VOLUME 116, ISSUE 6, P1580-1589
Arnaud Fauconnier, M.D., Ph.D., Hocine Drioueche, M.Sc., Cyrille Huchon, M.D., Ph.D., Joseph Du Cheyron, B.Sc., Emilie Indersie, Ph.D., Yasmine Candau, M.B.A., Pierre Panel, M.D., Xavier Fritel, M.D., Ph.D.
To assess the value of a self-completed questionnaire based on patients’ verbal descriptors of pelvic painful symptoms to identify women with endometriosis.
Prospective 1:2 nonmatched case-control study.
Three French endometriosis referral centers.
Endometriosis cases were women aged 18–45 years with endometriosis confirmed by histology. Controls were as follows: asymptomatic women attending a gynecologic consultation for routine examination; women without evidence of endometriosis consulting for pain/infertility; and population-based controls from the same urban locations.
All women completed the 21-item yes/no questionnaire about painful symptoms.
Main Outcome Measure(s)
The area under the receiver operating characteristic curve of the full question set model based on binary logistic regression and the diagnostic accuracy of low- and high-risk classification rules based on selected threshold of the prediction model.
We included 105 cases and 197 controls (45 asymptomatic consultation-based controls, 66 women without endometriosis consulting for pain/infertility, and 86 population-based controls). The full question set prediction model, including age, had an area under the receiver operating characteristic curve of 0.92 (95% confidence interval, 0.87–0.95) after internal validation. The high-risk classification rule had a specificity of 98.0% and a positive likelihood ratio of 30.5. The low-risk classification rule had a sensitivity of 98.1% and a negative likelihood ratio of 0.03. For a hypothesized pretest prevalence of 10%, the high- and low-risk prediction rules ascertained endometriosis with posttest probability rates of 77.2% and 0.3%, respectively.
A self-completed patient-centered questionnaire can identify women at low or high risk of endometriosis with a high diagnostic accuracy and, thus, may help early identification of women with endometriosis.