Optimal gestational age and birth-weight cutoffs to predict neonatal morbidity

Background. Gestational age (GA) and birth weight (BW) criteria are used to identify newborns at risk for neonatal morbidity. Currently, preterm is GA less than 37 weeks; low birth weight is BW less than 2,500 grams; and smal l for gestational age (SGA) is BW less than the tenth percentile weight for the infant's GA. The optimal classification system balances the misclassif ication cost of false negatives against the cost of false positives. Object ive. To calculate the relative misclassification costs implied by the curre nt 37-week and 2,500-gram cutoffs, and to test the validity of the current definition of SGA as a predictor of term morbidities. Methods. GA, BW, and morbidity information were collected for 22,606 infants born between July 1 981 and December 1992. Using this dataset, logistic regression coefficients were obtained modeling GA or BW as predictors of morbidities associated wi th prematurity. For a subset of 18,813 infants with GAs between 37 and 41 w eeks, coefficients were obtained modeling both GA and BW as independent pre dictors of term morbidities. The logistic regression coefficients were used to calculate optimal birth weight, gestational age, and birth-weight-for-g estational-age cutoffs. Results. The current definitions of low birth weigh t and preterm imply that it is 18 to 28 times more costly to misclassify a sick infant as tow-risk than to misclassify a well infant as high-risk. Con clusions. Gestational age atone is better than birth weight alone at predic ting preterm morbidities. No birth-weight cutoff can adequately; predict te rm morbidities. A single weight-percentile cutoff for all gestational ages should not be used to identify newborns at high risk for neonatal morbidity .