Can sleep and wakefulness be distinguished in children by cardiorespiratory and videotape recordings?

Polysomnography, including EEG recording, is the standard method to diagnose obstructive sleep apnea (OSA) in children and adults. Diagnosis of OSA would be considerably simplified if it was shown that sleep could be distinguished from wakefulness without EEG recordings. Therefore, we compared sleep/wakefulness classification using a simplified cardiorespiratory-video (CRV) method with standard polysomnography in 20 children undergoing in-hospital evaluation for OSA. The channels for the simplified montage were chosen because they (1) were suitable for unattended, home recordings, (2) allowed the diagnosis of apneas, hypopneas, desaturation, and movement/arousals, and (3) did not require attachment to the head or face that might disturb the childs sleep. Sleep staging by the two methods was blinded to results of the other method. We evaluated 21,832 30-s epochs--1,092+/-111 (SD) per child. Across 20 subjects, 79.7+/-7.1% of the epochs were sleep. The simplified montage agreed with polysomnographic classification of sleep/wakefulness for 93.8+/-2.5% of the epochs. Of all sleep epochs, 97.7 (96.4, 98.1%) median (interquartile range), were correctly classified; sleep predictive value of the CRV method was 95.2+/-2.8%. Of all epochs classified as wakefulness by polysomnography, 80.1+/-12.8% were correctly classified by the CRV method. The wakeful predictive value was 88.7+/-2.6%. Kappa values averaged 0.8+/-0.1, indicating that agreement between the CRV method and polysomnography did not occur by chance and that the level of agreement was excellent. Thus, sleep can be distinguished from wakefulness in children being evaluated for OSA using a combination of cardiorespiratory and videotape recordings. These results suggest that the CRV method would be useful in a pediatric laboratory setting where EEG recordings are not always possible. They also suggest that overnight, unattended CRV recordings in a childs own home could correctly distinguish sleep from wakefulness.