Development of a Score to Predict the Paroxysmal Atrial Fibrillation in Stroke Patients: The Screening for Atrial Fibrillation Scale

Background and PurposeAn individual selection of ischemic stroke patients at higher risk of atrial fibrillation (AF) might increase the diagnostic yield of prolonged cardiac monitoring and render it cost-effective. MethodsThe clinical, laboratory, and brain/cardiac imaging characteristics of consecutive ischemic stroke patients without documented AF were recorded. All patients underwent at least 72 h of cardiac monitoring unless AF was diagnosed before, transthoracic echocardiogram, blood biomarkers, and intracranial vessels imaging. A predictive grading was developed by logistic regression analysis, the screening for atrial fibrillation scale (SAFE). ResultsA total of 460 stroke patients were analyzed to develop the SAFE scale, a 7-items score (possible total score 0-10): age >= 65 years (2 points); history of chronic obstructive pulmonary disease or obstructive sleep apnea (1 point); thyroid disease (1 point); NT-proBNP >= 250 pg/ml (2 points); left atrial enlargement (2 points); cortical topography of stroke, including hemispheric or cerebellar cortex (1 point); and intracranial large vessel occlusion (1 point). A score = 5 identified patients with paroxysmal AF with a sensitivity of 83% and a specificity of 80%. ConclusionScreening for atrial fibrillation scale (SAFE) is a novel and simple strategy for selecting ischemic stroke patients at higher risk of having AF who can benefit from a more thorough etiological evaluation. External validation of SAFE in a multicenter study, with a larger number of patients, is warranted.