Terminological similarity between medical error taxonomies and incident reports: impact on taxonomy usability, reliability, and usefulness for preventing medical errors

摘要

Medical error taxonomies are commonly incorporated into incident reporting systems so that a database of incidents can be built to guide clinical risk management. Not all medical error taxonomies are the same, as different medical error taxonomies may include different categories or have a different scope of classification. Our human factors based review of medical error taxonomies indicated that the majority of published medical error taxonomies, particularly those that adopt theoretical error concepts, classified systemic factors of medical errors. In another review, we demonstrated that medical error taxonomies that vary according to degree of granularity of categories may not effectively share or integrate information with one another, despite belonging to the same domain. We conducted two studies that compared the usability and reliability of a generic and a domain-specific medical error taxonomy. In the first study, twenty nurses and twenty-one pharmacists classified two medication error incidents using a generic and a medication-specific error taxonomy. In the second study, twenty medical students used either the generic or the medication-specific error taxonomy to classify five medication error incidents. After that, we carried out a third study in order to investigate whether generic and domain-specific medical error taxonomies might lead to different interventions. Twenty healthcare workers, such as patient safety managers and officers, participated in the third study and proposed interventions based on generic and medication error classifications.The studies' findings indicate that in classifying domain-specific incidents, a domain-specific medical error taxonomy would be easier to use and more reliable, probably because the taxonomy was more terminologically similar to the domain-specific incident reports than a generic medical error taxonomy. Furthermore, interventions based on domain-specific classifications are more likely to target reported domain-specific incidents than those based on generic medical error classifications. Therefore, the terminological similarity between a medical error taxonomy and incident reports is an important consideration in adopting a medical error taxonomy. Based on the findings, we modelled how incident reporting systems can incorporate both generic and domain-specific medical error taxonomies, and modelled the relationship between the scope of a medical error taxonomy and the degree of granularity of the taxonomy's categories.