Abstract 17109: Real World Assessment of Atrial Fibrillation and Other Arrhythmias After Cryptogenic Stroke

Abstract

Introduction: Long-term monitoring for atrial fibrillation (AF) is recommended in patients, who have experienced a cryptogenic stroke (CS). Clinical trials have identified AF in ~30% of patients after 3 years of continuous monitoring with insertable cardiac monitors (ICMs). Hypothesis: In a real-world analysis from a large academic healthcare system, we sought to evaluate a CS population with ICMs and a) determine the yield of AF and subsequent initiation of anticoagulation; and b) identify the presence of other arrhythmias. Methods: We evaluated all CS patients who had received an ICM between October 2014 and April 2020. We manually reviewed all stored electrocardiograms that were automatically labeled as AF by the ICM and adjudicated them as either a) AF or b) other cardiac arrhythmia including premature atrial contractions (PAC), premature ventricular contractions (PVC), supraventricular tachycardia (SVT), or nonsustained ventricular tachycardia (NSVT). Results: A total of 84 CS patients with ICMs were included: 51% men, mean age 63 years, and mean CHA 2 DS 2 -VASc 4.1. Over a median follow-up duration of 15.7 months, there were 34 patients (40% of the cohort) who did not have any AF alerts. In the remaining 50 patients, there were 960 stored electrograms that were adjudicated. Only 154 recordings from 16 patients (19% of the entire cohort) were adjudicated as AF. Oral anticoagulation was initiated in all these patients with adjudicated AF. The remaining tracings, which had been automatically categorized by the ICM as AF alerts, represented 34 patients (40% of the cohort). These patients had other arrhythmias including frequent PACs or PVCs, SVT, or NSVT. Conclusions: Compared to clinical trials, our real-world assessment suggests that the yield of AF following CS is lower - approximately 20%. Our findings highlight the importance for reviewing device tracings given the high rates of false positive for AF. Further research to refine AF detection algorithms in ICMs is needed.