Predicting neonatal outcomes: birthweight, body mass index or ponderal index?

Aims: To compare birthweight (BW), body mass index (BMI) and ponderal index (PI) as predictors of selected short-term (ST) outcomes, namely Neonatal Intensive Care Unit (NICU) admission and prolonged hospitalization (PH), among newborn (NB) infants.
Methods: Data was collected prospectively on 9,226 infants born during one year at nine tertiary care hospitals in Greater Beirut, Lebanon. The predictive abilities of BW, BMI and PI were compared using the area under the receiver operator characteristic (ROC) curves and sensitivity analysis was performed at the optimal cut-off points for the best anthropometric measurement.
Results: The area under the ROC curve suggested superior discriminative power for BW as compared to BMI or PI, as a predictor of NICU admission (ROC area = 0.73) and PH (ROC area = 0.74). The optimal BW cut-off point was 2750 g (sensitivity: 0.49; specificity: 0.89) and 2950 g (sensitivity: 0.62; specificity: 0.78) for NICU admission and PH, respectively.
Conclusion: In our population of NB infants, BW--a crude measure of fetal growth--is a better predictor than either BMI--a measure of adiposity in adults and children--or PI--a measure of thinness at birth--for selected ST outcomes in NB infants.