Validity of diabetes self-reports in the Women's Health Initiative

Area of Science:

• Clinical Research
• Epidemiology
• Women's Health

Background:

• Accurate self-reporting of health conditions is crucial in large clinical trials.
• The Women's Health Initiative (WHI) collected self-reported data on diabetes among participants.
• Validation of self-reported data against medical records is essential for data integrity.

Purpose of the Study:

• To determine the positive and negative predictive values of self-reported diabetes.
• To assess the accuracy of self-reported diabetes in the context of the WHI clinical trials.
• To evaluate the reliability of self-reported diabetes data for prevalent and incident cases.

Main Methods:

• Identified WHI participants who self-reported diabetes or a random sample who did not.
• Surveyed participants regarding diabetes diagnosis and treatment.
• Reviewed medical records for antidiabetes medication use or diagnostic laboratory glucose measurements.

Main Results:

• Medical records confirmed 91.8% of self-reported prevalent diabetes and 82.2% of incident diabetes cases.
• 94.5% of women who never self-reported diabetes had no medical evidence of the condition.
• Higher body mass index correlated with more accurate self-reporting of incident diabetes.

Conclusions:

• Self-reported diabetes in the WHI trials demonstrates high positive predictive values for both prevalent (91.8%) and incident (82.2%) cases.
• A high negative predictive value (94.5%) was observed for individuals not reporting diabetes.
• A claims algorithm showed high predictive values for diabetes classification in fee-for-service Medicare participants.