Stability in waking-sleep states in neonates as a predictor of long-term neurologic outcome.

Thirty-three full-term neonates were ranked blindly on a scale ranging from the least to the highest "risk" for future neurologic complications on the basis of available perinatal biographies, tests, and examinations performed during the newborn period. Four prolonged polygraphic-behavioral recordings were obtained one week apart beginning at ten days after birth. Five waking and sleep states were scored in each session as percentages of total observation time, giving a total of 20 scores for each baby to be subjected to analysis of variance. These measures also provided individual profile consistency or variability in maintaining waking-sleep states over the selected period of postnatal time. The whole cohort, except three infants who could not be followed adequately, was then reexamined periodically over a period ranging from 3 to 4 years (+/- 6 months) for neurologic and developmental assessments. Except for two scores that produced a low level of statistical significance (P less than .05), the other 18 scores were found to be not associated with long-term outcomes. Even the first two scores were not satisfactory discriminators for the outcome of the individual babies. However, when coefficients of concordance (W) were computed from each individual baby profile, significant statistics (P less than .001) emerged to indicate good correlations between high or low W values in the newborn period and long-term outcomes. All 17 newborns who had W scores greater than 0.9 were found to be normal at follow-up regardless of the poor ranking given several of them during the newborn period. Among the 13 newborns who had W scores less than 0.9, 11 had a poor clinical outcome at follow-up, though several had been ranked initially as falling within the least "risk" group.