論文アブストラクト：Interruptions in the emergency department (ED) can have serious patient safety consequences, and few solutions exist to mitigate the disruptiveness of interruptions. We developed a theoretically motivated model to predict the likelihood of emergency physicians returning to an interrupted task. Eighteen emergency physicians were observed individually for two-hour blocks of time, resulting in a total of 2160 minutes of observation and 231 interruptions. We used a mixed effects logistic regression model to predict the likelihood of primary task resumption after interruptions. The likelihood of primary task resumption was predicted by memory decay, measured by the duration of the interruption, workload, measured by the patient volume during the shift, and whether shift was day or night. With a better understanding of these interruptions, we can help design interventions to manage interruptions, minimize medical errors, and improve patient safety.