Predicting neurodevelopmental outcomes of preterm infants: A holy grail

Abstract

A major concern of physicians, parents and society is the high frequency of a wide range of adverse neurodevelopmental outcomes for those infants that survive very preterm birth. More and more information indicates that assessments at discharge poorly predict assessments at 2 years, and that these early assessments do not predict school age outcomes very well. Ambalavanan et al used classification tree analysis to evaluate if a structured analysis of clinical variables improved the prediction of neurodevelopmental outcomes. The surprising result was that simply using antenatal data was as good as antenatal plus 5 days of post-delivery information or antenatal plus 8 days of post-delivery information. The implication is that postnatal care is quite irrelevant to two year neurodevelopmental outcomes. In a provocative editorial, MacKendrick questions the assumptions of “linear models” for a process as complex as neurodevelopment. He suggests chaos theory may help us understand the varied outcomes that result from preterm delivery.