Article
Validation of International Classification of Diseases, 10th Revision, German Modification (ICD-10-GM) Codes Used to Detect Adverse Drug Events in Hospital Routine Data
Search Medline for
Authors
-
Nils Kuklik
- Universitätsklinikum Essen, Institut für Medizinische Informatik, Biometrie und Epidemiologie (IMIBE), Essen, Deutschland; Universitätsklinikum Essen, Zentrum für Klinische Studien Essen (ZKSE), Essen, Deutschland
-
Jürgen Stausberg
- Universitätsklinikum Essen, Institut für Medizinische Informatik, Biometrie und Epidemiologie (IMIBE), Essen, Deutschland
-
Karl-Heinz Jöckel
- Universitätsklinikum Essen, Institut für Medizinische Informatik, Biometrie und Epidemiologie (IMIBE), Essen, Deutschland
| Published: | August 29, 2017 |
|---|
Outline
Text
Introduction: Adverse drug events (ADEs) pose a burden to healthcare systems in terms of harm to patients and financial resources [1]. Although several surveillance systems are available and established, they lack completeness, suffer from acceptance problems, or are cost intensive [2]. As diseases are routinely coded with ICD-10-GM diagnoses, using this database could usefully complement the existing systems for ADE identification. To evaluate the potential of routine data in detecting inpatient ADEs, the validity of ICD diagnoses associated with ADEs was determined by calculation of positive predictive value (PPV) and sensitivity.
Methods: The study utilized routine data from four German hospitals covering the years 2014 and 2015, including secondary diagnoses coded with the ICD-10-GM. For PPV calculation, a set of ICD “code groups” previously identified as codes indicating an ADE with high certainty were analyzed [3]. 807 cases were selected by random sampling and validated by retrospective chart review. The sensitivity was generally evaluated for inpatient ADEs identified by chart review of 1,510 inpatient stays and by comparison with the routine data of the respective patient. Chart review was performed by experienced staff of the participating hospitals using standardized forms.
Results: A total of 91.2% of the reviewed ADE codes were identified as physical symptom in the patient records, and 65.1% of the cases were confirmed as ADE in the records. When analyzing the subset of ICD-10 codes comprising predominantly hospital-acquired events, a PPV for ADEs of 77.4% was calculated. For sensitivity analysis, 495 hospital-acquired ADEs were identified by chart review, of which 49.7% events were coded as an ICD diagnosis in the routine data describing the disease. A subgroup of 12.1% ADEs were further recorded as ADE in the routine data.
Discussion: The results give important insights into the data quality of routine data with regard to ADE coding. The calculated PPVs confirm its suitability for the usage in ADE detection. However, the sensitivity of inpatient ADEs was relatively low, with roughly 1 of every 8 ADEs coded as drug-associated disease. Improvements in ADE coding and patient record documentation is crucial to further increase data quality, and the reasons for under-reporting of ADEs need to be discussed.
Die Autoren geben an, dass kein Interessenkonflikt besteht.
Die Autoren geben an, dass ein positives Ethikvotum vorliegt.
References
- 1.
- Jha AK, Larizgoitia I, Audera-Lopez C, et al. The global burden of unsafe medical care: analytic modelling of observational studies. BMJ Qual Saf. 2013;22(10):809-15.
- 2.
- Lopez-Gonzalez E, Herdeiro MT, Figueiras A. Determinants of Under-Reporting of Adverse Drug Reactions A Systematic Review. Drug Safety. 2009;32(1):19-31.
- 3.
- Stausberg J, Hasford J. Drug-related admissions and hospital-acquired adverse drug events in Germany: a longitudinal analysis from 2003 to 2007 of ICD-10-coded routine data. BMC Health Serv Res. 2011;11:134.
