The Reveal LINQ ICM atrial fibrillation detection algorithm demonstrates high positive predictive value for episodes ≥1 hour, though performance varies significantly based on the underlying patient indication and episode duration.
BackgroundInsertable cardiac monitors (ICMs) are used for long-term ECG monitoring. The Reveal LINQ ICM has an improved atrial fibrillation (AF) detection algorithm.ObjectiveThe purpose of this study was to investigate the algorithm’s real-world performance in patients with syncope, cryptogenic stroke, and known AF.MethodsConsecutive patients with implanted ICM and AF detection parameters automatically set and maintained depending on the indication for monitoring were included. A single reviewer annotated all stored episodes after ICM implant. A second reviewer annotated a random sample of 10% of all detected AF episodes. The episode detection positive predictive value as well as true and false detection rates were determined for AF episodes of different durations.ResultsThe study enrolled 3759 patients (1604 43% with syncope, 1049 28% with known AF, 1106 29% with cryptogenic stroke). Overall, 20,659 AF episodes were detected in 1020 patients. The gross episode detection positive predictive value was 84%, 73%, and 26% for all episodes (≥2 minutes) and improved to 97%, 95%, and 91% for detected AF episodes ≥1 hour in the syncope, known-AF, and cryptogenic stroke patient cohorts, respectively. The true (and false) detection rate was 0.23 (0.05), 3.8 (1.4), and 0.23 (0.65) per patient-month of monitoring for the syncope, known-AF, and cryptogenic stroke patient cohorts, respectively. Limiting ECG storage to the longest detected AF episode significantly reduced the burden of episode adjudication without significantly compromising the identification of patients with true AF.ConclusionThe performance of LINQ ICM is dependent on the AF incidence rate in the population being monitored, the programmed sensitivity of AF algorithm, and the duration of detected AF episodes. Insertable cardiac monitors (ICMs) are used for long-term ECG monitoring. The Reveal LINQ ICM has an improved atrial fibrillation (AF) detection algorithm. The purpose of this study was to investigate the algorithm’s real-world performance in patients with syncope, cryptogenic stroke, and known AF. Consecutive patients with implanted ICM and AF detection parameters automatically set and maintained depending on the indication for monitoring were included. A single reviewer annotated all stored episodes after ICM implant. A second reviewer annotated a random sample of 10% of all detected AF episodes. The episode detection positive predictive value as well as true and false detection rates were determined for AF episodes of different durations. The study enrolled 3759 patients (1604 43% with syncope, 1049 28% with known AF, 1106 29% with cryptogenic stroke). Overall, 20,659 AF episodes were detected in 1020 patients. The gross episode detection positive predictive value was 84%, 73%, and 26% for all episodes (≥2 minutes) and improved to 97%, 95%, and 91% for detected AF episodes ≥1 hour in the syncope, known-AF, and cryptogenic stroke patient cohorts, respectively. The true (and false) detection rate was 0.23 (0.05), 3.8 (1.4), and 0.23 (0.65) per patient-month of monitoring for the syncope, known-AF, and cryptogenic stroke patient cohorts, respectively. Limiting ECG storage to the longest detected AF episode significantly reduced the burden of episode adjudication without significantly compromising the identification of patients with true AF. The performance of LINQ ICM is dependent on the AF incidence rate in the population being monitored, the programmed sensitivity of AF algorithm, and the duration of detected AF episodes.
Mittal et al. (Sun,) studied this question.