Medication errors and adverse drug events:Enhancing detection through Spontaneous Reporting Systems and Electronic Health Records

Abstract

In this thesis, we applied various approaches to create an epidemiological overview of medication errors (MEs) and to study select adverse drug events (ADEs). There was a knowledge gap in regards to previously unknown MEs characteristics, undiscovered safety signals, and lack of data standardization of spontaneous reporting systems (SRSs). Unstandardized datasets from SRSs were standardized with in-house developed software in an effort to make analyses possible. Observational data from SRSs and electronic health records (EHRs) was used to provide more insight on MEs and safety signals. This thesis has contributed to extensive characterization of MEs from observational data and unveiled potentially new safety signals to be further investigated. We conclude that while we managed to fill the knowledge gap to some extent in the field of MEs and ADEs, improvements can still be made by addressing standardization and mapping of drug names, data missingness, lack of data registration, and analysis on unstructured clinical text data from EHRs with novel methodologies such as Large Language Models (LLM).