gms | German Medical Science

Introduction: Patient registries represent collections of structured patient data and allow the integration of quality-assured information from multiple sources [1], [2]. However, populating a registry is time-consuming and requires additional documentation efforts for clinicians. It is therefore desirable to automatically transfer data routinely documented in a hospital information system to a registry.

This work aims to design and prototypically implement a process for data transfer from the local hospital information system ORBIS into Open-Source Registry System for Rare Diseases (OSSE)-based registries, in compliance with data protection, technical and organizational requirements.

State of the art: OSSE is a customizable framework for setting up disease-specific registries [3], [4]. Electronic data capture (EDC) primarily occurs via a web-based user interface. In addition, OSSE offers a REST API for data import via a registry-specific XML format.

To make routine data available for research, the Frankfurt Data Integration Centre (DIC) was founded as part of the Medical Informatics Initiative. Standard operating procedures are in place to provide standardized formatted data upon a data use request, subject to ethics and privacy approval.

Concept: Based on an analysis of the technical and organizational requirements of OSSE, the DIC and data protection, we created a concept for data flow from the local hospital information system into an OSSE registry. The following steps were identified: data extraction, pseudonymisation and record linkage, validation, transformation, and data import. For greater flexibility, we opted for a modular design that allows transferring parts of the process to other use cases in the future.

Implementation: Using a representative test registry and artificial dataset, the concept was implemented as an ETL (extraction, transformation, loading) process. Each step is represented as an individual module, including database queries for data extraction and Python-based scripts for transformation into an OSSE-compatible XML format and data validation against definitions in the central metadata repository. Linking a patient’s routine and registry data occurs requires communication with the registry’s identity management.

The resulting prototype is now being deployed in the “Nachsorgeambulanz Intensivmedizin - NIM” study of the Department of Anaesthesiology, Intensive Care Medicine and Pain Therapy, University Hospital Frankfurt. In this study, former ICU patients (6 months to 2.5 years post-stay) undergo a comprehensive assessment. Study-specific assessments are recorded directly in OSSE while routine examinations are documented in ORBIS (approximately 30% of data items) for later transfer to OSSE.

Lessons learned: In conclusion, we have designed and implemented a prototype of a modular ETL process to facilitate the transfer of routine data from ORBIS into an OSSE-based registry. It is currently tested in a local pilot study to be offered as a service for local OSSE registries.

Record linkage and the lack of data standards were identified as major challenges for the implementation. Although many steps are automatable, the process still requires manual input for mapping routine and registry data. Based on future developments of OSSE, we will further standardize the process, using interoperability standards like FHIR and common terminologies. This will also facilitate the future application of the process in other use cases.

The authors declare that they have no competing interests.

The authors declare that a positive ethics committee vote has been obtained.