Are self-reported medical conditions and medicine use from middle-aged and older adults credible? A validation study comparing Danish SHARE-data with National Health Register data

Abstract

Purpose: Self-reported information on medical conditions and medicine use is widely used in epidemiological research. We investigated the validity of self-reported medical conditions and medicine use from Danish participants in the Survey of Health, Ageing and Retirement in Europe.

Methods: Self-reported health data were linked at individual level with the National Patient Registry and the Danish National Prescription Registry (n = 5572). Agreement was assessed by kappa value and presented as sensitivity, specificity, predictive values, and odds ratios.

Results: The agreement of self-reported medical conditions varied by type (κ = 0.23-0.67), lowest for cholesterol-related diseases, highest for diabetes. Self-reported medicine use varied by medication type (κ = 0.33-0.90) lowest for painkillers, highest for diabetes. Women had lower odds for self-reporting medicine use correctly but higher odds for self-reported diseases. Higher age was associated with lower agreement, while higher education with higher agreement.

Conclusions: Although Danish data from the Survey of Health, Ageing and Retirement in Europe on medical conditions and medicine vary, the data are valid for epidemiological use for most medicine data, and for diseases associated with hospitalizations or influencing everyday life. Caution is warranted for self-reported data from 70+ year olds and people with lower educational level.

Keywords: Disease; Epidemiological studies; Medicine use; Register-based; Self-reporting; Validation.