Comparison of the Effect of Atrial Fibrillation Detection Algorithms in Patients With Cryptogenic Stroke Using Implantable Loop Recorders

Occult atrial fibrillation (AF) can be the underlying cause for cryptogenic stroke (CS). Implantable loop recorders (ILRs) have become an important tool for long-term arrhythmia monitoring in CS patients. Office-based ILR implantation by nonelectrophysiologist physicians is increasingly common. To report the real world diagnostic yield and accuracy of remote ILR monitoring in high risk CS patients, we retrospectively analyzed 145 consecutive patients with CS who underwent ILR implantation between October 2014 and October 2018 at New York University Langone Health. A certified device technician and an electrophysiologist adjudicated all transmissions. The yield and accuracy of Reveal LINQ Intra Cardiac Monitor (ICM), a fourth generation device, was compared to that of TruRhythm Detection algorithm (fifth generation device). AF was diagnosed in 17 patients (12%) over a mean follow-up of 28 +/- 12 months. The median time to diagnosis was 7.4 +/- 21.3 months. A total of 1,637 remote transmissions (scheduled- and auto-triggered alerts: 756; patient-triggered: 881) were adjudicated. The positive predictive value for AF episodes in the scheduled interrogations increased from 4% in the Reveal LINQ ICM to 16% in the TruRhythm LINQ. Of 881 patient-triggered transmissions, none were found to be true positive. In the Reveal LINQ ICM, for scheduled transmissions, primary causes of false positive (FP) were atrial ventricular premature complexes (80%). In the TruRhythm LINQ, for scheduled transmissions, primary cause of FP were T-wave over-sensing (87%). In conclusion, the real world diagnostic yield of ILR for patients with CS remains suboptimal, with at least 84% of AF alerts being FP. Patient-riggered events did not correlate with arrhythmia and the necessity of patient triggering in this population should be questioned. Expert interpretation of recordings is critical to assure accurate diagnosis. (C) 2020 Elsevier Inc. All rights reserved.