Development of phenotype algorithms for the detection of adverse events in electronic health record data: a multicentre study

Clinical Snapshot

Key Findings

• P Value: Not reported

• Effect Size: Not applicable — no empirical performance data reported; this is a development study

• Primary Outcome: 82 phenotype algorithms developed for 48 adverse events using structured EHR data; 31 AEs covered exclusively by newly developed algorithms, 17 AEs by at least one modified existing algorithm

• Nnt Or Sensitivity: Not reported — sensitivity, specificity, PPV, and NPV are absent; empirical validation is planned as a subsequent study phase

• Confidence Interval: Not reported

Clinical Application

High feasibility for implementation within the German MII infrastructure given the TOP Framework's multi-format export capability and the institutional commitment of 39 sites. Feasibility in other health systems requires local adaptation of German-specific codes (PZN, OPS) and validation against local EHR data structures Direct applicability to Australian practice is limited at this stage. Australian hospitals use ICD-10-AM (Australian Modification) rather than the German ICD-10-GM, and medication coding differs substantially (PBS item codes rather than PZN). LOINC is increasingly used in Australian pathology, providing some alignment. The ACSQHC and Australian Commission on Safety and Quality in Health Care have active programs in ADE surveillance, and the conceptual framework of EHR-based phenotype algorithms is highly relevant to Australian digital health initiatives (My Health Record, ADHA). However, Australian implementation would require: (1) adaptation of ICD-10-AM codes, (2) replacement of PZN with PBS/AMT codes, (3) local empirical validation. The TGA's pharmacovigilance program could benefit from validated versions of such algorithms. RACGP and SHPA members involved in polypharmacy management would find the underlying AE taxonomy clinically relevant. Patients with structured EHR data in hospital settings where ICD-10, ATC, LOINC, and procedure coding are routinely used; primarily applicable to polypharmacy patients at risk of adverse drug events in the German MII context

Abstract

OBJECTIVES: To develop phenotype algorithms for the detection of adverse events (AEs) or AE-proxies in electronic health records (EHRs), accounting for varying data availability. DESIGN: Multicentre study conducted as part of the Use Case POLAR_MI (POLypharmacy, drug interActions, Risks) of the German Medical Informatics Initiative (MII). SETTING: Germany. PARTICIPANTS: Multidisciplinary teams from 10 German university sites within the MII. INTERVENTIONS: Not applicable. MAIN OUTCOME MEASURES: Literature- and consensus-based development and operationalisation of AE algorithms using structured EHR data, including a standardised, multicentre expert review process. Data categories used: International Classification of Diseases, 10th Revision (ICD-10) codes for diagnoses; Anatomical Therapeutic Chemical (ATC) codes and 'Pharmazentralnummern' (PZN; German eight-digit identification code for pharmaceutical products) for medications (used in the treatment of AEs); Logical Observation Identifiers Names and Codes (LOINC) for laboratory values and medical findings; and 'Operationen- und Prozedurenschlüssel' (OPS; German procedure classification) codes for medical and surgical procedures. RESULTS: We developed 82 algorithms for 48 AEs. Algorithms for the same AE varied by data categories or code selections. At the AE level, 31 AEs were covered exclusively by newly developed algorithms, and 17 AEs by at least one modified algorithm.Overall, 52 algorithms were based on a single data category, while 30 required multiple categories. ICD-10 codes were most commonly used (n=65 AE algorithms), followed by LOINC (n=27), ATC codes (n=18), OPS codes (n=11) and PZN (n=2). All phenotype algorithms were semantically modelled and can be executed using the publicly available Terminology- and Ontology-based Phenotyping (TOP) Framework, which supports export in various formats. CONCLUSION: We present a peer-reviewed set of algorithms for a large number of AEs, which can be implemented in structured routine electronic data sources and (pending validation studies) may support pharmacoepidemiologic research. The algorithms will be implementable across all 39 participating sites of the German MII. As a next step, we will empirically validate the algorithms against all information (including free text) contained in EHRs. TRIAL REGISTRATION: Not applicable.

References

1. 1.Redeker, L., Haerdtlein, A., Wermund, A. M., Mussawy, B., Rottenkolber, M., Coenen, M., Dürr, P., Federbusch, M., Jüttner, C. P., Schuster, A. K., Seidling, H. M., Uciteli, A., Beger, C., Neumann, D., Loeffler, M., Dreischulte, T., & Schmiedl, S. (2026). Development of phenotype algorithms for the detection of adverse events in electronic health record data: a multicentre study. BMJ Open. https://doi.org/10.1136/bmjopen-2025-111175