gms | German Medical Science

68. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e. V. (GMDS)

17.09. - 21.09.23, Heilbronn

Central monitoring of a decentralized registry – lessons learned from the AKTIN Emergency Department Data Registry

Meeting Abstract

  • Jonas Bienzeisler - Medizinische Fakultät der RWTH Aachen, Aachen, Germany
  • Alexander Kombeiz - Medizinische Fakultät der RWTH Aachen, Aachen, Germany
  • Lucas Triefenbach - Medizinische Fakultät der RWTH Aachen, Aachen, Germany
  • Rainer Röhrig - Medizinische Fakultät der RWTH Aachen, Aachen, Germany
  • Raphael W. Majeed - Medizinische Fakultät der RWTH Aachen, Aachen, Germany
  • Notaufnahmeregister AKTIN-Research Group - AKTIN e.V., Aachen, Germany

Deutsche Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie. 68. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e. V. (GMDS). Heilbronn, 17.-21.09.2023. Düsseldorf: German Medical Science GMS Publishing House; 2023. DocAbstr. 5

doi: 10.3205/23gmds130, urn:nbn:de:0183-23gmds1302

Veröffentlicht: 15. September 2023

© 2023 Bienzeisler et al.
Dieser Artikel ist ein Open-Access-Artikel und steht unter den Lizenzbedingungen der Creative Commons Attribution 4.0 License (Namensnennung). Lizenz-Angaben siehe http://creativecommons.org/licenses/by/4.0/.


Gliederung

Text

Introduction: The AKTIN emergency department (ED) registry gives access to electronic health records from German EDs [1], [2], [3]. As of March 2023, 58 emergency departments are participating. Data are stored in local AKTIN data warehouses (DWH) which communicate with a central component, the AKTIN Broker. Data stem from the ED information system and are imported into the AKTIN DWHs using an HL7 CDA (Clinical Document Architecture) interface [4]. Incidents such as local network changes or updates in the CDA interface CDA (Clinical Document Architecture) may cause the transfer to stop. EDs need to identify and fix the cause, then initiate a reimport to fill the data gap. The longer the downtime, the more difficult it is to locate the cause. Therefore, it is necessary to monitor participating EDs. The objective of this work is to present our monitoring concept and the lessons learned from operation.

Methods: Each AKTIN DWH periodically sends import statistics to the AKTIN broker. To monitor the process, we developed two Python scripts that act as microservices. The first script retrieves import statistics, import errors, connection and import timestamps, and stores them in a CSV file. The second script analyzes the saved CSV files, calculating error rates and checking for disconnections. The results are presented on a confluence page. If issues are identified, the AKTIN team manually notifies the affected ED.

We examined import statistics and import errors, as well as disconnections from participating EDs from January 28, 2022 to March 28, 2023. Daily Mean, median, and interquartile range (IQR) were calculated for import statistics and disconnections. Months with >90% inactivity were excluded from analysis. Based on the error message, import errors were manually classified as "validation error" when the CDA was syntactically incorrect, or "database error" when database import of a syntactically correct CDA failed.

??????Results: On average, each ED imported 116 encounters daily (median: 86, IQR: 50-113) and updated 286 cases daily (median: 41, IQR: 0-374); 20 encounters were rejected due to invalid data syntax (median: 0, IQR: 0-1). Import failure after validation due to semantics occurred daily in 1 encounter (median: 0, IQR: 0). Disconnections lasted 1D10:09:42 on average (median: 01:40:00, IQR: 00:40:00-10:55:00). Most frequent errors were during HL7 CDA validation (93.4%), with static element and value errors being common. Database errors accounted for 1.9% of errors; 4.8% could not be assigned to a category.

Discussion & conclusion: The ED monitoring effectively demonstrates the stability of the registry infrastructure. However, we observed reoccurring disconnections. The disconnection duration pertains only to the connection with the AKTIN Broker; the DWH may stay active. Effects of an inactive DWH can be mitigated by reimporting cases. A future script will automatically notify EDs when disconnections are detected.

A majority of import failures resulted from HL7 CDA validation, indicating compatibility issues or structural data quality issues. Once resolved, usually by correcting mapping errors, interfaces remained stable. A large number of failed imports was primarily due to unsuccessful reimports of individual encounters. The results of this work will be incorporated in a future data quality reporting system.

The authors declare that they have no competing interests.

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


References

1.
Ahlbrandt J, Brammen D, Majeed RW, Lefering R, et al. Balancing the need for big data and patient data privacy-an IT infrastructure for a decentralized emergency care research database. Stud Health Technol Inform. 2014;205:750-4. DOI: 10.3233/978-1-61499-432-9-750 Externer Link
2.
Brammen D, Greiner F, Kulla M, Otto R, et al. AKTIN – The German Emergency Department Data Registry – real-time data from emergency medicine. Med Klin Intensivmed Notfmed. 2022 Feb;117(1):24-33. DOI: 10.1007/s00063-020-00764-2 Externer Link
3.
Kulla M, Baacke M, Schöpke T, Walcher F, et al. Core Dataset “Emergency Department” of the German Interdisciplinary Association of Critical Care an Emergency Medicin (DIVI). Notfall Rettungsmed. 2014 Dec 1;17(8):671–81. DOI: 10.1007/s10049-014-1860-9. Externer Link
4.
AKTIN. Publications for the joint research project AKTIN [Internet]. 2023 [cited 2023 Mar 29]. Available from: https://aktin.art-decor.pub/aktin-html-20180323T201638/project.html Externer Link