Validation of manually scored multichannel frontal electroencephalography against polysomnography in a paediatric cohort

Abstract

Polysomnography is the only internationally recognized method to diagnose paediatric obstructive sleep apnea, thus, simpler and more cost-effective diagnostic tools are urgently needed. This study aimed to validate the manual scoring of frontal self-applicable electroencephalography against polysomnography in a paediatric cohort. The polysomnography and the frontal electroencephalography were simultaneously recorded for 1 night (n = 102) in 10-13-year-old children. Scoring was performed according to the American Academy of Sleep Medicine rules, with minor adjustments to the frontal electroencephalography. Manual scorings of sleep stages were compared in an epoch-by-epoch manner using Cohen's kappa (κ) and confusion matrices using three different models: the three-stage (wake/non-rapid eye movement/rapid eye movement), the four-stage (wake/sleep stage 1 + sleep stage 2/deep sleep Stage 3/rapid eye movement) and the five-stage model (wake/sleep stage 1/sleep stage 2/deep sleep Stage 3/rapid eye movement). The inter-scorer agreements were assessed, and the intraclass correlation coefficient was used for common sleep variables: total sleep time, wake after sleep onset, sleep efficiency, sleep-onset latency and arousal index. Cohen's κ values for the three-stage, four-stage and five-stage models were 0.85, 0.73 and 0.70, respectively. The agreement for the sleep variables studied ranged from 0.87 to 0.99. The inter-rater agreement (n = 10) was κ = 0.78 for the polysomnography and κ = 0.70 for the frontal electroencephalography. Sleep staging from the frontal electroencephalography was comparable to that of a standard polysomnography in a paediatric cohort, and showed promising results in estimating sleep time and sleep architecture.