To examine the validity of self-reported endometriosis and to improve the reliability of questionnaires by including endometriosis-related questions.
Analysis of survey questionnaire data.
Cohort of 26, 898 female twins aged 20-60 years at interview, who participated in either of two surveys (1998-2002 or 2005-2006).
Endometriosis diagnosis in the Swedish National Inpatient Registry (IPR).
The self-reported endometriosis diagnoses and endometriosis-related questions from a nationwide population-based twin registry were linked with the IPR. Fairly good agreement was found between the self-reported and IPR data on endometriosis. The receiver operating characteristics (ROC) curves showed fairly good predictive ability of self-reported endometriosis to have a confirmed endometriosis diagnosis in the IPR with an area under the curve (AUC) 0.79 (95% confidence interval [CI], 0.77-0.81). Further, the predictive ability increased to AUC 0.89 (95% CI, 0.88-0.90) when there was additional information about infertility and age.
Our results indicate that self-reported data on endometriosis are moderately accurate and may be useful in studies when register data are not available.