Failure to validate the San Francisco Syncope Rule in an independent emergency department population.

Study Objective: We conduct a prospective independent validation of the San Francisco Syncope Rule to identify emergency department (ED) syncope patients with short-term serious outcomes.
Methods: This was a prospective observational cohort study of adult patients presenting to a university hospital ED with acute syncope or near syncope. Patients meeting inclusion criteria as defined in the San Francisco Syncope Rule derivation were evaluated for 5 previously derived predictor variables: abnormal ECG result, shortness of breath, hematocrit level less than 30%, triage systolic blood pressure less than 90 mm Hg, and history of congestive heart failure. Hospital admission occurred at the discretion of the emergency physician, independent of the decision rule. Follow-up occurred through contact with the inpatient attending physician for admitted patients and by telephone contact with patients not hospitalized or those hospitalized and discharged before day 7. Predetermined outcome measures as defined by the San Francisco Syncope Rule were death, myocardial infarction, arrhythmia, pulmonary embolism, stroke, subarachnoid hemorrhage, significant hemorrhage, or any condition causing or likely to cause a return ED visit and hospitalization for a related event.
Results: Complete predictor and follow-up data were available for 713 of 743 (96%) enrolled patients. Sixty-one of 713 (9%) patients met predetermined criteria for serious outcome. Sixteen of 61 (26%; 95% confidence interval [CI] 16% to 39%) patients with a serious outcome were not identified as high risk by the rule. Rule performance to predict serious outcomes was sensitivity 74% (95% CI 61% to 84%), specificity 57% (95% CI 53% to 61%); negative likelihood ratio 0.5 (95% CI 0.3 to 0.7) and positive likelihood ratio 1.7 (95% CI 1.4 to 2.0).
Conclusion: In this independent validation study, sensitivity and negative likelihood ratio of the San Francisco Syncope Rule were substantially lower than reported in the original studies and suggest that the rule has limited generalizability.