Screening for endometriosis: a scoping review of screening measures that could support early diagnosis

Background Endometriosis is prevalent in approximately 6–10% of all women of reproductive age and is associated with pelvic pain, heavy menstrual bleeding, infertility, and pain during intercourse. Despite reporting symptoms, women wait around 11 years before receiving a diagnosis, further interfering with their mental and physical health. Patient reported screening measures can promote faster diagnosis, however their measurement quality remains unknown. Our objective was to identify and assess the measurement properties of endometriosis screening tools in a clinical setting. Methods We searched Medline, Embase, and CINAHL from January 2010 until February 15th, 2024, as well as the reference list of all included studies. Two reviewers independently assessed eligibility at all stages of the review. Study quality was assessed with a modified COSMIN framework and stoplight system in which the measurement properties of each Patient Reported Outcome Measure (PROM) were ranked and scored as positive (green), negative (red), or unknown (yellow). Results Of the 6082 studies that were collected, 24 were assessed for eligibility and eleven PROMs met our inclusion criteria and had their data extracted. A majority of the included studies assessed very few measurement properties (e.g., measurement error, structural validity, construct validity or responsiveness, etc…) of the PROM, leaving their quality unknown. The ENDOPAIN-4D received a positive rating in six out of ten measurement properties, ranking highest among the included studies. A Machine Learning Algorithm (MLA) developed by Bendifallah et al. (2022) also received good content and criterion validity, however required both patient report and clinical indicators. Conclusion Of the included PROMs, the ENDOPAIN-4D was found the be the highest quality and could be adopted for a primary care setting. While the MLA could be used in a tertiary or specialist care setting reliance on more advanced data. However, like most studies included, the scope of its application is limited due to the potential homogeneity of ethnicity, gender, and socioeconomic status of the sample.