gms | German Medical Science

Introduction: Routine medical data from general practitioners are an eminently important source of information for health services research. Although almost all data of routine health care are nowadays recorded as electronic patient records, access to these data for research purposes in Germany is limited. This can be explained by three largely unsolved challenges: (1) data privacy protection legislation, (2) technical, and (3) procedural difficulties. The project RADAR (Routine Anonymized Data for Advanced Ambulatory Health Services Research) is looking for solutions to overcome these challenges to achieve secondary use of computerized medical records for research [1].

In the project, access to data stored in the software systems of German general practitioners is obtained via the file-based export Behandlungsdatentransfer (BDT) interface. BDT was originally designed in 1994 for the complete migration of all electronic health records between software systems of general practitioners in case of a software product change. Despite its age and the different original purpose, the BDT interface has the potential to make ambulatory health records accessible for research purposes.

Methods: A central requirement for processing BDT data stems from data protection requirements: BDT data must be processed on-site before leaving the general practice for research. The RADAR project explores and compares two scenarios: providing (a) an anonymous selection of treatment data without informed consent of patients and (b) an exclusive export of pseudonymous data from patients with informed consent.

Although previous research projects also processed BDT data [2], [3], [4], no software solution matched our requirements for parsing BDT into an object-oriented representation. Therefore, to make health records accessible within both scenarios, we implemented a BDT parser in Java programming language that uses an object-oriented model created according to the BDT specification [5].

Results: The implemented BDT parser successfully processes BDT files from general practitioners into object-oriented models. These models provide several advantages for further processing. Implemented helper functions facilitate the programming activity and data conversion. This is an important requirement for the technical implementation of the RADAR project, as software interfaces are aimed to be connected. Within the scenarios pseudonyms are generated by connecting the RADAR software with the web interface of a Trusted Third Party. Anonymization is achieved by using the programming application interface (API) of the anonymization software ARX [6]. Further, the presence of meta-information of the specification in the virtual model allows validation of field lengths, data type and rule checks of the field syntax and the sentence order structure.

Discussion: The workflows developed for the implementation of a BDT data export out of the general practice into a data storage center are complex due to high demands for data privacy protection and data security. Processing BDT data into an objectified model makes it possible to react flexibly to project requirements.

The developed BDT parser is one technical outcome of the developmental project phase in RADAR. Its evaluation and real-life application will be conducted in RADARs proof of concept phase, where data from real general practitioners are collected and evaluated in the two exemplary scenarios.

Acknowledgments: The RADAR project is supported by grants of the German Research Foundation (DFG) - grant numbers: HU 1587/2-1, HO 1937/7-1, RI 1000/7-1, YA 191/8-1 (29.04.2016).

The authors declare that they have no competing interests.

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