How to measure harm

Amongst the many challenges we face in trying to improve safety of care is the paucity of knowledge we possess about the rate and types of harm occurring to our patients. The traditional approach has been to identify sentinel cases, either because they are obvious or they are reported via traditional risk reporting structures. This approach has many limitations. A more recent approach has been to systematically review charts looking for evidence of adverse events, which allows one to both measure the rate of harm, identify areas of greatest risk and then one can determine whether changes made are having any meaningful effect in reducing rates of harm.
A paper just published (free here) describes an alternative approach in which a trained observer directly measures rates and type of harm occurring. The authors found almost 14% of patients suffered at least one adverse event, and 6% of all patients suffered a preventable adverse event. Overall 33% of adverse events were judged to be preventable. The rate of harm varied widely by specialty with <2% of obstetric patients being harmed to 11% of medical and ICU patients being harmed. The severity of harm also varied, with almost 5% of cardiac ICU patients dying or being permanently disabled due to an AE.