Abstract
This article shows how health care organizations can analyze their reported near-miss and sentinel events using probabilistic risk analysis (PRA). The specific aims are to model risks and identify known hazards that threaten patient safety. This article provides an introduction to PRA. To demonstrate how the approach could work with small data sets, we provide a tutorial for analysis of 10 incidences of medication errors. We show that time to errors can be used to measure progress in reducing medication errors. We show that the influence of various causes of errors can be quantified on the basis of the prevalence of the cause among a handful of incidences and among the error-free cases. PRA enables health care organizations to (1) incorporate objective data into the deliberations of the safety teams, (2) gauge even small progress in incidences of sentinel events, and (3) set objective priorities for risk-reduction strategies. Critics may argue that PRA is too complicated to learn and has onerous data requirements. This is not true. We report on training of health care analysts who successfully analyzed risks within health care organization within 2 months time using the PRA. The training of these analysts and their success in applying their new learned skills show that the approach is practical.