Modeling development of sleep-wake behaviors: I. Using the mixed general linear model

The purpose of this paper is to demonstrate the use of the mixed general linear model (MixMod) for modeling development of sleep-wake behaviors in preterm infants. The mixed general linear model allows the concurrent identification of both group and individual developmental patterns in longitudinal data sets with inconsistently timed data, irregularly timed data, and randomly missing values. This statistical technique is well suited to data from preterm infants because these infants enter and leave longitudinal studies at varying times depending on their health status. One sleep organizational variable--the regularity of respiration in quiet sleep--obtained from a study of 37 preterm infants was used as an example. Seven infant characteristics were used as covariates. The various steps involved in conducting a mixed model analysis of this variable are illustrated. The strengths and limitations of this technique are discussed.