Results
Baseline characteristics: There were 57,206 patients who had at least one record indicating an AF ablation procedure, 2,402 (4.2%) of which had a Medtronic CIED implanted prior to the first AF ablation procedure on record, and 665 (1.2%) of which satisfied the full inclusion/exclusion criteria. Of those, 417 (62.7%) patients had persistent AF (mean age 67.3 ± 9.0, 73.6% male) and 248 (37.3%) patients had paroxysmal AF (mean age 66.2 ± 9.3, 73.0% male) as defined by their AF pattern within 30 days prior to ablation (Table 1). The mean time between AF ablation procedure and last observed daily AF burden was 22.7±14.7 months.
Recurrence of Atrial Fibrillation: In patients with paroxysmal AF (Figure 1, Panel A), survival free from AF recurrence at 1-year post blanking period varied from 28.2% (using a definition of AF recurrence of >6 minutes) to 72.1% (using the definition >23 hours). Similarly, in patients with persistent AF, survival free from AF recurrence at 1-year post blanking period ranged from 24.9% (using a definition of AF recurrence of >6 minutes) to 60.0% (using the definition of 7 consecutive days >23 hours; Figure 1, Panel B).
Atrial Fibrillation Burden Reduction Following Ablation: Of the 248 patients with paroxysmal AF, the, overall median percentage of time in AF was reduced by 99.6% (from 23.7% prior to ablation to 0.1% following ablation) and the overall median daily AF burden was reduced from 0.28 to 0 hours (Figure 2). Of the 417 patients with persistent AF at baseline, the overall median percentage of time in AF was reduced by 99.3% (from 67.5% prior to ablation to 0.5% following ablation) and the overall median of median daily AF burden was reduced from 23.98 to 0 hours (Figure 3).
Monitoring Strategy Simulation: Out of the 665 patients who satisfied the inclusion/exclusion criteria, 369 (55.5%) had at least 1 year of uninterrupted daily AF observations (thus included in the monitoring strategy simulation), of which 137 had paroxysmal AF and 232 had persistent AF at baseline. Of the 138 patients with paroxysmal AF at baseline, 101 (73.7%), 72 (52.6%) and 41 (29.9%) had a least one day with AF > 6 min, AF > 6 hours and AF > 23 hours, respectively within 1-year post blanking period. Of the 232 patients with persistent AF at baseline, 174 (75.0%), 134 (57.8%) and 112 (48.3%) had a least one day with AF > 6 min, AF > 6 hours and AF > 23 hours respectively within 1-year post blanking period. Sensitivity and NPV estimates for all monitoring strategies are graphically displayed in Figure 4 (Panel A for patients with paroxysmal AF and Panel B patients with persistent AF). For patients with paroxysmal AF, the 30-day monitoring strategy had the highest estimates for sensitivity (64.8%) and NPV (50.4%), whereas for patients with persistent AF, the quarterly 7-day monitoring strategy had the highest estimates for sensitivity (71.8%) and NPV (54.1%%) for detecting AF events > 6 min. A single 7-day monitor (a strategy commonly used in clinical practice) had a low sensitivity (45.2% and 43.4%) and negative predictive value (39.5% and 37%) for detecting AF events > 6 min in patients with paroxysmal and persistent AF, respectively. Heat maps showing daily AF burden within 1 year post blanking period in patients with paroxysmal and persistent AF are shown in Supplement Figure 1 and Supplement Figure 2, respectively.