Development of a comprehensive real-world evidence dataset to inform diabetes policy: the diabetes data cell project.

backgroundDiabetes is a highly prevalent non-communicable disease with a growing global burden, posing significant challenges to healthcare systems. Effective management requires comprehensive monitoring of prevalence, risk factors, and outcomes across diverse diabetes populations. While several European countries have established or are establishing national diabetes registries to guide care and policy, Belgium's diabetes data landscape remains fragmented, lacking a unified national diabetes registry. Existing datasets cover limited populations and care settings, hindering a complete understanding of diabetes care and prevalence. To address this, the Belgian Diabetes Forum initiated the Diabetes Data Cell Project, aiming to integrate multiple healthcare datasets and create a comprehensive national diabetes registry.

methodsAn inventory of existing databases was compiled, and the quality of the datasets was assessed based on predefined criteria. A selection of core datasets was made to ensure reliable identification of individuals with diabetes, based on clinical diagnoses and coverage of the entire healthcare spectrum. Complementary datasets were selected to enrich the core datasets with additional information on healthcare utilization and mortality. All datasets had to allow for linkage at the individual level and provide longitudinal data.

resultsWe identified two large core datasets of people with diabetes who are treated in primary care and in hospital outpatient clinics. We obtained permission from the Information Security Commission to link both sets for the period 2017-2027 at the individual level to the national insurance dataset - where all medical procedures are recorded. That insurance dataset is also linked to mortality data. The first data extraction, covering 2017-2024, resulted in a linked dataset of 76 964 unique individuals. The combined dataset provides a comprehensive overview of diabetes diagnosis and type, diabetes duration, demographic characteristics, clinical outcomes, laboratory results, medications and other treatments, medical procedures and hospitalizations, complications, healthcare utilization, mortality and costs of care.

conclusionsThis study demonstrates that large-scale linkage of Belgian health data is feasible within the current legal and technical framework and enables the identification of people living with diabetes across both primary and specialized care. By integrating clinical and administrative datasets at the individual level, the project provides a comprehensive and longitudinal view of diabetes care, supporting analyses of treatment patterns, care processes, outcomes, and healthcare utilization. This work lays the foundation for a continuously updated, 'living' national diabetes registry in Belgium and represents an important step towards more integrated, data-driven healthcare policy.