Polysomnography in newborns and young infants: sleep architecture.

Sleep architecture derived from long-term polysomnographic recordings during the first year of life is characterized by clear developmental trends against a backdrop of variability. Variability is due to differences in state definitions and data collection and analysis strategies but probably also to an intrinsic characteristic of the maturing central nervous system (functional plasticity). Changes in sleep and wakefulness probably constitute nonspecific responses to a variety of stimuli. The variability has frustrated efforts to use specific features of sleep architecture for diagnostic or prognostic purposes. At present, polysomnographic studies of sleep architecture independent from EEG and cardiorespiratory studies are not indicated for diagnosing specific medical conditions or prognoses of good/adverse outcomes. For accurate interpretation for cardio-respiratory data, however, studies of sleep and wakefulness are indispensable. Furthermore, the study of neonatal seizures, in particular the coherence of state-defining variables or the evolution of sleep morphology, may benefit from attention to sleep architecture. Initial findings from some laboratories suggest that the very feature of excessive instability, which can be measured by repetitive long-term polysomnographic monitoring, signals a poor prognosis. In addition, fragmented sleep and the evolving interrelationship between ultradian and circadian rhythms may contain useful information that has yet to be mined. The advent of computer technologies can make the clinical laboratory into a setting where both research and clinical studies contribute to an elucidation of risk for sudden infant death syndrome and sequelae of neonatal seizures.