gms | German Medical Science

Introduction: Hospital information systems (HIS) are widely used at German hospitals to save information about patients, diagnostics, and therapies. Often this data is not saved in a structured way. Therefore, it is difficult to be compared or used for analyses. To help answer specific questions in medical research, large quantities of data have to be structured to be comparable. This is the reason why data integration centres (DIC) of the Medical Informatics Initiative (MII) transform medical data from university hospitals into structured data formats that can be used by researchers for their specific projects [1], [2].

One use case within the MII is “Polypharmacy, drug interactions, risks” (POLAR). This use case examines which combinations of medication can cause drug interactions. To analyze this problem structured medication data is provided by the DIC [3].

To access the structured data in a uniformed way all DIC implement the interoperable standard Fast Healthcare Interoperability Resources (FHIR) together with the MII core data set (KDS) [2].

Methods: The university hospital in Giessen uses the HIS Meona for documentation. For this project information about medications is extracted from the Meona SAP MaxDB database and transformed into FHIR, particularly into the FHIR resource “Medication” [4], [5]. To transform this data to FHIR resources we use the ETL-Software Mirth Connect in version 3.9 which includes a FHIR Plugin that allows the user to easily generate FHIR resources. We create a Mirth Channel in which database requests are executed, FHIR resources are built and sent to a FHIR-compliant server [6].

Results: Approximately 250.000 Medication resources were sent to the FHIR server. The generated resources contain PZN- and/or ATC-Codes for the identification of the medication. Furthermore, the strength, pharmaceutical form and all ingredients are included. To additionally identify not only the medication but also all specific ingredients ASK-Codes are also extracted and saved into the FHIR resource.

About 1.800 of the generated resources have a PZN-Code as a unique identification, all 250.000 resources contain ATC-Codes. The ASK-Code is saved in approximately 56.000 Medication resources, the rest contains a textual description of the ingredients.

Discussion: The absence of the PZN- and ASK-Code in some resources is due to the information provided by the Meona database. In this context, it will be important that future research investigates if the PZN- and ASK-Codes are needed in the evaluation of data or if the ATC-Code and the textual description of the ingredients suffices. Additionally, it should be analyzed how many of the medications without PZN- or ASK-Code are being administered to any patients and if the amount is significant for the evaluation.

Conclusion: This work is a necessary step on the way to run analysis about drug interactions in POLAR. The next step will be the construction of MedicationAdministration resources from Meona. This resource describes the administration of a medication, with the information about the patient, medication, administered dose and at what time the medication was administered.

The authors declare that they have no competing interests.

The authors declare that an ethics committee vote is not required.