TOPLINE:

Analysis of electronic health records (EHRs) of more than 43,500 patients with endometriosis across six University of California (UC) medical centers revealed hundreds of significantly associated conditions, including genitourinary disorders, neoplasms, and autoimmune diseases. Patient clustering identified distinct subgroups with unique comorbidity patterns, suggesting potential pathways for personalized disease management.

METHODOLOGY:

• Researchers analyzed de-identified EHR data from 19,059 patients with endometriosis at UC San Francisco (UCSF) and 24,453 patients across five other UC medical centers.
• Analysis included case-control comparison using 1:1 propensity score matching against control individuals on the basis of age, sex, race, ethnicity, and location, with additional matching for healthcare utilization in some analyses.
• Investigators performed odds ratio (OR) analysis and unsupervised clustering techniques to identify and characterize patterns of diagnoses associated with endometriosis.
• Patient selection utilized 49 standard Systematized Nomenclature of Medicine condition IDs descended from endometriosis in the Observational Medical Outcomes Partnership Common Data Model schema.

TAKEAWAY:

• Analysis revealed 661 significantly enriched comorbidities at UCSF spanning multiple disease categories, with the strongest associations found for uterine adenomyosis (odds ratio [OR], 181), pelvic peritoneal adhesions (OR, 51.1), and noninflammatory disorders of female genital organs (OR, 30.2).
• Researchers identified 302 conditions (45% of the complete set) that were significantly enriched across UCSF and UC-wide datasets, with statistically significant correlation of the log ORs (Pearson r = 0.864; P = 2.38 × 10^-91).
• According to the authors, protective associations with hyperlipidemia (OR, 0.67) and mixed hyperlipidemia (OR, 0.67) in the UC-wide dataset are particularly interesting, given literature identifying statins as potential therapeutic avenues.
• The researchers found that migraines remained a significant association both before and after endometriosis diagnosis, suggesting potential shared pathways or treatment opportunities.

IN PRACTICE:

“Integrating genomic, clinical, and patient-reported data with EHR-based findings may further enhance our understanding and ultimately support the development of targeted diagnostic tools and treatment strategies, including with novel machine learning approaches. By advancing knowledge of endometriosis and its comorbidities, this research contributes to ongoing efforts to improve patient care, reduce diagnostic delays, and address the significant burden of this disease,” the authors of the study wrote.

SOURCE:

The study was led by Umair Khan, Bakar Computational Health Sciences Institute, UCSF. It was published online in Cell Reports Medicine.

LIMITATIONS:

The researchers acknowledged several inherent data issues, including missing information, patients moving between healthcare systems, and coding differences across institutions. The analysis is geographically constrained to medical centers in California, potentially limiting applicability to other regions or populations. Additionally, the case-control design precludes conclusions about causality or temporality in the observed associations. The selection criteria were deliberately permissive, defining endometriosis cases based on EHR-documented diagnoses rather than surgically confirmed cases, which may have introduced misclassification bias.

DISCLOSURES:

The study received support from the Eunice Kennedy Shriver National Institute for Child Health and Human Development, P01HD 106414, and the National Institute of General Medical Sciences, T32GM067547 and T32GM142516. Linda C. Giudice disclosed serving as a consultant to Myovant Sciences, Gesynta Pharma, Celmatix, NextGen Jane, and Chugai Pharmaceutical Co. The remaining authors reported no competing interests.

This article was created using several editorial tools, including AI, as part of the process. Human editors reviewed this content before publication.