Aims: The aim of the study was to verify in what proportion of patients, device-detected atrial high rate (AHR) episodes are indeed atrial arrhythmias (AAs). We investigated also the reasons for inappropriate arrhythmia classification and assessed if patients with misdiagnosed arrhythmias have distinct characteristics that would help to identify them.
Methods and results: The study population consisted of 304 consecutive patients implanted with cardiac resynchronization therapy defibrillators (CRT-Ds) and subsequently monitored via remote monitoring for a median follow-up (FU) of 30.5 months. Intracardiac electrograms of every recorded AHR episode were assessed and classified (AA vs. no AA) by two experienced cardiologists. During FU, 14 386 episodes of AHR were recorded and classified in 176 (57.9%) patients. In 89.2% of them, these episodes were true AA (94% atrial fibrillation, 62% de novo). The reasons for AHR misdiagnosis were atrial far-field signals (89.5%) and noise (10.5%). The mean per cent of day spent in AHR (54.9 vs. 5.86%; P < 0.001) and the occurrence of periods with low CRT pacing (82.8 vs. 55%; P = 0.003) were significantly higher in AA subjects than in those with misdiagnosed AHR. Episode duration of properly detected AHRs was longer than that of misdiagnosed AHRs. Higher per cent of time spent in AHR was an independent marker of appropriate arrhythmia detection [adjusted hazard ratio (HR) 1.04; P = 0.023].
Conclusion: Nearly two-thirds of CRT-D patients had AHR episodes within 2.5 years after implantation. Almost 90% of AHRs were indeed AA. Misdetections were caused by far-field sensing or noise. A two-step diagnostic algorithm (>9% of time spent in AHRs and episode duration >36 s) allowed for proper detection of AA with a high hit-rate and specificity.
Keywords: Atrial far-field sensing; Atrial fibrillation; Atrial high rate episode; Cardiac resynchronization therapy; Heart failure.
Published on behalf of the European Society of Cardiology. All rights reserved. © The Author 2015. For permissions please email: journals.permissions@oup.com.