A multistate model of clinical trajectories of heart failure patients with implantable electronic devices followed with remote monitoring: What outcome implications of alerts at long term follow up?

BACKGROUND: Understanding how baseline conditions influence alerts and adverse events in patients with heart failure (HF) and cardiac implantable electronic devices (CIEDs) under remote monitoring (RM) is crucial for patient management. However, the impact of HF alerts on clinical outcomes remains poorly defined. OBJECTIVE: To assess the impact of HF alerts on adverse outcomes and to identify clinical trajectories based on individual baseline risk factors. METHODS: We conducted a single-center, retrospective study including HF patients implanted with a CIED and provided with RM. We modeled patients trajectories using a semi-Markov, four-state framework with six possible transitions, encompassing a HF score alert state, HF Hospitalization (HFH) state and all-cause death. Cox proportional hazards models assessed the effects of baseline covariates on transition rates and evaluated the impact of HF alert occurrence on outcomes, treated as time-dependent covariates. RESULTS: A total of 511 patients (median age 69.9 years IQR 61.4-77.1) were included. During a median follow-up of 1.8 years IQR 0.7-3.9, 60 patients (11.7%) transitioned to an HF score alert state, of whom 6 (10%) and 12 (20%) respectively experienced an HFH and died without a prior HFH. From the baseline group, 29 patients (5.7%) transitioned directly to death and 53 patients (10.4%) had an HFH. Of the 59 total patients who experienced an HFH, 19 (32%) died during follow-up. The transition to a HF score alert state was significantly associated with all-cause death (HR 6.99, 95% CI 1.89-25.94; p = 0.004) while not statistically significant associated with HFH (HR 0.58, 95% CI 0.13-2.69; p = 0.65). CONCLUSIONS: Using a multistate model, we characterized clinical trajectories in HF patients and observed the effects of different covariates on transition rates.