gms | German Medical Science

Introduction: Typically, structured observational data of particular diseases for epidemiological and clinical research are collected in clinical registries. From our experience, the technical implementation and the design of such registers are a lengthy process that can easily take months. However, during the SARS-CoV-2-pandemic it was required to be immediately able to collect data in two newly established research databases and later link it to data from the RWTH Aachen Biobank and the data integration centers established during the SMITH Project [1], [2]. Potentially, the data should be connected to non-local research projects e.g. the German medical informatics initiative (MII).

Based on a two-step approach we thus drafted a primitive, yet fully functional manual setup mimicking out-of-the-box solutions, that can potentially be advanced into a state-of-the-art clinical registry.

State of the Art: The substructure for many registries in Germany are the generic concepts of the telematics platform for medical research networks (TMF) [3]. The TMF and the MII are creating a framework for the sustainable use of healthcare data and advise the use of FHIR and I2B2 for clinical research databases.

Concept: We conceptualized an intermediate layer for pooling local data sources and potentially connecting them to non-local sources. We first identified processes that could be handled manually and with out-of-the-box solutions for fast set-up. The dataflow, pseudonymization in cooperation with a trusted third party, and consent management will be an implementation of the generic TMF concepts [3]. These can be implemented as proprietary files, that can be distributed ad-hoc but also allow for the later use of the (automated) open-source solutions developed in the mosaic project [4]. The only technical prerequisite for data-linkage and data capture is a patient log consisting of full name and birthday as well as a completed patient consent.

Data will be later collected in a data warehouse employing the OMOP common data model. The data are processed in an ETL pipeline, that can be first implemented as a script and subsequently integrated into an ETL layer. Perspectively, the collected data can be copied into the data integration centers by providing a FHIR facade for the OMOP CDM [5].

Implementation: After conceptualizing our approach, we start first data capture for linkage after ten days of preparation and intend to implement the DWH for data integration within a few weeks. Study protocols and consent forms can now be finalized accordingly. Studies Using this dataset are published, for example [6], [7].

Lessons learned: We drafted a research infrastructure, the COvid19RegisterINtegrationAachen (CORINA), as an intermediate layer for the sustainable capture of data. In a crisis, immediate roll-out has priority to capture data electronically at all. A manually working framework is required, as it determines the study protocol and the design of the informed consent, blocking data capture. Technical deficits cannot be avoided. The technical architecture has to be chosen in a manner, that doesn't block later development; additional resources have to be budgeted for revisal. Thus, the design of technical and organizational measures is a balance of maintaining structural prerequisites for later development and rapid set-up in small steps.

[*] Covid-Aachen-STC: Marx N, Dreher M, Marx G, Müller-Wieland D, Dahl E, Kuhl C, Röhrig R, Stingl J, Uhlig S

The authors declare that they have no competing interests.

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