An accessible, non-invasive tool for endometriosis diagnosis reveals an association between age at symptom onset and endometriosis symptom prevalence

Abstract

Objective: To determine what symptom differences are prevalent in patients with differing ages of endometriosis symptom onset. Material and methods: We obtained clinical and demographic data from 1560 individuals with suspected pelvic conditions undergoing laparoscopy from the Endometriosis Patient Registry at Ponce Health Science University-Ponce Research Institute. We then generated predictive models by fitting logistic regressions to the patient data. We determined association between symptoms and age at symptom onset in patients with endometriosis by generating predictive linear and multinomial logistic regression models. Results: Our best model had an accuracy of 81.76%, with a sensitivity of 89.32% and a specificity of 64.57% at an optimal threshold of 0.75. Classic endometriosis symptoms such as dyspareunia and pelvic pain showed different prevalence rates based on patient age at onset of symptoms. Conclusion: Symptom-based predictive models are able to predict patients’ likelihood of having endometriosis in a non-invasive and accessible manner. Gynecologic and pelvic symptoms including dyspareunia and presence of uterine fibroids are significantly associated with age at symptom onset.