Extraction of respiratory myogram interference from the ECG and its application to characterize sleep-related breathing disorders in atrial fibrillation.

BACKGROUND AND PURPOSE

Present methods to extract respiratory myogram interference (RMI) from the Holter-ECG and assess effect of supraventricular arrhythmias (SVAs) onto ECG-based detection of sleep-related breathing disorders (SRBDs) and AHI estimation.

METHODS

RMI was quantified as residual energy after ECG cancellation or high-pass filtering for different windowing constellations. In 140 cases without (SET_A) and 10 cases with persistent SVAs (SET_B), respiratory polysomnogram annotations served as reference for SRDB detection from Holter-ECGs. We applied our previously published method to identify SRDBs in 1-min epochs and estimate the AHI based on joint modulations in RMI and QRS-area.

RESULTS

Sensitivity and specificity of 0.855/0.860 in SET_A dropped to 0.831/0.75 in SET_B. A significantly higher number of wake events in SET_B likely contribute to the asymmetric decrease and is consistent with a tendency to overestimate the AHI.

CONCLUSIONS

Despite reduced accuracy, RMI and QRS-area appear relatively robust against SVA and promise Holter-based detection at least of medium to severe SRBDs also in patients with SVAs.