213: Intrapartum risk stratification for early-onset neonatal sepsis


      Intrapartum chorioamnionitis, a diagnosis usually prompted by maternal fever and often based on soft clinical findings, leads to neonatal sepsis workup, antibiotic administration and longer hospital stay. Given that most neonates born to women with this diagnosis do not actually have an infection, a more objective diagnostic approach is needed to prevent overtreatment. Our aim was to develop a multivariable approach to improve risk-stratification of early onset neonatal sepsis (EOS).

      Study Design

      This is a secondary analysis of a multicenter observational cohort. Trained and certified research personnel abstracted the maternal and neonatal records of women delivering at 25 hospitals over a 3-year period, and perinatal outcomes were ascertained according to pre-specified guidelines. Liveborn, non-anomalous singleton neonates, or firstborn in case of twins, were included. Maximum maternal temperature, maternal antibiotic use and culture-proven EOS were recorded. Multivariable analysis using k-fold cross validation identified antepartum and intrapartum factors that were independently associated with EOS. The best final multivariable model was then compared with a model that included the same variables plus clinically-diagnosed chorioamnionitis and a model with clinically-diagnosed chorioamnionitis only.


      Among 111,593 women who delivered during the study period, 175 neonates experienced EOS (0.16%, 95%CI 0.13-0.18). EOS occurred in 0.15% of the cases without fever and in 1.57% of those with temperature >102.2 oF (39 oC). EOS was rare in term neonates. The variables listed in the Table produced the best-fit model for prediction of EOS (area under the curve [AUC] = 0.92). This best-fit model was superior to clinical chorioamnionitis alone (AUC = 0.55; p<0.001; Figure), and adding clinical chorioamnionitis to the best model did not improve it (AUC = 0.92).


      In this large cohort with rigorously collected data, proven EOS was rare, particularly in term neonates, and clinical diagnosis of chorioamnionitis was not a good predictor. A better predictor of EOS is a combination of maximum maternal temperature, BMI, gestational age and birthweight. Using this risk-stratification model should improve identification of neonates at risk for EOS and prevent unnecessary neonatal testing and intervention, as well as decrease cost.
      Figure thumbnail fx1
      Figure thumbnail fx2