Privacy-preserving record linkage by a federated trusted third party (fTTP) – unlocking medical research potential in Germany

Introduction: In the context of the COVID-19 pandemic, the urgent need for centralized research infrastructures became clear in Germany. Based thereon, large multicenter medical research projects, such as the Network of University Medicine (NUM) were initiated with the aim to collaboratively consolidate and harmonize the medical data of the individual healthcare providers for research purposes. As a part of the Medical Informatics Initiative (MII), data integration centers (DIC) have already been established at all university hospitals to provide standardized data sets. In order to comply with the relevant data protection requirements, a privacy-preserving record linkage (PPRL) is required to enable cross-site merging of patient records. A federated trusted third party (fTTP) was implemented. Material and methods: Generic use cases were identified, conceptualized and implemented and provide the basis for the research work. Different scenarios such as rare diseases were considered, with the option to extend the PPRL using additional record linkage methods. Existing tools such as identity and pseudonym management systems were expanded regarding their respective functionalities. Results: The fTTP enables PPRL and pseudonymization of patient records. It uses well-established PPRL methods. A pseudonymization hierarchy enables the individual’s re-identification e.g., in cases in which incidental findings require contacting a participant in research projects. A clearing procedure is set up to manage potential matches in order to minimize homonym and synonym errors. Conclusion: The fTTP enables a unified, project-specific pseudonymization across multiple participating sites and PPRL. The fTTP enables secure linkage of further medical data sources, such as data from clinical studies or claims data, on a patient-specific basis. The outlined concepts and technical implementation can serve as blueprint to further use cases.