FHIRJoin: A Prototypical Concept for Federated Queries Across Distributed FHIR Servers

The Health Level 7 Fast Healthcare Interoperability Resources (FHIR) standard has become the widely adopted framework of the German Data Integration Centers that enables interoperable access to healthcare data through standardized REST-based interfaces. In many real-world research environments however, clinical data are often distributed across multiple independent FHIR servers. This limits the ability to execute automated search queries that span multiple data repositories. Standard FHIR servers lack native capabilities to perform aggregate distributed searches resulting in significant hurdles for researchers in cohort discovery. This paper presents FHIRJoin, a concept and prototype for enabling federated FHIR queries across distributed FHIR servers. A Python-based prototype was developed using the FastAPI framework which decomposes complex search queries into server-specific subqueries. It retrieves patient identifiers from each server and aggregates the results using set-based operations. The resulting patient cohort is ultimately used to retrieve and return a FHIR resource bundle to the client. FHIRJoin was evaluated in a distributed setup with multiple FHIR servers containing synthetic clinical data by utilizing queries provided by the DIC Frankfurt. The proposed approach provides a scalable and lightweight solution for federated cohort discovery without centralized data integration, making it suitable for distributed healthcare research infrastructures.