Analysing Medication Errors Reported to the Danish Incident Reporting System & Methodological Approaches in Contemporary Research

Medication errors (MEs) in healthcare are common. MEs occur in all healthcare sectors and may cause harm, disability or death. MEs are contributing to rising economic expenses in healthcare, reaching EUR 35 billion globally. To prevent MEs, healthcare authorities and drug surveillance organizations have introduced reporting systems for gathering information related to medication safety events. In Denmark, the national mandatory incident reporting system, the Danish Patient Safety Database (DPSD), was introduced in 2005. Since then, large volumes of safety reports have been collected, most of which were related to medication.

In this PhD thesis, we analysed medication safety reports from the DPSD. We used two different approaches: a descriptive (counts and percentages) in Study 1 (‘Medication incidents and medication errors in Danish healthcare: A descriptive study based on medication incident reports from the Danish Patient Safety Database, 2014–2018’), and a method of disproportionality in Study 2 (‘Identifying high-risk medications and error types in Danish patient safety database using disproportionality analysis’). Furthermore, in our literature study we identified and mapped analytical methods for ME analysis in reporting systems (Study 3, ‘Methodological approaches for medication error analyses in patient safety and pharmacovigilance reporting systems a scoping review protocol ’ and ‘Methodological approaches for analysing medication errors in incident reports: a scoping review ’). 

The objective of Study 1 was to report numbers and characteristics of medication incidents and MEs submitted to the DPSD, focusing on medications, their severity, and the trends therein. We reviewed medication-related reports from the DPSD on adult populations from all healthcare sectors over a five-year period (2014–2018). The analysis was performed at two levels. First, at the incident-report level, where we described incidents in relation to healthcare sectors, patient demographic characteristics, and overall harm. Second, at the medication level, as multiple drugs were often involved in the same incident. Out of 479,814 incident reports, 61% (n=293,536) were related to individuals over 70, and 44.6% (n=213,974) were reported from nursing homes. Harmless events accounted for 71% of the reports, and less than one percent (n=3859) was associated with severe harm or death. Paracetamol and furosemide were the most commonly reported drugs. High-risk medicines, such as blood-thinning medicines (anticoagulants), methotrexate (used in chemotherapy or treatment of autoimmune diseases), potassium chloride (to prevent or treat low levels of potassium in the blood), and morphine (used to relieve severe pain) were identified as drugs involved in incidents with severe harm or death. 

In Study 2, we aimed to analyse MEs using a disproportionality analysis to identify signals of harmful drug-medication error combinations, focusing on exploring the drugs' associations with the stages in medication processes and error types. Disproportionality analysis is a statistical tool, originally developed and used for detection of adverse drug reactions within medication safety surveillance. It has proven to be sensitive in identifying seldom errors in large databases, known for underreporting of events, missing data, and reporting bias. We applied a method of disproportionality to analyse harmful MEs identified in our first study. The analysis found signals of disproportional reporting for the majority of high-risk medications identified in Study 1 and additional information on associations of medications – ME types and stages in the medication process.

Our scoping review (Study 3) aimed to identify, explore and map the existing publications/scientific literature on methods of ME analysis in reporting systems (Papers 3 and 4). We searched studies from electronic databases such as Embase (Ovid), Medline (Ovid), Cinahl (EBSCOhost) and Cochrane Central, Google Scholar, major national healthcare safety agencies and pharmacovigilance centres’ websites. Literature published from January 2017 to August 2022 was screened and extracted by two researchers. We found European and North American countries to be highly represented. The findings from 50 extracted publications revealed that the largest proportion of publications explored ME reports in hospitals (46%), in the general population (70%), and national patient safety reporting systems (70%). We identified a range of analytical methods to analyse MEs, including quantitative (40%), qualitative (8%), mixed methods (38%), and advanced computerized methods (14%). While descriptive quantitative analyses for categorized data were most prevalent, a number of studies adopted a newer approach exploring disproportionality analysis. Content analysis emerged as a commonly used method for managing free-text data. Mixed methods combined qualitative and quantitative approaches for analyzing both categorized and free-text data. Moreover, some studies utilized advanced computerized methods such as text mining, natural language processing, and artificial intelligence.

The results of this PhD project suggest that descriptive analysis alone is limited to identifying all potentially dangerous medications and MEs situations reported to the database due to underreporting, missing data, and reporting bias. Disproportionality analysis proved to find the majority of MEs related to high-risk medications identified descriptively. Furthermore, disproportionality analysis pointed to previously unknown and risky drug-ME situations. Employing this method as an additional screening tool for medication reports might help to focus on some areas in need of attention and spur further investigation in the DPSD. The literature review identified a tendency towards using methods of disproportionality and modern computerized methods for ME analyses. Future research should examine these methods for their utilization, suitability, applicability, and effectiveness in the analysis of reporting systems.