Design: Prospective time-series intervention study.

Setting: A university hospital emergency department.

Participants/patients: 68 physicians, all of whom were responsible for the triage of at least one of 252 patients presenting to the emergency department with a chief complaint of acute chest pain.

Intervention: A previously validated algorithmic prediction rule that was attached to the back of patient data forms in the emergency department.

Measurements: Patients’ clinical data were recorded by the examining physician in the emergency department or by a research nurse blinded to patient outcome. The physicians recorded their own estimates of the risk of acute myocardial infarction and their reactions to the prediction rule in a self-administered questionnaire completed at the time of triage.

Main results and conclusions: The physicians reported that they looked at the prediction rule during the triage of 115 (46%) of the 252 patients. The likelihood of using the prediction rule decreased significantly with increasing level of physician training. The most common reasons given for disregarding the prediction rule were confidence in unaided decision making and lack of time. The physicians reported that of the 115 cases for which the prediction rule was used, only one triage decision (1% ) was changed by it. Future research should explore how prediction rules can be designed and implemented to surmount the barriers highlighted by these data.