Implementing a federated regional diabetes register in a decentralized health system: implications for healthcare organization and the European Health Data Space

Background: Fragmentation of health information systems remains a major barrier to effective health system performance across Europe, particularly in decentralized settings. Federated data infrastructures have been proposed as a scalable solution, but evidence on their implementation at population level remains limited. Objectives: We aimed to describe the implementation of a federated information infrastructure at regional level and assess the completeness and variability of linked data across participating centers, discussing implications of scaling up the approach within the European Health Data Space. Materials and methods: We carried out a population-based cohort study linking administrative, clinical, and patient-reported data across three healthcare organizations in Emilia-Romagna, Italy, covering over 2.1 million residents. Data were analyzed using a federated architecture without sharing individual-level information. Baseline characteristics were assessed at 1 January 2019, with longitudinal follow-up over 6 years. Results: We identified 116,552 individuals with diabetes (prevalence 5.6%). Among individuals with available classification, 90.8% had type 2 diabetes. Clinical data were available for 43.5% of patients in charge of diabetes clinics, with substantial heterogeneity across centers (12.8-80.6%). Among those with clinical data, 27.5% had baseline HbA1c levels below 48 mmol/mol, while 69% had elevated systolic pressure (≥130 mmHg) and 56.6% had high diastolic pressure (≥80 mmHg). Sociodemographic variables were largely missing. Patient-reported outcomes were collected in 521 individuals, demonstrating feasibility but limited scalability. Discussion: Federated linkage of administrative, clinical, and patient-reported data is feasible at regional scale and enables population-level monitoring of diabetes care. However, variability in data completeness was primarily driven by organizational and governance factors rather than technical capacity. These findings provide empirical evidence that strengthening health data systems requires alignment of healthcare organization and service delivery models, beyond technical solutions alone. Conclusion: The REWINDER project built a collaborative information infrastructure using federated linkage of different data sources and person-reported outcomes, independently managed by local healthcare organizations. The project has made available a large database to inform policy and planning of diabetes care across the region. The system may be used as a model that can be conveniently scaled up to other geographical areas and chronic diseases.