Atrial fibrillation detection by the subcutaneous defibrillator: real-world clinical performances and implications from a multicentre study.

<bold>Aims: </bold>No data exist concerning the clinical performances of the subcutaneous implantable cardioverter-defibrillator (S-ICD) atrial fibrillation (AF) detection algorithm. We aimed to study the performances and implications of the latter in a 'real-world' setting.<bold>Methods and Results: </bold>Between July 2017 and August 2019, 155 consecutive S-ICD recipients were included. Endpoint of the study was the incidence of de novo or recurrent AF using a combined on-site and remote-monitoring follow-up approach. After a mean follow-up of 13 ± 8 months, 2531 AF alerts were generated for 55 patients. A blinded analysis of the 1950 subcutaneous electrocardiograms available was performed. Among them 47% were true AF, 23% were premature atrial contractions or non-sustained AF, 29% were premature ventricular contractions or non-sustained ventricular tachycardia, and 1% were misdetection. Fourteen percent (21/155) patients had at least one correct diagnosis of AF by the S-ICD algorithm. One patient presented symptomatic paroxysmal AF not diagnosed by the S-ICD algorithm (false negative patient). Patient-based sensitivity, specificity, positive, and negative predictive values were respectively 95%, 74%, 38%, and 99%. Among patients with at least one correct diagnosis of AF, 38% (8/21) had subsequent clinical implications (anticoagulation initiation or rhythm control therapies).<bold>Conclusion: </bold>The S-ICD AF detection algorithm yields a high sensitivity for AF diagnosis. Low specificity and positive predictive value contribute to a high remote monitoring-notification workload and underline the necessity of a manual analysis. Atrial fibrillation diagnosis by the S-ICD AF detection algorithm might lead to significant therapeutic adjustments. [ABSTRACT FROM AUTHOR]