RESUMO
BACKGROUND AND OBJECTIVE: Recent randomized studies have shown the potential of prolonged cardiac rhythm monitoring to detect silent paroxysmal atrial fibrillation (PAF) in patients with cryptogenic stroke. Our aim was to identify clinical and magnetic resonance imaging (MRI) factors that predict the detection of PAF during long-duration rhythm cardiac Holter (LDRCH) monitoring. METHODS: A retrospective analysis was performed using data from 171 patients with cryptogenic stroke or transient ischemic attack who underwent LDRCH monitoring (i.e., 21 days) from 2009 to 2013. Clinical, echocardiographic, and imaging predictors of atrial fibrillation (AF) were determined by multivariable analysis. RESULTS: PAF lasting more than or equal to 30 seconds was diagnosed in 26 of 171 (15%) patients. Median delay to first detected AF episode was 7 (interquartile range: 2-13) days. Four factors were independently predictive of PAF detection on LDRCH: age more than 70 years (odds ratio [OR], 4.6; 95% confidence interval [CI], 1.5-13.8); premature atrial complex on electrocardiogram (OR, 4.6; 95% CI, 1.1-19.6); left ventricular hypertrophy on transthoracic echocardiography (OR, 6.4, 95% CI, 1.6-26.4); and previous white matter lesions on brain MRI (OR, 4.2; 95% CI, 1.2-15.6). Recent brain infarction pattern on diffusion-weighted imaging was not associated with PAF detection on LDRCH. CONCLUSION: LDRCH is a noninvasive and inexpensive test with a high rate of AF detection in patients with cryptogenic stroke or transient ischemic attack. Occult PAF was more commonly diagnosed in older patients with premature atrial complex on the baseline electrocardiogram, left ventricular hypertrophy on echocardiography, and with previous white matter lesions on brain MRI.