Abstract 150: Outpatient Cardiac Telemetry Detects A High Rate of Atrial Fibrillation Among Patients with Cryptogenic Cerebral Ischemia
Introduction: In one-third of patients discharged with stroke or TIA the etiology of the event remains undetermined. Occult paroxysmal atrial fibrillation (PAF) was reported as a potential etiology. The factors predicting PAF in these patients have not been defined.
Hypothesis: 1) A high rate of PAF can be detected during 21 days of mobile cardiac outpatient telemetry (MCOT). 2) There will be a set of risk factors which can predict the detection of PAF with MCOT.
Methods: We performed a retrospective analysis on patients evaluated by MCOT monitoring within 6 months of a cryptogenic stroke or TIA. The ECG data were analyzed with an algorithm verified against the MIT-BIH arrhythmia database. Multivariate analysis with survival regression analysis was performed using demographic, clinical, neuroimaging and cardiac imaging characteristics to determine predictive risk factors for detection of PAF on MCOT. Kaplan-Meier estimates were computed for the 21-day PAF rates.
Results: We analyzed 156 records; 97% of patients were not on anticoagulation. PAF occurred in 27 of 156 (17.3%) patients during MCOT monitoring of up to 30 days. The rate of PAF detection significantly increased from 3.9% in the initial 48hrs, to 9.2% at 7 days, to 15.1% at 14 days, and to 19.5% by 21 days (p<0.05) (figure). Multivariate analysis defined several independent predictors of PAF detection (table). Premature atrial complex (PAC) on baseline ECG held the strongest correlation with PAF (HR 13.65, p=0.001).
Conclusion: MCOT is a robust diagnostic tool for the identification of PAF in patients with cryptogenic stroke and TIA. Length of monitoring is strongly associated with the likelihood of detection of PAF. Female gender, PAC on ECG, increased left atrial diameter (LAD), reduced left ventricular ejection fraction (LV EF) and greater NIHSS increase the likelihood of PAF detection.
- © 2012 by American Heart Association, Inc.