it is accompanied by progesterone resistance, immune dysregulation, neuroinflammation, angiogenesis, and progressive tissue fibrosis [ 2 , 3 ]. The estimates of its global burden vary substantially depending on the data source. Although the GBD 2021 study documented 22.3 million prevalent cases of formally diagnosed endometriosis [ 4 ], population-based and clinical estimates suggest that the true prevalence may reach approximately 10% of women of reproductive age—corresponding to up to 190 million affected women worldwide [ 5 ]. This discrepancy is not contradictory but reflects the well-documented diagnostic gap discussed throughout this review. The clinical significance of endometriosis is defined not only by its high prevalence but also by the breadth of its medical and social consequences: chronic pelvic pain, dysmenorrhea, dyspareunia, infertility, reduced quality of life, psychological morbidity, and elevated risks of malignant neoplasms and systemic diseases [ 6 , 7 ]. The economic burden of endometriosis is comparable to that of type 2 diabetes mellitus: in the United States alone, direct and indirect costs associated with endometriosis exceed USD 80 billion annually [ 8 ]. A central paradox of contemporary medicine is that, despite such a high prevalence and severe consequences, an average of 4 to 12 years elapse from the appearance of first symptoms to diagnostic confirmation [ 9 , 10 ]. This diagnostic delay is driven by the cultural normalization of pain symptoms, insufficient awareness among primary-care physicians, absence of reliable non-invasive biomarkers, and continued requirement for laparoscopy as the diagnostic gold standard [ 9 , 11 ]. Over the past decade, the number of systematic reviews (SRs) and meta-analyses (MAs) on endometriosis has grown exponentially [ 12 ]. Nevertheless, these data remain scattered across individual clinical domains, frequently contradict one another, and vary substantially in methodological quality [ 13 ]. The umbrella review format—a systematic review of systematic reviews—provides a fundamentally different level of evidence synthesis, enabling not only the summation but also the critical appraisal, hierarchical organization, and integration of all available meta-analytic data [ 14 , 15 ]. In this review, we adopt the formal umbrella review methodology as defined by the Joanna Briggs Institute and codified in the PRIOR 2023 (Preferred Reporting Items for Overviews of Reviews) [ 16 ] and PRISMA-OvR 2021 [ 17 ] reporting guidelines. This framework requires the following: (i) a pre-registered protocol; (ii) a systematic search of multiple databases; (iii) a methodological quality appraisal of every included review using a validated instrument—in our case AMSTAR-2 [ 18 ]; (iv) an explicit assessment of primary-study overlap; and (v) a hierarchical classification of evidence strength. Accordingly, in this review, the term critical synthesis refers to this formal framework and not to a narrative review. This umbrella review has four objectives: (1) to systematize the current evidence on the global burden, risk factors, pathogenesis, diagnosis, treatment, and long-term consequences of endometriosis; (2) to critically appraise the methodological quality of all included reviews using AMSTAR-2 [ 18 ]; (3) to classify the strength of identified associations using the adapted Fusar-Poli and Radua evidence-grading scheme [ 19 ]; and (4) to formulate clinical recommendations and research priorities grounded in this hierarchical synthesis.

The systematic search retrieved 5653 records from four electronic databases (PubMed: 1847; Embase: 2103; Cochrane Library: 412; Scopus: 1291). An additional 47 records were identified through hand-searching of reference lists of included reviews and the PROSPERO registry. After the removal of duplicates present within the database records ( n = 1435) and across the databases ( n = 12), 4253 unique records remained for title/abstract screening. Of these, 3929 were excluded as not relevant, and 324 full-text articles were assessed for eligibility. Following the exclusion of 272 articles (reasons detailed in Figure 1 ), 52 systematic reviews and meta-analyses were included in the final synthesis. The 52 included reviews (39 with quantitative meta-analysis and 13 without) covered publications from 2017 to 2026, with a combined primary sample of more than 6,000,000 participants. Most reviews originated from the United States, the United Kingdom, China, Australia, and EU countries. A complete characterization of all 52 included systematic reviews and meta-analyses—including first author, year, journal, topic domain, number of primary studies, total participants, primary outcome, effect estimate with 95% confidence interval, between-study heterogeneity (I 2 ), publication bias assessment (Egger test where reported), AMSTAR-2 rating, and assigned evidence class—is provided in Supplementary Table S1 . Table 2 in the main text presents a thematic summary of these 52 reviews across six clinical domains. The primary-study overlap across the 39 quantitative meta-analyses was generally moderate. The domain-specific CCA values were as follows: epidemiology 4.2% (slight); risk factors 7.8% (moderate); diagnosis and biomarkers 11.3% (high); medical treatment 9.6% (moderate); surgery and fertility 6.4% (moderate); and comorbidities 3.1% (slight). In the diagnostic biomarker domain, where overlap was highest, three reviews were identified that drew on substantially shared primary cohorts; in these cases, the most recent and methodologically strongest review was prioritized for evidence-class assignment, and contributions of the others are presented descriptively only. The methodological quality varied across the 52 included reviews: AMSTAR-2 ratings were high for 13 reviews (25.0%), moderate for 21 (40.4%), low for 15 (28.8%), and critically low for 3 (5.8%). The complete domain-by-domain AMSTAR-2 appraisal for all 52 reviews is provided in Supplementary Table S2 , in which each of the 16 domains is rated as Yes, Partial Yes, No, or Not Applicable, and the seven critical domains are indicated. For readability, Figure 2 in the main text displays a representative subset of 22 reviews, selected to span all six clinical domains and all four overall confidence categories (high, moderate, low, critically low). The PRISMA-OvR flow diagram is presented in Figure 1 . The final umbrella review included 52 systematic reviews and meta-analyses, distributed across six clinical domains. The epidemiology of endometriosis is fundamentally dependent on the diagnostic standard employed and the study population, which explains the substantial variation in reported estimates [ 4 , 6 ]. According to GBD 2021 22,275,015 prevalent cases of endometriosis were recorded in 2021, with an age-standardized prevalence rate (ASPR) of 275.57 per 100,000 persons [ 4 ]. Despite a 1.07% reduction in the age-standardized incidence rate (ASIR) between 1990 and 2021, the absolute number of new cases continues to rise [ 4 ]. ARIMA and exponential-smoothing projection models predict a continued increase in disability-adjusted life-years (DALYs) through 2050, particularly in developing countries [ 5 ]. The wide reported range of prevalence—from approximately 5–10% in unselected reproductive-age population samples, to 25–50% among women presenting with infertility, and up to 87% in tertiary referral cohorts with chronic pelvic pain—reflects three principal sources of heterogeneity rather than true biological variation. First, case ascertainment differs markedly: studies relying on administrative or insurance claims data identify only formally coded diagnoses (yielding ~1%), whereas studies requiring laparoscopic confirmation capture both symptomatic and incidentally discovered disease (yielding higher rates). Second, population selection introduces strong upward bias in specialist-clinic cohorts compared with general-population samples. Third, diagnostic criteria have evolved: contemporary studies increasingly include sonographically diagnosed deep infiltrating endometriosis and adenomyosis under the broader entity, whereas older studies were largely restricted to surgically confirmed pelvic endometriosis. Together, these factors fully account for the observed range and underscore the need for standardized diagnostic criteria in future epidemiological work. A clear diagnostic gap is therefore evident: insurance data report prevalence of approximately 1%, while clinical studies yield 6.8% and self-reported surveys 6.6% [ 6 ]. This gap reflects not true epidemiology but a systemic diagnostic deficit, with up to 65% of women with endometriosis initially receiving an incorrect diagnosis [ 9 , 10 ]. Prevalence stratified by population subgroup is shown in Figure 3 A. GBD 2021 analyses revealed that the highest ASIR and ASDR values were recorded in regions with a low socio-demographic index (SDI) [ 5 ]. This apparent paradox is explained by the inverse relationship between diagnostic accessibility and case detection in resource-constrained settings [ 5 ]. DALYs attributable to endometriosis-associated infertility are concentrated in the reproductive age group (peak 25–29 years) and are projected to increase through 2044 [ 7 ]. The evidence hierarchy for risk factors in endometriosis is presented in Table 3 . The first methodologically rigorous umbrella review of risk factors (354 observational studies, >5,000,000 participants) identified the following evidence hierarchy [ 22 ]. The contemporary SRs and molecular studies support a multilevel pathogenetic model in which no single theory explains the complete clinical picture [ 23 , 40 , 41 ]. The classical retrograde menstruation theory explains most cases of pelvic endometriosis but cannot account for extra-pelvic locations or cases arising in the absence of a uterus [ 40 ]. The current evidence supports an integrative concept encompassing: (I) local estrogenic dominance with overexpression of aromatase (CYP19A1) and increased ERβ/ERα ratio due to epigenetic regulation [ 23 ]; (II) progesterone resistance involving impaired HOXA10/HOXA11 expression and defective stromal cell decidualization [ 41 ]; (III) immune dysregulation with reduced NK-cell cytotoxic activity, Th1/Th2 shift toward a pro-inflammatory phenotype, and elevated peritoneal fluid IL-6, IL-8, and TNF-α [ 40 ]; (IV) epigenetic aberrations—hypomethylation of the ERβ and CYP19A1 promoters, hypermethylation of HOXA10, and histone acetylation—collectively constituting the “epigenetic memory” of endometriotic lesions [ 23 ]; (V) oxidative stress and ferroptosis, whereby excess free iron from cyclical bleeding induces Fenton-reaction oxidative damage, impairing granulosa cells and oocyte quality [ 41 ]; and (VI) microbiota dysbiosis, with SRs confirming an association between endometriosis and altered intestinal and vaginal microbiota modulating immune responses via an entero-gonadal axis [ 24 ]. The diagnostic delay in endometriosis is one of the most reproducible and clinically significant phenomena in reproductive medicine [ 9 ]. The systematic review by De Corte et al. (2025, 17 studies, 2018–2023) confirmed that delay persists even in contemporary cohorts: median time from symptom onset to diagnosis ranges from 0.3 to 12 years, with the highest values in Europe (6–10 years) and the lowest in specialized referral centers [ 9 ]. The authors stratify delay into three components: (1) primary delay—from symptoms to first medical consultation (cultural normalization of dysmenorrhea); (2) clinical delay—from first consultation to diagnosis (insufficient primary-care physician awareness); and (3) systemic delay—shortage of specialized centers [ 9 , 10 ]. Notably, the diagnostic delay has not meaningfully decreased over 20 years despite growing awareness, indicating a structural rather than technological nature of the problem [ 9 ]. The laparoscopy with histological verification remains the diagnostic gold standard—a paradox in the era of precision medicine [ 11 ]. SRs describe the following biomarker landscape (summarized in Table 4 and Figure 4 ). Sensitivity, specificity, and AUC values are extracted from the meta-analyses and primary validation studies cited in Table 4 . Clinical readiness and reproducibility scores represent consensus expert estimates assigned by the authors based on the maturity of validation in the included systematic reviews; these are descriptive heuristics intended to aid interpretation and should not be interpreted as quantitative performance metrics. The very high AUC values reported for some emerging biomarker platforms (e.g., 0.997 for PromarkerEndo in the most advanced reported analysis; 0.906 for the IMAGENDO AI-assisted imaging platform) warrant particular caution in interpretation for several reasons. (i) Sample sizes in the pivotal validation studies remain modest ( n = 79 for the Oxford CA-125 + BDNF panel; n = 704 for PromarkerEndo across its development and validation cohorts), with limited representation of important subgroups including adolescents, perimenopausal women, and ethnic minorities. (ii) Most reported performance metrics derive from internal cross-validation or from single-center external validation; multi-center, geographically diverse, and prospectively recruited validation cohorts are largely lacking. (iii) The risk of overfitting in machine-learning-based diagnostic models is well documented, particularly when feature selection and model tuning are performed on the same dataset used for performance estimation. (iv) Spectrum bias is a substantial concern: case–control designs comparing symptomatic women with surgical confirmation against healthy controls systematically overestimate real-world diagnostic performance in screening settings. Until prospective, multi-center validation in unselected clinical populations is available, these tools should be regarded as promising research-stage technologies rather than implementation-ready diagnostics. Comparative treatment effectiveness with evidence classification is presented in Table 5 . Endometriosis is an estrogen-dependent disease, making hormonal suppression the cornerstone of medical management [ 29 , 30 ]. The meta-analysis ( n = 2137, 14 RCTs) by Zakhari et al. (2021) established that post-operative hormonal suppression significantly reduces recurrence risk compared with expectant management [ 29 ]. Dienogest (2 mg/day) demonstrated a recurrence OR of 0.30 (95% CI 0.18–0.53; p < 0.001)—a 70% relative risk reduction—together with VAS pain reduction SMD −1.04 at 3 months and SMD −0.79 at 12 months [ 29 ]. The network meta-analysis by Wattanayingcharoenchai et al. (2021) ranked dienogest highest for endometrioma recurrence prevention and probability of achieving pregnancy among all compared agents [ 30 ]. The PRE-EMPT randomized controlled trial confirmed equivalence between extended-cycle progestins and COCs for pain-recurrence prevention, with superior tolerability of progestins [ 31 ]. Laparoscopy remains the dual standard—simultaneously diagnostic and therapeutic. Systematic reviews demonstrate that surgical resection provides pain relief in 60–80% of patients; however, without maintenance hormonal therapy, recurrence rates reach 40–50% within five years [ 32 , 39 ]. For ovarian endometriomas, cystectomy versus fenestration or ablation provides a significantly lower recurrence risk (OR 0.41; 95% CI 0.24–0.71), despite a greater risk of reduced ovarian reserve [ 39 ]. For deep infiltrating endometriosis (DIE), laparoscopic resection reduces pain intensity; however, fertility outcomes after DIE surgery remain heterogeneous, and decisions should be individualized. Key findings from meta-analyses on the impact of endometriosis on IVF outcomes include: (i) a reduction in retrieved oocytes (MD −1.22; p < 0.05) and in mature oocytes (MD −2.24 [ 34 ]; (ii) IVF outcomes comparable to those of unexplained infertility at ASRM stages I–II, but significantly inferior at stages III–IV [ 34 ]; (iii) surgery before IVF for stages I–II does not meaningfully improve live-birth rates (evidence level A) [ 43 ]; (iv) surgery for stages III–IV is indicated for endometriomas > 5 cm and pain, but is not mandatory prior to IVF in other clinical scenarios (evidence level B) [ 43 ]; (v) post-operative hormonal suppression combined with IVF improves pregnancy rates by 10–15% [ 32 ]; and (vi) fertility preservation (oocyte vitrification) is recommended from the time of diagnosis in cases of bilateral endometriomas, recurrent disease, or prior ovarian surgery [ 33 ]. A meta-analysis (nine studies, random-effects model) established that the risk of anxiety disorders in women with endometriosis was RR = 2.82 (95% CI 1.69–4.68; p < 0.001)—a nearly threefold increase [ 35 ]. The pooled relative risk for depression was 2.78 (95% CI 1.63–5.25); however, between-study heterogeneity was extreme (I 2 = 100%), indicating that the pooled estimate should be interpreted with substantial caution and likely reflects a wide and clinically meaningful range of true effects across populations rather than a single underlying risk ratio [ 38 ]. The direction and consistency of effect across all included studies (every primary study reported an increased risk) supports the qualitative conclusion that depression risk is elevated, but a precise magnitude cannot be reliably estimated from these data. Likely sources of heterogeneity include differences in depression ascertainment, study population, and timing relative to diagnosis. Maladaptive emotion-regulation strategies (catastrophizing, suppression) act as significant mediators of the pain–depression relationship [ 38 ]. Systematic reviews document the following associations: clear-cell ovarian carcinoma (RR 2.10; 95% CI 1.50–2.93; Class II–III) [ 37 ]; endometrioid ovarian carcinoma (RR 1.55; 95% CI 1.22–1.97; Class III) [ 37 ]; and endometrial carcinoma with a moderate risk increase (RR ≈ 1.35) in atypical endometriosis [ 37 ]. The proposed mechanism involves direct malignant transformation through ARID1A/PIK3CA mutation accumulation and oxidative stress in retention cysts [ 37 ]. Meta-analyses document an elevated frequency of autoimmune diseases in women with endometriosis: OR 1.3–1.8 for autoimmune conditions overall [ 37 ]. The most reproducible associations include rheumatoid arthritis, systemic lupus erythematosus, autoimmune thyroiditis, irritable bowel syndrome, and migraine [ 37 ]. The accumulated data support the concept of endometriosis as a systemic immuno-inflammatory condition [ 24 , 37 ]. The classification of association strength by evidence class and a forest plot of key associations identified across the 52 included meta-analyses are presented in Figure 5 and Figure 6 .

A pervasive theme of the present umbrella review is delay at every stage of the clinical trajectory: delay in symptom recognition (cultural normalization of dysmenorrhea), delay in diagnosis (4–12 years across regions), delay in treatment (hormonal therapy frequently prescribed empirically without diagnostic confirmation), and delay in research (randomized data on many key questions remain notably limited) [ 9 , 10 , 36 ]. Of particular significance is the finding that diagnostic delay has not decreased over two decades despite growing awareness [ 9 ]. This indicates a structural—rather than technological—nature of the problem. Its resolution requires primary-care physician training, standardization of screening protocols, validation of non-invasive biomarkers, and establishment of specialized centers [ 36 ]. The accumulated meta-analytic evidence supports dienogest as one of the preferred options for post-operative hormonal maintenance therapy, with consistent effect estimates across the largest available meta-analyses (recurrence OR~0.30) and a favorable tolerability profile relative to GnRH agonists in the available comparative data [ 29 , 30 ]. However, characterization of dienogest as a universal first-line standard would currently exceed the underlying evidence: most included trials were heterogeneous in inclusion criteria (mixed stages, mixed surgical approaches), follow-up duration was generally limited to 12–24 months, and head-to-head data against extended-cycle combined oral contraceptives—which the PRE-EMPT trial suggests are non-inferior for pain recurrence—remain limited [ 32 , 34 ]. Clinical decisions should therefore be individualized according to patient priorities (pain control, fertility timing, side-effect profile), and inclusion of dienogest in formal first-line guidance should await further high-quality comparative effectiveness research and confirmation by guideline committees. None of the reviewed biomarkers is ready for immediate routine implementation as a standalone test [ 27 , 28 ]. However, the totality of data on PromarkerEndo (AUC 0.997), AI-assisted multimodal platforms (AUC 0.906), and miRNA panels (accuracy ≥ 90%) creates, for the first time, real prerequisites for non-invasive screening within the next 5–10 years [ 28 , 42 ]. The key prerequisite is multi-biomarker approaches integrating molecular, imaging, and clinical data. The Yale breakthrough (Vash-Margita & Taylor, 2025) with an adolescent miRNA signature opens the prospect of early screening before irreversible reproductive damage occurs [ 26 ]. One of the key conceptual shifts of this review is that consistent evidence supports the view that endometriosis is a systemic, rather than a localized, disease [ 24 , 40 , 41 ]. It affects the immune system (NK-cell dysregulation, Th1/Th2 imbalance), neuroinflammatory pathways (central sensitization, glial activation), metabolism, microbiota, mental health (anxiety RR 2.82; depression RR 2.78), and oncogenesis [ 35 , 37 , 38 , 40 ]. This demands a fundamentally different multidisciplinary approach: gynecologist, reproductive endocrinologist, psychologist, and immunologist as a unified care team [ 36 ]. Strengths of the present umbrella review include: systematic methodology in accordance with PRIOR 2023 and PRISMA-OvR 2021; AMSTAR-2 quality assessment across all 52 included reviews; application of a validated evidence-classification scheme; comprehensive coverage of six major clinical domains; and a prospectively registered protocol [ 16 , 17 , 18 , 19 ]. Limitations include: heterogeneity of primary studies (most included MAs demonstrate moderate-to-high I 2 > 50%), restricting the interpretation of pooled estimates; variability in diagnostic criteria across reviews (laparoscopy, ultrasound, self-report); potential publication bias, which, although low-to-moderate by Egger test in most included MAs, cannot be fully excluded; geographic inequality (majority of primary studies are conducted in Europe, North America, and East Asia, limiting generalizability to resource-constrained settings); and a paucity of RCTs for certain aspects of pathogenesis and therapy.

This umbrella review synthesizes evidence from 52 systematic reviews and meta-analyses encompassing more than 6,000,000 participants. The integral conclusion is unequivocal: endometriosis remains substantially under-diagnosed relative to its true population prevalence, frequently under-treated, and more biologically complex than has traditionally been recognized. Six key conclusions structure the evidence base are as follows: Global burden is increasing—22.3 million formally diagnosed cases (GBD 2021), with DALYs projected to rise through 2050, supporting prioritization of endometriosis within public health agendas [ 4 , 5 ]. Diagnostic delay (4–12 years) appears largely structural in nature and is unlikely to be resolved by technological advances alone [ 9 , 10 ]. Dienogest is among the preferred options for post-operative maintenance therapy (recurrence OR 0.30; Class II evidence), although individualized decision-making remains essential [ 29 ]. Proactive fertility counseling—including consideration of oocyte cryopreservation and timely initiation of assisted reproduction where indicated—should be offered from the time of diagnosis [ 32 , 33 ]. Multi-biomarker and AI-assisted diagnostic platforms represent the most promising current avenue toward non-invasive diagnosis, though external validation in diverse cohorts is required before clinical implementation [ 28 , 42 ]. Psychological assessment and support should be integrated into multidisciplinary care, recognizing the substantially elevated risk of anxiety and depressive symptoms (RR~2.8) [ 35 , 38 ]. Research priorities include: multicenter validation studies of non-invasive biomarkers in ethnically diverse cohorts; development of personalized treatment algorithms based on molecular phenotyping; long-term cohort studies of oncological risks; and randomized trials of psychological and neuromodulator interventions.