Abstract 4146087: Danish Pragmatic Randomized Trial to Evaluate the Effect of HeartLogic-Guided Management on Heart Failure Outcomes – The DANLOGIC-HF Preliminary Registry Findings

Background: Hospitalizations due to heart failure (HF) are costly and detrimental to patient health. The HeartLogic (HL) algorithm has been found to be predictive of impending HF decompensation. Purpose: To present characteristics of the Danish nationwide DANLOGIC-HF population and preliminary findings on the performance of the HL algorithm. Methods: All Danish patients with a HL compatible device implanted after 2017 were identified. Their retrospective HL index values were used to calculate active alert periods. HF status and other diagnoses were based on ICD10 diagnoses codes, and patients with a CRT-D were categorized as having known HF regardless of ICD10 diagnoses. The association between HL alerts and HF hospitalizations were assessed using a Cox proportional hazard model with alert status as a time-dependent covariate and clustered standard errors. Results: In this population of 1599 adult patients with a CRT-D or ICD, mean age was 59.6 years,75% were male, and 26.9% had ischemic heart disease. While 43% had known HF, only 24% of the population had a CRT-D. Aside from diabetes (7.4% vs 2.7%, p<0.001) and chronic obstructive pulmonary lung disease (3.5% vs 0.5%, p>0.001) being more frequent in patients with HF, the prevalence of cardiovascular comorbidities was similar between patients. Patients with HF had a lower left ventricular ejection fraction (LVEF) (29% vs 44%, p<0.001) and higher NYHA class at implantation compared to patients without HF. Patients with HF also had a higher HF-related hospitalization rate (30 vs 7 per 100 patient-year, p<0.001). At a mean follow-up time of 2.9 patient-years, HF at baseline and LVEF significantly modified the association between HL alert status and risk of HF hospitalization. The hazard ratio of being hospitalized with HF while in active HeartLogic alert was 7.9 [5.7, 11.1] for the entire population, 5.4 [3.8, 7.6] for patients with known HF and 14.7 [6.0, 36.4] for patients without known HF (p for interaction <0.001). For patients with a LVEF<40%, the hazard ratio for being hospitalized with HF if having an active HL alert was 7.1 [4.9, 10.6] compared with 33.7 [7.3, 157] for patients with a LVEF>40% (p for interaction = 0.026). Conclusion: Being in active alert was associated with a high risk of being hospitalized with HF, especially among patients without known HF. These findings emphasize the potential utility of HL-guided management in improving HF outcomes, both amongst patients with and without known HF.