48Remote monitoring of Heart Failure patients with a Multisensor ICD Algorithm: value of an alert-based follow-up strategy

Abstract Background The HeartLogic algorithm measures and combines multiple parameters, i.e. heart sounds, intrathoracic impedance, respiration pattern, night heart rate, and patient activity, in a single index. The associated alert has proved to be a sensitive and timely predictor of impending heart failure (HF) decompensation, and the HeartLogic alert condition was shown to identify patients during periods of significantly increased risk of HF events. Purpose To report the results of a multicenter experience of remote HF management with HeartLogic algorithm and appraise the value of an alert-based follow-up strategy. Methods The HeartLogic feature was activated in 104 patients (76 male, 71 ± 10 years, left ventricular ejection fraction 29 ± 7%). All patients were followed according to a standardized protocol that included remote data reviews and patient phone contacts every month and at the time of HeartLogic alerts. In-office visits were performed every 6 months or when deemed necessary. Results During a median follow-up of 13[11-18] months, centers performed remote follow-up at the time of 1284 scheduled monthly transmissions (10.5 per pt-year) and 100 HeartLogic alerts (0.82 alerts/pt-year). The mean delay from alert to the next monthly remote data review was 14 ± 8 days. Overall, the patient time in the alert state (i.e. HeartLogic index above the threshold) was 14% of the total observation period. HF events requiring active clinical actions were detected at the time of 11 (0.9%) monthly remote data reviews and at 43 (43%, p < 0.001) HeartLogic alerts. Moderate to severe symptoms of HF were reported during 2% of remote visits when the patient was out of HeartLogic alert condition and during 15% of remote visits performed in alert condition (p < 0.001). Out of 100 alerts, 17 required an in-office visit and 5 a hospitalization to manage the clinical condition. Overall, 282 scheduled and 56 unscheduled in-office visits were performed during follow-up. Any HF sign (i.e. S3 gallop, rales, jugular venous distension, edema) was detected during 18% of in-office visits when the patient was out of HeartLogic alert condition and during 34% of visits performed in alert condition (p = 0.002). Conclusions HeartLogic alerts are frequently associated with relevant actionable HF events. Events are detected earlier and the volume of alert-driven remote follow-ups is limited when compared with a monthly remote follow-up scheme. The probability of detecting common signs and symptoms of HF at regular remote or in-office assessment is extremely low when the patient is out of HeartLogic alert state. These results support the adoption of an alert-based follow-up strategy.