Luna, DanielRizzato Lede, DanielRubin, LucianaOtero, CarlosOrtiz, Juan M.García, Mónica G.Rapisarda, Romina P.Risk, MarceloGonzález Bernaldo de Quirós, Fernán2019-08-212019-08-212017-08978-1614-99-829-7http://ri.itba.edu.ar/handle/123456789/1724"Decision support systems can alert physicians to the existence of drug interactions. The Hospital Italiano de Buenos Aires, Argentina, has an in-house electronic health record with computerized physician order entry and clinical decision support. It includes a drug-drug interaction alert system, initially developed under traditional engineering techniques. As we detected a high alert override rate, we rebuilt the knowledge database and redesigned the alert interface with User-Centered Design techniques. A laboratory crossover study using clinical vignettes showed that new alerts were more usable than traditional ones. This paper aimed to validate these results through a controlled and randomized experimental study with two branches (old vs. new design) in a real setting. We analyzed, quantitatively, every fired alert between April 2015 and September 2016. Finally, we performed user surveys and qualitative interviews to inquire about their satisfaction and perceptions. In real scenarios, user-centered design alerts were more usable, being more effective and satisfactory, but less efficient than traditional alerts. "Safe omission", as a new concept, emerged from our stratified analyses and interviews."enINTERACCION DE DROGASERRORES DE MEDICACIONPRESCRIPCION ELECTRONICAINTERACCION PERSONA COMPUTADORATOMA DE DECISIONESinfo:eu-repo/semantics/acceptedVersion