Conclusion IRM-derived AFB estimates are unreliable estimators of the true AFB. Particularly for paroxysmal AF patients, IRM-derived AFB estimates should not be used to evaluate outcomes after AF interventions.
Background Estimates of atrial fibrillation (AF) burden (AFB) derived from intermittent rhythm monitoring (IRM) are increasingly being used as an outcome measure after therapeutic interventions; however, their accuracy has never been validated. The aim of this study was to compare IRM-derived AFB estimates to the true AFB as measured by implantable continuous monitoring (CM) devices.
Methods Rhythm histories from 647 patients (mean AFB: 12 ± 22%; 687 patient·years) with CM devices were analyzed. IRM of various frequencies and durations were simulated and the obtained IRM-derived AF burdens were compared to the true AFB measured by CM.
Results The relative error of the IRM burden estimates was dependent on the IRM length (P < 0.001), frequency of IRM (P < 0.001), the true AFB (P < 0.001), and its temporal aggregation (AF density, P < 0.001). In paroxysmal AF patients, the relative error even with aggressive IRM strategies was >80% of the true AFB. The relative error decreased with higher true AF burdens, lower AF densities, and higher IRM frequency or duration (P < 0.001). However, even in patients with high AF burdens and/or low AF densities, IRM estimates of AFB significantly deviated from the true AFB (relative error >20%, P < 0.001) and resulted in a substantial measurement error.