Validation of self-reported endometriosis

Abstract

Objectives

To assess the validity and reliability of self-reported diagnoses of endometriosis.

Study design

The study included 8572 women from two Australian birth cohorts (1989–95 and 1973–78), using data from the Genetic variants, Early Life exposures, and Longitudinal Endometriosis Symptoms Study (GELLES), a sub-study of the Australian Longitudinal Study on Women's Health. Validity was assessed using predictive values, and reliability was evaluated with kappa statistics.

Main outcome measures

Self-reported endometriosis diagnoses from the GELLES question “Has a doctor or healthcare provider ever diagnosed you with endometriosis?” were validated using a longitudinal survey and linked administrative records as the gold standard.

Results

The findings from the 1989–95 and 1973–78 cohorts showed good agreement for overall self-reported endometriosis diagnosis, with uncorrected kappa values of 0.67 and 0.70 (or 0.86 and 0.83 after correction for bias and prevalence effects), and positive predictive values of 0.75 and 0.84, respectively. The agreement was notably stronger for self-reported surgical diagnoses, with uncorrected kappa values of 0.87 and 0.79, and positive predictive values of 0.84 and 0.85, respectively. However, the validity of clinically suspected diagnoses was lower due to the higher occurrence of discrepancies in case assignments.

Conclusions

The findings indicate that self-reported endometriosis obtained from two questions about diagnoses made by doctors and specified methods (surgical or clinical) are valid and reliable, particularly for surgical diagnoses. Women's responses to these questions can be reliable for identifying endometriosis cases in epidemiological research. Self-reported data of clinically suspected diagnoses (i.e. without surgical confirmation) should be interpreted cautiously.