Biases in a medical incident causation database: a quantitative evaluation using PRISMA-Medical

The incident database of the Dutch Health Care Inspectorate (IGZ) was suspected to be biased. To determine the type and extent of such biases three empirical analyses have been performed. In these analyses the current working method of the IGZ regarding the analysis and registration of incidents has been compared with PRISMA-Medical. The results of the analyses indicate that the number of registered root causes per incident is too small in the current incident database. Moreover, human failure is overrepresented, whereas organisational (latent) failure is underrepresented. When PRISMA-Medical is used for the analysis and registration of incidents, a more valid and reliable insight into the causes of incidents can be obtained