Discrimination of sleep states using continuous cerebral bedside monitoring (amplitude-integrated electroencephalography) compared to polysomnography in infants

Aim: Limited two-channel electroencephalography (EEG) and amplitude-integrated EEG (aEEG) monitorings are being increasingly used; however, these measurements have not been compared with polysomnographic monitoring, the gold standard for determining infant sleep states. We aimed to determine the accuracy of two-channel EEG and aEEG recordings in defining sleep states and wakefulness in term infants compared to polysomnographic monitoring. Methods: Sleep was assessed in eight healthy term born infants (mean: 34 ± 3 days), using simultaneous polysomnography (Compumedics S-Series) and a two-channel EEG monitor (Brainz BRM2). EEG intensity, 90% spectral edge frequency (SEF), aEEG amplitude frequency bands were analysed in 30-second epochs during quiet sleep, active sleep and awake as determined by polysomnography. Results: BRM2-recorded EEG accurately identified quiet sleep from active sleep for EEG intensity (p = 0.003), SEF (p = 0.001) and aEEG amplitude (p = 0.003) and quiet sleep from awake, but not active sleep from awake. Frequency band analysis showed that wake could be identified by changes in absolute power (p = 0.015) and frequency as a percentage of total power (p = 0.03). Conclusion: We demonstrate that limited two-channel EEG monitoring can distinguish quiet sleep from active sleep and may be suitable for investigating the development of sleep in infants in the neonatal intensive care setting.