Implantable defibrillator-detected heart failure status predicts atrial fibrillation occurrence

Background: In heart failure (HF) patients, atrial fibrillation (AF) is associated with a worse prognosis. Implantable cardioverter-defibrillator (ICD) diagnostics allow continuous monitoring of AF and are equipped with algorithms for HF monitoring. Objective: We evaluated the association between the values of the multisensor HF HeartLogic index and the incidence of AF, and assessed the performance of the index in detecting follow-up periods of significantly increased AF risk. Methods: The HeartLogic feature was activated in 568 ICD patients. Median follow-up was 25 months [25th–75th percentile (15–35)]. The HeartLogic algorithm calculates a daily HF index and identifies periods of IN-alert state on the basis of a configurable threshold. The endpoints were daily AF burden ≥5 minutes, ≥6 hours, and ≥23 hours. Results: The HeartLogic index crossed the threshold value 1200 times. AF burden ≥5 minutes/day was documented in 183 patients (32%), ≥6 hours/day in 118 patients (21%), and ≥23 hours/day in 89 patients (16%). The weekly time of IN-alert state was independently associated with AF burden ≥5 minutes/day (hazard ratio [HR] 1.95; 95% confidence interval [CI] 1.22–3.13; P = .005), ≥6 hours/day (HR 2.66; 95% CI 1.60–4.44; P <.001), and ≥23 hours/day (HR 3.32; 95% CI 1.83–6.02; P <.001), after correction for baseline confounders. Comparison of the episode rates in the IN-alert state with those in the OUT-of-alert state yielded HR ranging from 1.57 to 3.11 for AF burden from ≥5 minutes to ≥23 hours. Conclusions: The HeartLogic alert state was independently associated with AF occurrence. The intervals of time defined by the algorithm as periods of increased risk of HF allow risk stratification of AF according to various thresholds of daily burden.