INTRODUCTION: Previous reports demonstrate that individual patients may have distinctive electrocardiographic patterns preceding atrial fibrillation (AF) onset. However, these observations are based on single recordings, and are thus limited by their "snapshot" derivation; it is unclear whether patterns observed on these recordings are consistent over time. We hypothesized that the use of an implantable loop recorder (ILR) would be feasible for serial, long-term characterization of electrocardiographic events preceding AF onset. METHODS AND RESULTS: An ILR was implanted in 29 patients with AF. Patients were instructed to activate the device immediately upon experiencing AF symptoms. Analysis focused on the electrocardiographic segment immediately preceding AF; each segment was classified as one of six patterns: short/long, normal sinus rhythm, absolute sinus bradycardia, relative sinus bradycardia, absolute sinus tachycardia, or relative sinus tachycardia. From a total of 318 device activations, 105 AF onsets were suitable for analysis, acquired from 21 patients over a mean follow-up interval of 7 months. Although AF was frequently triggered by an atrial premature beat that emerged during normal sinus rhythm, most patients demonstrated a variety of onset patterns. Degeneration of uniform atrial tachycardia to AF was rare. Significant delay in device activation relative to AF onset was common, as was activation in the absence of AF. CONCLUSIONS: Using the ILR, serial, long-term characterization of electrocardiographic events preceding AF onset was feasible, and typically demonstrated a varied pattern within individuals. There was significant incongruity between symptomatic and electrocardiographic AF burden. As described here, ILR technology is inadequate for the purpose of accurate characterization of AF burden.
INTRODUCTION: Previous reports demonstrate that individual patients may have distinctive electrocardiographic patterns preceding atrial fibrillation (AF) onset. However, these observations are based on single recordings, and are thus limited by their "snapshot" derivation; it is unclear whether patterns observed on these recordings are consistent over time. We hypothesized that the use of an implantable loop recorder (ILR) would be feasible for serial, long-term characterization of electrocardiographic events preceding AF onset. METHODS AND RESULTS: An ILR was implanted in 29 patients with AF. Patients were instructed to activate the device immediately upon experiencing AF symptoms. Analysis focused on the electrocardiographic segment immediately preceding AF; each segment was classified as one of six patterns: short/long, normal sinus rhythm, absolute sinus bradycardia, relative sinus bradycardia, absolute sinus tachycardia, or relative sinus tachycardia. From a total of 318 device activations, 105 AF onsets were suitable for analysis, acquired from 21 patients over a mean follow-up interval of 7 months. Although AF was frequently triggered by an atrial premature beat that emerged during normal sinus rhythm, most patients demonstrated a variety of onset patterns. Degeneration of uniform atrial tachycardia to AF was rare. Significant delay in device activation relative to AF onset was common, as was activation in the absence of AF. CONCLUSIONS: Using the ILR, serial, long-term characterization of electrocardiographic events preceding AF onset was feasible, and typically demonstrated a varied pattern within individuals. There was significant incongruity between symptomatic and electrocardiographic AF burden. As described here, ILR technology is inadequate for the purpose of accurate characterization of AF burden.