Validation of Self-Reported Disease History in Korea: A National Health Insurance Cohort Study

Self-reported disease history is widely used in epidemiological studies but is susceptible to misclassification. This study aimed to evaluate the accuracy of self-reported history of hypertension, diabetes, dyslipidemia, and stroke in Korea and to identify factors associated with reporting accuracy.

We analyzed data from the National Health Insurance Service (NHIS)-National Sample Cohort (2002-2019) for participants aged ≥ 40 years who underwent health check-ups (n = 312,367 for hypertension, 291,356 for diabetes, 289,596 for dyslipidemia, and 278,296 for stroke). Self-reported disease history was compared with medical claims records, which served as the reference standard for diagnosis history. Sensitivity, specificity, positive predictive value, negative predictive value, and observed agreement were calculated. Logistic regression models identified factors associated with false negative responses.

Self-reported prevalence was substantially lower than claims-based prevalence across all conditions: hypertension (33.9% vs. 47.9%), diabetes (15.0% vs. 36.1%), dyslipidemia (14.7% vs. 66.8%), and stroke (2.0% vs. 7.7%). Sensitivity was highest for hypertension (69.9%), followed by diabetes (40.7%), dyslipidemia (21.6%), and stroke (21.8%). Specificities exceeded 99% for all conditions. Factors reflecting greater healthcare engagement-advanced age and higher comorbidity burden-were associated with better reporting accuracy. Longer duration since initial diagnosis improved sensitivity for chronic conditions, while longer time since last hospital visit decreased sensitivity for hypertension and dyslipidemia.

Self-reported disease history in the NHIS health check-up demonstrates high specificity but variable sensitivity across conditions, with underreporting being the primary concern. Temporal factors and healthcare engagement patterns significantly influence reporting accuracy. Our disease-specific validation parameters provide essential benchmarks for bias correction in epidemiological research and clinical practice in Korea, particularly for conditions with lower reporting sensitivities.