Triage of patients with chest pain in the emergency department: a comparative study of physicians' decisions.
Academic Article
Overview
abstract
PURPOSE: Little is known about physicians' triage decisions for patients with chest pain in the emergency department. We sought to understand better the variability and accuracy of physicians' triage decisions. SUBJECTS AND METHODS: We used 20 simulated cases to compare triage decisions by 147 physicians (46 emergency medicine, 87 internal medicine, and 14 cardiology physicians) with triage decisions recommended by a previously validated prediction rule. We calculated triage sensitivity and specificity using the prediction rule to estimate the likelihood that each of the simulated patients would suffer a major complication. Triage sensitivity was defined as the proportion of all patients expected to have major complications who were triaged to the coronary care or inpatient telemetry unit. RESULTS: Triage specificity was defined as the proportion of all patients without complications who were triaged to sites other than the coronary care or inpatient telemetry unit.Physicians' triage decisions were less sensitive (85% vs. 96%, P <0.001) and less specific (38% vs. 41%, P = 0.02) than decisions recommended by the prediction rule. Physicians overestimated patients' risk of complications and triaged more patients to inpatient monitored beds. Despite their preference for inpatient monitored beds, physicians' decisions would have resulted in four times as many major complications in patients who were not triaged to inpatient monitored beds, compared with decisions recommended by the prediction rule (2.4% vs. 0.6%, P <0.001). Although physicians' decisions were best explained by their provisional diagnoses, interphysician agreement about triage decisions (kappa = 0.34) and diagnosis (kappa = 0.31) was only fair. CONCLUSIONS: In simulated cases, physicians' triage decisions varied widely and their predictions of patient outcomes differed markedly from that of the validated prediction rule, suggesting that use of the prediction rule in the emergency department could improve physicians' decisions and patients' outcomes.
