35: Correlation between initial neonatal and early childhood outcomes among children delivered <34 weeks gestation


      Obstetric researchers commonly use neonatal morbidities as surrogate endpoints for longer-term outcomes. We sought to correlate neonatal diagnoses prior to hospital discharge with early childhood cognitive and motor function.

      Study Design

      Secondary analysis of a multicenter RCT of antenatal magnesium sulfate (Mg) vs. placebo administered to women at imminent risk for early PTB to prevent death and cerebral palsy (CP) in their offspring. All women were at high risk for PTB <32.0 wks. Singletons delivered 24.0-33.9 wks who survived to hospital discharge post-birth and had 2-year-old outcome data were included. Those surviving to age 2 were assessed by trained physicians and Bayley Scales of Infant Development Mental Development and Psychomotor Development Indices (MDI, PDI). Neonatal diagnoses at each baby’s initial hospital discharge were examined singly and in combination to determine those most predictive of severe composite childhood morbidity, defined as a childhood diagnosis of moderate/severe CP and/or Bayley MDI and/or PDI scores >2 SD below the mean. Data were analyzed by multiple logistic regression and area under ROC curves (AUC).


      1400 children met criteria. Children were delivered at a mean of 29.9 (range: 24.0-33.9) wks gestation. 58 (4.1%) had moderate/severe CP. On Bayley testing, 245 (19.2%) had a MDI Score >2 SD and 229 (17.8%) a PDI Score >2SD below the mean. A total of 349 (24.9%) had severe composite childhood morbidity. Multivariable regression results demonstrating the relationship between neonatal diagnoses and severe childhood morbidity are shown in the Table.


      Approximately 1 in 4 children born <34 weeks had severe childhood morbidity at age 2. Individual neonatal morbidities (BPD, NEC, sepsis, severe IVH, and PVL) had modest predictive value for subsequent adverse early childhood outcomes; combinations of multiple morbidities were only marginally more prognostic. Prediction of childhood outcomes from neonatal diagnoses remains imperfect.
      Tabled 1Multivariable regression results
      Figure thumbnail fx1
      Relationship between neonatal diagnoses and the probability of severe composite childhood morbidity are shown. All models are shown for individual predictors; the best model is shown for the combination of 2, 3, and 4 neonatal morbidities.
      *brain injury = severe intraventricular hemorrhage and/or periventricular leukomalacia
      Other co-variables in regression models included delivery gestational age, maternal education, randomization to magnesium, and chorioamnionitis.