Abstract 1939: Predictors of Paroxysmal Atrial Fibrillation on Holter Monitoring in Patients with Stroke and Transient Ischemic Attack
INTRODUCTION: Atrial fibrillation is the most common cause of cardioembolic stroke. Holter Monitoring (HM) for detection of Paroxysmal Atrial Fibrillation (PAF) is being increasingly performed. No study has yet identified the high risk subsets of patients in whom HM testing might be higher yield and more cost-effective. The aim of the study is to determine the yield of HM for detection of PAF and to determine clinical predictors for a positive HM to allow more focused use of the test.
METHODS: The records of 1128 consecutive patients attending the Stroke Prevention Clinic in our institute from September 2005 to September 2006 were reviewed. All patients with definite ischemic stroke or transient ischemic attack (TIA) were enrolled. We recorded demographic data, risk factors, cardiac status, medications, symptoms, stroke or TIA type, localization, detailed clinical and neuroimaging findings, HM finding, ECG, Tranthoracic (TTE) and Tranesophageal (TEE) Echocardiography findings and therapeutic interventions. Logistic regression analysis (crude and adjusted for age) was performed to determine predictors of PAF on HM.
RESULTS: Out of 426 patients (65±14 years) (male 49.7%) with definite TIA (50.3%) or stroke (46.7%), 413 patients (97%) had HM for a mean of 22.6 hours. Bursts of definite PAF occurred in 39 patients (9.5%), all of whom were anticoagulated. Age was a significant predictor of bursts of PAF on HM (OR 1.078 for each year increase; 95% CI 1.043,1.115; p < 0.0001). After adjustment for age the other significant predictors were: number of acute (OR 1.538 for each 1 unit increase in lesion number; 95% CI 1.063, 2.226; p = 0.0223) and chronic (OR 1.858 for each 1 lesion increase; 95% CI 1.358, 2.542; p=0.0001) infarcts on brain CT, number of chronic infarcts on MRI (OR 2.8 for each 1 unit increase in lesion number; 95% CI 1.7,4.7; p<0.001), and the presence of acute cortical infarcts on MRI (OR 6.086; 95% CI 1.083, 34.21; p =0.0404). All positive HM occurred in patients over the age of 55. None of the other variables was a significant predictor.
CONCLUSIONS: Age and neuroimaging (number of infarcts and cortical infarcts) are strong predictors of PAF bursts on HM. The detection of these bursts led to a change in management in all cases and did not occur in those under 55 years of age.