Diagnostic Disparities in Endometriosis and Adenomyosis: Investigating Social Vulnerability and Access to Care in an Ancestrally Diverse Population

Following Institutional Review Board approval, we collected genomic and clinical information on all female participants aged 18–51 years who had been enrolled in the Penn Medicine BioBank (PMBB) between 2008 and 2020. PMBB is a prospectively collected biorepository linked to electronic health records (EHR). 18 Participants were consented and enrolled during any outpatient visit across the University of Pennsylvania Health System and provided access to their medical records and at least one blood sample for genomic sequencing. 18 Patient records were reviewed for International Classification of Diseases (ICD-9/10) codes for endometriosis and/or adenomyosis at any encounter before or after enrollment, through March 2024. These codes included peritoneal (617.3A, 617.3B/N80.2, N80.3), ovarian (617.1A/N80.1), deep (617.4A, 617.5A/N80.4, N80.5, N80.80, N80.81), other/unspecified (617.8X/N80.8, N80.89, N80.9), and uterine endometriosis (617.0/N80.0), which contains a subcategory for adenomyosis (N80.03). Codes labeled as “uterine endometriosis” were classified as adenomyosis. Cases were independently reviewed by two clinicians and classified as surgically confirmed (visual diagnosis on operative report and/or histopathological confirmation), imaging confirmed (ultrasound or MRI showing endometrioma, deeply infiltrating disease, or criteria for adenomyosis), or clinically diagnosed (characteristic symptoms empirically treated by a gynecologist). Among patients with a surgical diagnosis of endometriosis, disease severity was also assessed using the American Association of Gynecologic Laparoscopists (AAGL) Endometriosis Classification (score and stage I–IV). To reduce bias due to differential access to surgery, all diagnostic categories were included in the primary analyses. As an additional validation, we analyzed a stricter case definition requiring surgical and/or imaging confirmation. A random audit of 9% of the control charts was performed to assess false negatives, and sensitivity and specificity of ICD codes for endometriosis and adenomyosis were calculated with 95% confidence intervals using exact binomial methods. Genotyping was performed on all PMBB participants, with quality control and genotype imputation to increase marker density. 18 Ancestry was inferred using ancestry-informative single nucleotide polymorphisms, with reference populations drawn from the 1000 Genomes Project. Principal component analysis (PCA) was performed to capture the primary axes of genetic variation, and individual participants were projected alongside reference populations. Based on clustering with these references, participants were assigned to one of five continental ancestry groups: African (AFR), Admixed American (AMR), East Asian (EAS), European (EUR), and South Asian (SAS). 19 , 20 Clustering along the first two principal components is shown in Figure 1 . Social Vulnerability Index (SVI) domain scores were assigned based on the Census tract of the patient’s residential address at enrollment. 21 , 22 The SVI, developed by the Centers for Disease Control and Prevention, incorporates 16 sociodemographic factors grouped into four domains ( Fig 2 ). This study included three domains: (1) socioeconomic status (poverty, unemployment, high housing cost burden, no high school diploma, lack of insurance), (2) household composition and disability (age distribution, disability status, single-parent households, English proficiency), and (3) housing and transportation (multi-unit structures, crowding, lack of vehicle access, group quarters). For each variable, tracts are ranked relative to all other U.S. Census tracts, yielding a percentile rank between 0 and 1. SVI values were analyzed continuously and in quartiles (0–0.25: low; 0.25–0.50: mild; 0.50–0.75: moderate; 0.75–1.0: high vulnerability). The race/ethnicity SVI domain was excluded, since race and ethnicity were captured directly via self-report. Cohort characteristics were summarized overall and by genetic ancestry. Concordance between genetic ancestry and self-reported race/ethnicity was described, and distributions of SVI domains were compared across ancestry groups. Logistic regression was used to model associations between ancestry and (a) any endometriosis diagnosis and (b) surgical diagnosis of endometriosis, with separate parallel analyses for adenomyosis. Multivariable models were adjusted for age, body mass index (BMI), socioeconomic SVI, and family/household SVI. We also tested interaction terms involving ancestry and SVI and conducted stratified analyses by ancestry and SVI quartiles. Results are reported as odds ratios (ORs) with 95% confidence intervals (CIs), using two-sided α=0.05.

Of 9,258 total female participants age 18–51 who underwent genomic sequencing, genetic analysis identified 5,260 (56.8%) with European, 3,243 (35.0%) African, 296 (3.2%) East Asian, 264 (2.9%) Admixed American, and 195 (2.1%) South Asian ancestry. Genetic ancestry correlated with self-reported race and ethnicity groups (e.g., European with White/non-Hispanic and African with Black/non-Hispanic) but showed incomplete overlap, particularly among Admixed American and Asian groups ( Table 1 ). Baseline characteristics of participants are presented in Table 2 . Compared with those of European ancestry, participants of African and Admixed American ancestry had higher BMI and were more likely to reside in areas with greater social vulnerability across all domains, including socioeconomic status, household composition, and housing and transportation (p < 0.001). Participants of African ancestry also had significantly higher gravidity, parity, and prevalence of fibroids (p < 0.001). Housing/transportation vulnerability was above average across all groups, consistent with the largely urban catchment area of the health system, in which multi-unit housing and lack of vehicle ownership were common. To strengthen the validity of our findings, we assessed the sensitivity and specificity of ICD codes for identifying endometriosis and/or adenomyosis. A total of 1,485 charts were reviewed for supporting documentation of the diagnosis (718 ICD-positive cases and 767 randomly selected ICD-negative controls). When any diagnostic modality was used as the reference (surgical, imaging, or clinical evaluation), ICD codes showed high sensitivity and specificity for identifying endometriosis (93.2% and 88.5%) and adenomyosis (85.8% and 87.5%). When a stricter definition requiring surgical diagnosis only (visual and/or histopathological evaluation) was used, sensitivity remained high but specificity declined (endometriosis 79.4%, adenomyosis 72.2%), indicating that some clinically significant cases were missed ( Suppl Table 1 ). The distribution of diagnosis types is shown in the supplementary material ( Suppl Table 2 ). When endometriosis diagnosis was defined broadly to include surgical, imaging, or clinical evaluation, the odds of diagnosis did not differ by genetic ancestry ( Table 3 ). However, when diagnosis was restricted to surgically confirmed cases, patients of African ancestry had lower odds of receiving a surgical diagnosis of endometriosis compared with those of European ancestry (OR 0.63, 95% CI 0.47–0.85). This association persisted after adjustment for age, BMI, and social vulnerability indices (adjusted OR [aOR] 0.57, 95% CI 0.38–0.85). Notably, the diagnosed prevalence of endometriosis among patients of African ancestry only approached that of European patients when imaging and clinical evaluation were also included, rather than surgery alone. An interaction term was incorporated into the models to examine whether the association between ancestry and diagnosis varied across quartiles of low, mild, moderate, and high vulnerability. Indeed, patients of African ancestry with the highest socioeconomic vulnerability had the lowest odds of any endometriosis diagnosis (aOR 0.29, 95% CI 0.14–0.61) and of surgical diagnosis specifically (aOR 0.31, 95% CI 0.12–0.80), compared with European-ancestry patients with the lowest vulnerability. By contrast, patients of African ancestry were more likely to be diagnosed with adenomyosis by imaging or at hysterectomy. Patients of African ancestry had markedly higher odds of adenomyosis (aOR 3.45, 95% CI 2.82–4.22), including surgical diagnosis at hysterectomy (aOR 3.41, 95% CI 2.49–4.69), compared with European-ancestry patients, independent of age, BMI, and social vulnerability indices. Figure 3 illustrates how socioeconomic vulnerability disproportionately affects diagnosis for patients of African ancestry. For European-ancestry patients, there is minimal change in the predicted probability of diagnosis for endometriosis or adenomyosis across socioeconomic strata. However, for patients of African ancestry, those with minimal socioeconomic vulnerability are more likely to obtain a diagnosis than those of high vulnerability. In other words, the probability of diagnosis decreased with increasing socioeconomic vulnerability (i.e., greater poverty, unemployment, and lack of insurance). Endometriosis severity scores and stage did not differ across ancestry groups, with most patients presenting with stage I–II disease (European: 71.6%; African: 70.8%) and fewer with stage III–IV ( Suppl Table 3 ). In multivariable regression, ancestry was not associated with severity at laparoscopy, after adjustment for age, BMI, and social vulnerability indices.

To our knowledge, this is the first study to evaluate genetically-defined ancestry and multiple measures of social vulnerability as key factors influencing disparities in endometriosis and adenomyosis diagnosis. Disparities in endometriosis diagnosis in this study were largely dictated by the diagnostic criteria used and by the social resources available to patients, particularly for those of African ancestry. When surgical confirmation was required, patients of African ancestry had lower odds of endometriosis diagnosis than patients of European ancestry, an effect that was more profound as socioeconomic vulnerability increased. In contrast, when a broader more inclusive definition of endometriosis diagnosis was applied (surgical, imaging, or clinical evaluation), the odds of endometriosis did not differ by ancestry. Overall, these findings suggest that previously reported higher rates of endometriosis among individuals of White race (which correlates with European ancestry) may reflect differences in socioeconomic resources and surgical evaluation of pelvic pain, rather than true differences in endometriosis susceptibility. Interestingly, in this same cohort, adenomyosis was more common among African-ancestry patients than European-ancestry patients. Concurrent adenomyosis along with endometriosis was also more frequently identified among patients of African ancestry. The diagnosis of adenomyosis may be less susceptible to structural and systemic biases, as the disease is often detected during evaluation for abnormal uterine bleeding or fibroids, and patients may progress to hysterectomy due to worsening non-pain-related symptoms, such as anemia. We suspect that differences in symptom presentation and diagnostic pathways may differentially shape disparities across both endometriosis and adenomyosis. While no prior studies have examined the interactions between ancestry, endometriosis or adenomyosis, and social vulnerability, several have reported higher incidence of endometriosis among patients of higher socioeconomic status or with private insurance. 10 , 23 , 24 Limited access to gynecologic subspecialists, underinsurance, medical mistrust, and exposure to racism and implicit bias may prevent or delay diagnosis in these patients. Conversely, in settings with better access across patients of diverse ancestry, such as infertility care populations with high insurance coverage and specialist involvement, or integrated health systems, disparities in diagnosis have not been observed. 25 , 26 Our study in an urban university-based system supports this: socioeconomic vulnerability affected diagnostic patterns and interacted with ancestry in ways that likely reflect other unmeasured social and structural factors. To reduce diagnostic delays, expedite treatment, and address inequities, there have been calls to prioritize clinical diagnosis and use of imaging rather than surgical diagnosis in patients with suspected disease. 17 , 27 Consistent with this framework, disparities in our cohort emerged only when laparoscopy was required, whereas they were not present when broader diagnostic definitions were used. Similarly, Zaritsky et al. found no disparities when using a highly inclusive definition of endometriosis—any endometriosis or adenomyosis ICD code, or two visits coded for pelvic pain (dysmenorrhea, dyspareunia, or other chronic pelvic pain). 26 This study has several limitations. First, underdiagnosis of both endometriosis and adenomyosis is likely due to missed or delayed recognition of symptoms in some patients and reliance on documentation in the electronic medical record. Although a major strength of this study is the rigor with which diagnoses, symptoms, disease severity, and concurrent conditions were categorized and verified, some diagnostic misclassification remains a concern because clinical, imaging-based, and surgically-confirmed diagnoses were all included. Second, given the single-institution study design, our findings may reflect patient demographics, referral patterns, and access to care issues specific to this academic health system or region, which may limit generalizability. Although this cohort included substantial multiethnic diversity compared with prior studies—with thousands of patients in the European and African ancestry groups allowing detection of relatively small absolute differences—sample sizes for Admixed American and Asian patients were insufficient to draw definitive conclusions. Third, although the neighborhood-level CDC Social Vulnerability Index is widely used and validated, we lacked access to patient-level metrics on income, education level, insurance coverage, access to a vehicle, and health care utilization. Finally, although genomic sequencing was used as an analytic tool to classify genetic ancestry independently of sociocultural identity, the findings of this study should not be misinterpreted as biologically deterministic. Genetic ancestry analysis offers insights about population history and anthropological patterns of human variation, but it does not measure genetic or biological susceptibility to disease. Based on the data in this study, no inferences can or should be made regarding genetic risk for endometriosis or adenomyosis. Recognizing the influence of social determinants, and their relationship with genetic ancestry, can be useful to inform strategies to promote more equitable recognition and diagnosis of endometriosis and adenomyosis in currently underrepresented populations.

As early as the 1930s, endometriosis was commonly characterized as a disease that primarily affected affluent White women. 1 – 4 In scientific articles, endometriosis was explicitly linked to wealth and race, described as “a disease of the rich,” more common among “white private patients than in the dispensary clientele,” and it was considered a “rarity” among Black patients. 2 – 4 This belief has been increasingly challenged, as recent evidence suggests that this observation was likely due to disparities in diagnosis and care rather than true differences in prevalence. 1 , 5 In the United States, Black and Hispanic patients have been more likely to experience misdiagnosis, delayed diagnosis, and undertreatment of endometriosis, as well as increased postoperative complications, compared with White patients. 6 – 8 Since histopathological confirmation via surgery has historically been the diagnostic gold standard for endometriosis, as well as its related condition adenomyosis, differences in access to gynecologic care and invasive procedures have directly impacted who receives a diagnosis and, by extension, the diversity of research datasets. For example, in a review of Kaiser Permanente records from 2006 to 2015, 70% of patients diagnosed with endometriosis were White, while only 6% were Hispanic, 9% Asian, and 5% non-Hispanic Black. 5 Many international registries similarly include fewer than 10% non-White or non-European participants. 9 – 12 Furthermore, although adenomyosis frequently coexists with endometriosis and shares overlapping symptoms and pathophysiological mechanisms, it is often excluded entirely from endometriosis studies, potentially exacerbating diagnostic biases and misclassification. 5 , 13 – 15 Identifying the underlying causes of racial and ethnic disparities in endometriosis and adenomyosis has been difficult due to three major methodological shortcomings in the literature: (1) inconsistent or non-validated diagnostic criteria that often do not distinguish between surgical, imaging-based, or clinically suspected disease; (2) reliance on broad race and ethnicity categories that largely reflect social and structural exposures in the community and are otherwise poor predictors of the genetic differences between individuals; and (3) lack of direct reporting on key social determinants of health, such as socioeconomic status, education, housing, and transportation, which can greatly influence access to care and diagnosis. 16 , 17 To address these limitations, we utilized genomic sequencing to more precisely classify genetic ancestry independent of self-reported race and ethnicity, allowing social factors—such as socioeconomic status—to be examined separately rather than conflated with racial or ethnic identity. Our primary objective was to evaluate differences in endometriosis and/or adenomyosis diagnosis across genetically-defined ancestry groups (African, Admixed American, East Asian, European, and South Asian), adjusting for relevant sociodemographic factors. We further examined whether diagnostic criteria, such as the requirement for surgical confirmation, might influence whether disparities were observed, since patients of different backgrounds or sociocultural circumstances may vary in their likelihood to undergo surgery or other gynecologic evaluation.