Estimation of atrial fibrillation burden as a function of monitoring time without atrial fibrillation

BACKGROUND: Atrial fibrillation (AF) burden is associated with adverse outcomes; in some patients, very low AF burden may not be actionable. OBJECTIVE: This study aimed to model AF burden as a function of monitoring time without AF. METHODS: We included 11,639 US patients (median age 62 years [interquartile range 46‒73], 40.6% male) without permanent/persistent AF, monitored for ≥14 days in 2021 with the PocketECG device (MEDICALgorithmics). The overall 14-day AF burden was modeled as a function of monitoring duration without AF episodes lasting ≥30 seconds to <6 minutes, ≥6 minutes to 1 hour, and ≥1 hour. Results are presented as 95th percentiles of AF burden. Confidence intervals (CIs) were bootstrapped with 10,000 replications. RESULTS: AF with a median burden of 2.6% (interquartile range 0.00‒0.13) was detected in 1186 patients (10.4%), of whom 985 had ≥1 episode ≥6 minutes, and 769 had ≥1 episode ≥1 hour. AF burden decreased rapidly with increasing monitoring time without AF. After 3 days, the 95th percentile of AF burden was 0.00% (95% CI 0.00%‒0.00%) in patients without any ≥30-second episodes, 0.008% (95% CI 0.00%‒0.02%) without any ≥6-minute episodes, and 0.08% (95% CI 0.03%‒0.18%) without any ≥1-hour episodes. The probability of an AF burden ≥0.1% after 3 days without ≥30 seconds AF was 3.1% (95% CI 2.7%‒3.4%). CONCLUSION: Monitoring time without AF can be used to predict 14-day AF burden and rule out AF burdens above certain thresholds. In this unselected patient sample, 3 days of monitoring without AF episodes ≥30 seconds exclude AF burdens ≥0.1% in 97% of patients.