RESUMO
INTRODUCTION: Up to one-third of ischemic strokes remained cryptogenic despite extensive investigations. Atrial fibrillation may be detected in a significant proportion of patients with embolic stroke of undetermined source, particularly after the introduction of implantable loop recorder in clinical practice. METHODS: We retrospectively included all the consecutive patients with embolic stroke of undetermined source referred to our units in the period November 2013 to December 2018 and in which an implantable loop recorder was positioned within 6 months from stroke event. Prevalence and predictors of atrial fibrillation were investigated. RESULTS: One hundred thirty-eight patients with embolic stroke of undetermined source fulfilling inclusion criteria were identified. The crude prevalence of atrial fibrillation at the end of observation period was of 45.7%. Incidence rates at 6, 12, 18, 24, and 36 months resulted, respectively, 31.8% (95% CI, 30.4-46.7), 38.0% (95% CI, 30.4-46.9), 42.6% (95% CI, 34.5-51.6), 46.6% (95% CI, 38.2-55.8), and 50.4% (95% CI, 41.6-59.9). On multivariate analysis, only excessive supraventricular electric activity and left atrial enlargement resulted to be significant predictors of atrial fibrillation (p = 0.037 and p < 0.0001, respectively). CONCLUSIONS: Atrial fibrillation may be detected in a relevant proportion (up to 50%) of patients with embolic stroke of undetermined source if a careful and extensive diagnostic work-up is employed. Excessive supraventricular electric activity and left atrial enlargement are significant predictors of the occurrence of atrial fibrillation in these patients.