Brain dysmaturity as expressed by EEG-sleep differences between healthy preterm and fullterm neonatal cohorts
Two statistical strategies to analyze EEG-sleep are described for a population of 129 healthy newborns. Seven EEG-sleep measures were selected to quantitate the degree of EEG dysmaturity between preterm and fullterm neonatal cohorts. These analytical tools were chosen since each procedure preserved each individual EEG-sleep measure's unique contribution to the overall neurophysiological expression of sleep. First a computerized graphics program depicted which combination of measures resulted in the maximum separation of the two neonatal cohorts in multidimensional 'physiological space.' A dysmaturity index was then defined as the Mahalanobis distance (MD) for each EEG-sleep measure in the preterm group when compared for each value in the fullterm group. Such an index is geometric in spirit and avoids the cancellation that can occur with summated measures. Values of specific EEG- sleep behaviors for normal fullterm infants tend to cluster fairly tightly around a location that depends on postconceptional age, while preterm infants are more disbursed and center on a different location. Receiver-operator characteristic curves (ROC curves) were constructed to display the results of multivariate analyses: greater differences between cohorts were displayed when more EEG-sleep measures were compared. Therefore, this index is the difference between the 'at-risk' child's vector and the center of the cluster for normal infants of the same conceptional age (CA). This index will be used to compare neonatal groups with perinatal factors at increasing CA for their risk for developmental delay and Sudden Infant Death Syndrome (SIDS). Such a neurophysiological sleep profile may be useful to compare a given individual with altered brain maturation who is at-risk to a control group studied at comparable ages.
