HeartLogic Alert Burden and Clinical Yield Under Passive Real-World Remote Monitoring

Background and Aims : Multiparametric remote monitoring algorithms in cardiac implantable electronic devices (CIEDs), such as HeartLogic, can identify early deterioration in heart failure (HF), but their real-world implementation may create substantial alert workload for device clinics. We quantified HeartLogic alert burden and clinical yield in a routine device-clinic setting where alerts were available but not embedded within a standardised HF response pathway. Methods : We performed a retrospective cohort study of 357 adult patients monitored with HeartLogic-enabled CIEDs between 2018 and 2023. HeartLogic alert periods were defined from device-derived index thresholds and linked to hospitalisation data. Alert burden was quantified by frequency, duration and cumulative time in alert state. HF hospitalisation rates were compared inside vs. outside alert-periods. The "Number Needed to Alert" (NNA) was calculated as a pragmatic workload-yield metric, defined as the number of alerts requiring review to identify one temporally associated hospitalisation. Results : Over a median follow-up of 1.44 years, 223 of 357 patients (63%) experienced ≥1 HeartLogic alert. Among these patients, the median alert frequency was 1.68 alerts/year and the median cumulative time in alert state was 60 days per patient-year, equivalent to 17% of monitoring time. Across 798 discrete alert periods, the median alert duration was 35.4 days. HF hospitalisation rates were 41-fold higher during alert-periods (98.2 vs 2.4 per 100 patient-years; IRR 41.2, 95% CI 23.8-71.4, p<0.001). The NNA was 8.5 for HF hospitalisation and 4.2 for all-cause hospitalisation. Higher peak HeartLogic score, atrial fibrillation, and chronic loop-diuretic therapy were associated with greater risk of HF hospitalisation. Conclusions : In routine passive remote monitoring, HeartLogic alerts were frequent, prolonged and strongly associated with HF hospitalisation risk. These findings define the operational burden and clinical yield of HeartLogic in real-world device-clinic practice and provide a benchmark against which structured triage and heart-failure response pathways can be prospectively evaluated. The study does not assess the effectiveness of alert-guided clinical intervention.