ABSTRACT
We sought to predict neonatal metabolic acidosis at birth using antepartum obstetric characteristics (model 1) and additional characteristics available during labor (model 2). In 5667 laboring women from a multicenter randomized trial that had a high-risk singleton pregnancy in cephalic presentation beyond 36 weeks of gestation, we predicted neonatal metabolic acidosis. Based on literature and clinical reasoning, we selected both antepartum characteristics and characteristics that became available during labor. After univariable analyses, the predictors of the multivariable models were identified by backward stepwise selection in a logistic regression analysis. Model performance was assessed by discrimination and calibration. To correct for potential overfitting, we (internally) validated the models with bootstrapping techniques. Of 5667 neonates born alive, 107 (1.9%) had metabolic acidosis. Antepartum predictors of metabolic acidosis were gestational age, nulliparity, previous cesarean delivery, and maternal diabetes. Additional intrapartum predictors were spontaneous onset of labor and meconium-stained amniotic fluid. Calibration and discrimination were acceptable for both models (c-statistic 0.64 and 0.66, respectively). In women with a high-risk singleton term pregnancy in cephalic presentation, we identified antepartum and intrapartum factors that predict neonatal metabolic acidosis at birth.