56-44: Evaluation of a High Pass Filter Designed to Reduce Oversensing in the S-ICD

Abstract

Introduction: Inappropriate shocks in transvenous ICDs are most often caused by supraventricular tachycardia (SVT), while in the subcutaneous ICD (S-ICD) the most common cause is T-wave oversensing (TWOS). We sought to evaluate the performance of a high-pass filter designed to reduce cardiac oversensing in the S-ICD. Methods: The algorithm was tested on a dataset of 626 adjudicated episodes (161 TWOS, 328 SVT, 137 VT/VF) from 161 patients. Episodes were evaluated for appropriate decision to treat. TWOS episodes were tested with and without a stored normal sinus rhythm reference ECG. Each episode was run through three devices to account for test system variability, for each device generation (9 evaluations per episode). Repeated measures logistic modeling compared odds of inappropriate decision to treat across generations. TWOS episodes were evaluated at nominal settings of dual zone with rate 200/220. VT/ VF and SVT episodes were evaluated at 170/250 for maximal algorithm exposure. Results: Odds of inappropriate decision for SVT (treated) and VT/VF (untreated) remained unchanged across device generations (table). Inappropriate therapy decision for TWOS (treated) was reduced by 82% compared to the previous generation and 71% compared to generation 1. The time to detect treatable arrhythmias was 8.60 ±3.81 sec in gen 2 and 8.86 ± 3.30 sec in gen 2.5. All episodes of VT/VF undersensed were treated by at least 1 device configuration for each generation. Conclusions: A new S-ICD algorithm demonstrated significant reduction in shocks for TWOS without a reduction in sensitivity to VT/VF. Results: The validation cohort included 100 patients with and 79 patients without AF. The algorithm correctly excluded AF in all 79 non-AF patients (specificity100%). Conversely, the AF algorithm correctly identified 94 of the 100 AF patients (sensitivity 94%). AF was not detected when episodes were quite short (<8 minutes; n=4) or associated with a stable ventricular response (n=2). Conclusions: A novel RR based AF algorithm was developed and tested using publically available ECG databases. The algorithm exhibited very high sensitivity and specificity. If incorporated within existing S ICD systems, it would offer clinicians the ability to monitor for AF without requirement of a transvenous atrial lead.