Background: To determine the accuracy of patient self-reports of specific cardiovascular diagnoses and to identify individual patient characteristics that influence the accuracy.
Methods: This investigation was conducted as a part of the randomized controlled ORBITAL study. Patients with hypercholesterolemia were enrolled in 1961 primary-care centers all over Germany. Self-reported questionnaire data of 7640 patients were compared with patients' case report forms (CRFs) and medical records on cardiovascular diseases, using kappa statistics and binomial logit models.
Results: kappa values ranged from 0.89 for diabetes to 0.04 for angina. The percentage of overreporting varied from 1% for diabetes to 17% for angina, whereas the percentage of underreporting varied from 8.0% for myocardial infarction to 57% for heart failure. Individual characteristics such as choice of individual general practitioner, male gender and age were associated with the accuracy of self-report data.
Conclusion: Since the agreement between patient self-report and CRFs/medical records varies with specific cardiovascular diagnoses in patients with hypercholesterolemia, the adequacy of this tool seems to be limited. However, the authors recommend additional data validation for certain patient groups and consideration of individual patient characteristics associated with over- and underreporting.