Relation of sleep state to hypoxemic episodes in ventilated extremely-low-birth-weight infants

Objective: To determine whether hypoxemic episodes in ventilated extremely-low-birth-weight infants correlate with specific behavioral states. Study design: Three-hour video-electroencephalography-polysomnography was performed on 13 ventilated extremely-low-birth-weight infants with mean postconceptional age of 28.3 weeks. The electroencephalogram was scored for discontinuity. Rapid eye movements, body, head, and limb movements were scored from synchronized video. Sleep states were defined from electroencephalography, rapid eye movements, and movement criteria. Nonparametric statistics were used to test for differences in the proportion of time hypoxemic (oxygen saturation less than or equal to85%) between behavioral states. Results: The proportions of time hypoxemic were 0.6% during quiet sleep, 4.4% during active sleep, 10.7% during indeterminate sleep, and 16.7% during arousal. There was a significant overall difference between the states (P =.004) and a significant difference between active sleep and indeterminate sleep in a pairwise comparison (P =.001). Conclusions: Higher proportions of hypoxemia were found during indeterminate sleep and arousal compared with active sleep and quiet sleep. We speculate that motor activity during sleep disruption could prevent effective mechanical delivery of ventilator breaths and contribute to episodes of hypoxemia. Our results suggest that strategies promoting uninterrupted sleep cycling analogous to the intrauterine state could improve ventilatory stability.