Abstract
Objective The objective of this study was to use data from the 20-center beneficial effect
of antenatal magnesium sulfate (BEAM) trial to assess the external validity of the
Neonatal Research Network (NRN) estimator, a widely employed web-based counseling
tool to estimate the probability of an adverse outcome for periviable infants given
intensive care.
Study Design The probability of different adverse outcomes predicted from the NRN estimator was
compared with observed rates at 18 to 22 months for ventilated, nonanomalous infants
born at 23 to 25 weeks and assessed in BEAM as in the NRN. Results were assessed using
rigorous validation methods for prediction models.
Results Among 289 eligible infants, 26% died, 40% died or had profound neurodevelopmental
impairment (PNDI), and 71% died or had NDI. The area under the receiver operating
characteristic curve was 0.70 (95% confidence interval [CI], 0.63–0.78) for death,
0.64 (95% CI, 0.56–0.71) for death or NDI, and 0.71 (95% CI, 0.65–0.78) for death
or PNDI. Observed and predicted rates were somewhat different for death or NDI but
quite similar for death and for death or PNDI in different risk groups. Brier scores
for accuracy were favorable (0.17–0.22) for all outcomes.
Conclusion Our results provide external validation of the NRN estimator for assessing the probability
of adverse outcomes at 18 to 22 months for periviable infants given intensive care.
Keywords
periviable birth - neonatal death - neonatal morbidity - extreme prematurity
