Towards reliable prevalence estimates for chronic conditions: a data linkage study

Abstract Background Reliable prevalence estimates of chronic conditions are essential for understanding population health, and developing guidelines and policy to keep healthcare affordable and sustainable. Electronic health records (EHR) from general practitioners (GPs) and claims data from health insurers are increasingly used for this purpose in the Netherlands. But to what extent do these sources complement each other? We investigated the degree of overlap and complementation between two data sources for prevalence estimates of chronic conditions. Methods EHR data from 919,619 patients of GPs participating in the Nivel Primary Care Database were linked at the individual level to claims data available at the National Healthcare Institute. Standardized prevalence estimates per 1,000 patients for 35 chronic conditions were calculated for 2019, based on single sources and combined. Subsequently, the percentage of overlap and complementation between prevalence estimates from both data sources was calculated. Results For all 35 conditions, the combined prevalence was higher than that based on a single data source. Overlap varied per chronic condition, ranging from 6% (chronic shoulder complaints) to 92% (diabetes mellitus). Of all identified cases per condition, EHR data complements claims data between 4% (AIDS/HIV) and 83% (liver diseases), while claims data complements EHR data between 1% (diabetes mellitus) and 54% (migraine). Conclusions Data linkage between EHR and claims data results in higher prevalence estimates for chronic conditions, but the extent to which data sources complement each other differs strongly per condition. These findings offer valuable insights to researchers and healthcare policymakers for improving the reliability of prevalence estimates and supporting the development of sustainable healthcare policies. Key messages • These findings imply that researchers and healthcare policymakers should be aware that epidemiologic measures based on a single data source could be an underestimation. • These findings indicate that, depending on the chronic condition, data linkage of EHR and claims data could provide added value for informing sustainable healthcare policy.