Adverse events in medicine: easy to count, complicated to understand, and complex to prevent.

The performance of patient safety initiatives has not met expected targets for reasons that are gradually being understood. They have been too hospital-centered and too process- and "silo"-driven in their search for the causes of adverse events (AEs). Information technology could help overcome many obstacles, but only if the tools developed are based on a relevant safety model. We have applied the distinction between easy, complicated, and complex problems and strategies in healthcare to changes that need to be made in the detection and analysis of AEs. We propose a triple shift: (i) adopting an outcome-driven rather than a process-driven policy when defining and counting AEs (relatively easy), (ii) applying a patient- and not silo-driven approach and extending the timeframe when analyzing AEs (more difficult), and (iii) taking a systemic view of all care delivered to a patient during their life-span in order to erect barriers against the risks identified (highly complex).