HeartLogic remote monitoring: Alert burden and its association with hospitalization in patients with cardiac implanted electronic devices

Abstract Background Heart failure(HF) hospitalisations impose a substantial burden on healthcare systems, with readmission rates reaching 30% at 30 days and 73% at 1-year. Early intervention of decompensation is crucial to prevent hospitalisations. Novel features in cardiac implanted electronic devices (CIEDs), such as Boston Scientific’s HeartLogic (HL), use multi-parameter algorithms to identify early HF events through remote monitoring (RM). Despite their potential, they are not widely implemented as workflows for responding to alerts are not well developed and these alerts add workload pressures on understaffed RM clinics. Purpose To evaluate HL alert burden and their association with hospitalisation outcomes. Methods We conducted a retrospective review of RM data and clinical records from consecutive patients with HL-enabled devices at a tertiary hospital CIED clinic. The study period was from January 1, 2018, to December 31, 2023. An HL alert is generated when a score from the multi parameter algorithm reached ≥16 and resolved when ≤6. Results Of 392 patients (mean monitoring period 2.25 ±1.4 years), 216 (55.1%) generated 791 HL alerts, averaging 1.4 ±1.75 alerts/patient/year. Median alert duration was 35 days [IQR 21-56] and median days-in-alert per year was 60 days [IQR 26-122]. In total, there were 88 alerts (11.3%) in 60 patients that resulted in 110 admissions (75 admissions were given the primary diagnosis of heart failure) with a median hospitalisation rate of 1.01 admissions per patient [IQR 0-2]. Higher maximum and mean HL scores (p<0.019), chronic kidney disease (p < 0.02) and both mild and moderate left ventricular systolic dysfunction (p <0.01) were significant predictors of hospitalisation. The median time to hospitalization from alert onset was 15 days [IQR 3-41], and length of stay of 6 days [IQR 3-11]. The number of alerts was significantly associated with increased HF admissions (IRR = 1.07, 95% CI: 1.00–1.15, p = 0.025). Each additional alert increased cardiac admission rates by 6.95%, suggesting alerts as markers for hospitalization risk. Significant differences were observed in maximum, mean, and median HL scores between hospitalized and non-hospitalized patients (p < 0.001). Alerts associated with admissions had a significantly longer duration (median 59 vs 39 days, p = 0.0005) with greater variability in alert duration among admitted alerts (Levene’s Test, p < 0.001). These findings suggest that prolonged alert durations may be a marker of higher hospitalization risk. Conclusion HL alerts are strongly associated with hospital admissions, demonstrating their utility in identifying early decompensation in high-risk HF patients. The frequency and prolonged duration of alerts contribute to the workload of understaffed RM clinics. Future studies should focus on optimizing RM workflows, prioritizing actionable alerts, and developing effective interventions for patients at highest risk.Table 1- Baseline Characteristics Alert Durations by Admission Status