Effect of cardiologist ECG review on emergency department practice

Study objective: To determine the effect of cardiology review of ECGs on emergency department practice. Methods: We carried out a prospective cohort study at an urban teaching ED. Our subjects were adult patients undergoing electrocardiography. We prospectively collected 1,000 consecutive ECGs and classified them by severity according to the following system: class 1, normal or minor abnormalities only; class 2, abnormalities with potential to alter case management; and class 3, potentially life-threatening abnormalities. Actual ECG readings by ED physicians (who had access to computerized interpretations at the time of treatment) were compared with those of staff cardiology quality-assurance reviewers; if they were not in agreement, an expert cardiology panel blindly chose the superior interpretation. Subsequently, an expert emergency physician panel reviewed discordant readings for discharged patients to determine the need for further action. Results: Of 1,000 ECGs, the readings for 190 (19%) were significantly discordant. The expert cardiology panel preferred the ED reading in 72 cases (38%) and the staff cardiology reading in 118 (62%). In 30 other cases no ED reading was recorded in the medical record. Of the 148 cases in which the expert cardiology panel agreed with the cardiology reading or there was no ED reading, 102 patients were admitted and 46 discharged. Of the 46 discharges, 8 cardiology readings were categorized as class 1, leaving only 38 cases in which the staff cardiology reading might have affected the ED decision to discharge a patient. All of these readings were in class 2, with the exception of one unclassifiable diagnosis. There were no class 3 readings. On expert emergency physician panel review of these 38 ECGs and interpretations, only 8 (.8%, 95% confidence interval, .3% to 1.6%) were considered sufficiently important to warrant chart review. In actual practice, none of these cases was affected by the ECG quality-assurance (QA) process. Two of these patients died during our 1-year follow-up. In one of these cases, the ECG QA process could have altered the patient's outcome. Conclusion: The existing ECG review process as mandated by the Joint Commission on Accreditation of Healthcare Organizations (JCAHO) will likely have minimal influence on patient outcomes at our institution. We should establish the effectiveness of this mandated QA process before committing scarce resources to its performance.