Predictors of heart failure events detected by a multisensor implantable defibrillator algorithm

Background: Cardiovascular and non-cardiovascular comorbidities are common in heart failure (HF) patients and determine disease severity and prognosis. Select modern implantable defibrillators (ICDs) are equipped with multisensor algorithms for HF monitoring. The HeartLogic index combines multiple ICD-based sensor data (heart rate, heart sounds, thoracic impedance, respiration, activity), and the associated alert has proved to be a sensitive and timely predictor of impending HF decompensation in cardiac resynchronization therapy (CRT) patients. Purpose: This analysis aims to investigate the performance of the algorithm in non-CRT patients, as well as in relation to the presence of comorbidities. Methods: The HeartLogic feature was activated in 568 ICD patients (410 with CRT) from 26 centers. The median follow-up was 25 months [25th–75th percentile: 15-35]. Results: We recorded 1200 HeartLogic alerts (0.71 alerts per patient-year) in 370 patients. Among patient characteristics, atrial fibrillation (AF) at implantation (HR: 1.62, 95%CI: 1.27-2.07, p<0.001) and chronic kidney disease (CKD) (HR: 1.53, 95%CI: 1.21-1.93, p<0.001) independently predicted alerts. HeartLogic alerts were not associated with CRT vs. non-CRT device implantation (HR: 1.03, 95%CI: 0.82-1.30, p=0.775). Comparing the combined index and all physiologic parameters during clinically stable periods we did not notice differences between CRT and non-CRT patients. Thoracic impedance was significantly lower in CKD than non-CKD patients. (46±11ohm versus 49±10ohm; p=0.047). We found a higher S3 amplitude (0.9±0.3mG versus 0.8±0.2mG; p=0.005) and nocturnal heart rate (72±9bpm versus 66±7bpm; p<0.001), and lower S1 amplitude (2.0±0.8mG versus 2.4±0.9mG; p<0.001) in AF patients vs non-AF. These differences persisted at the time of alerts (all p<0.05). In the overall population, and in patients stratified by device type, CKD and AF, we measured significant changes of all contributing sensors (paired t-test; p<0.05) from clinically stable periods to the time of alert. Conclusions: The burden of HeartLogic alerts appears similar between CRT and non-CRT patients, while patients with AF and CKD seem more exposed to alerts. ICD-measured thoracic impedance is sensitive to the fluid overload that characterizes kidney disease, as well as the f irst and third heart sound amplitudes seem sensitive to the reduced ventricular efficiency during AF. Nonetheless, ICD sensors seem to equally contribute to the HeartLogic alerts in all patient subgroups.