Electrocardiogram findings in emergency department patients with syncope.

Objectives: To determine the sensitivity and specificity of the San Francisco Syncope Rule (SFSR) electrocardiogram (ECG) criteria for determining cardiac outcomes and to define the specific ECG findings that are the most important in patients with syncope.

Methods: A consecutive cohort of emergency department (ED) patients with syncope or near syncope was considered. The treating emergency physicians assessed 50 predictor variables, including an ECG and rhythm assessment. For the ECG assessment, the physicians were asked to categorize the ECG as normal or abnormal based on any changes that were old or new. They also did a separate rhythm assessment and could use any of the ECGs or available monitoring strips, including prehospital strips, when making this assessment. All patients were followed up to determine a broad composite study outcome. The final ECG criterion for the SFSR was any nonsinus rhythm or new ECG changes. In this specific study, the initial assessments in the database were used to determine only cardiac-related outcomes (arrhythmia, myocardial infarction, structural, sudden death) based on set criteria, and the authors determined the sensitivity and specificity of the ECG criteria for cardiac outcomes only. All ECGs classified as "abnormal" by the study criteria were compared to the official cardiology reading to determine specific findings on the ECG. Univariate and multivariate analysis were used to determine important specific ECG and rhythm findings.

Results: A total of 684 consecutive patients were considered, with 218 having positive ECG criteria and 42 (6%) having important cardiac outcomes. ECG criteria predicted 36 of 42 patients with cardiac outcomes, with a sensitivity of 86% (95% confidence interval [CI] = 71% to 94%), a specificity of 70% (95% CI = 66% to 74%), and a negative predictive value of 99% (95% CI = 97% to 99%). Regarding specific ECG findings, any nonsinus rhythm from any source and any left bundle conduction problem (i.e., any left bundle branch block, left anterior fascicular block, left posterior fascicular block, or QRS widening) were 2.5 and 3.5 times more likely associated with significant cardiac outcomes.

Conclusions: The ECG criteria from the SFSR are relatively simple, and if used correctly can help predict which patients are at risk of cardiac outcomes. Furthermore, any left bundle branch block conduction problems or any nonsinus rhythms found during the ED stay should be especially concerning for physicians caring for patients presenting with syncope.