Objective: To build models that predict severe adverse outcome within 4 hours of birth in patients with postasphyxial hypoxic-ischemic encephalopathy. The goal was to develop models for selecting patients for therapeutic trials of neuroprotective medications.
Study design: Retrospective cohort study with follow-up to a minimum age of 12 months of 164 "outborn" term infants admitted to a tertiary neonatal intensive care unit, and 14 "inborn" term infants in the two tertiary perinatal centers in a regionalized setting. After performing univariate screening tests, multivariate models of association between risk factors and "severe adverse outcome" (death or major neurosensory impairment) were constructed.
Results: Of 178 infants with postasphyxial hypoxic-ischemic encephalopathy of defined severity admitted consecutively between 1985 and 1992, 48 died, 40 survived with major neurosensory impairment, and 13 were lost to follow-up. The important predictors of severe adverse outcome in the first 4 hours were delayed onset of breathing, administration of chest compressions, and seizures. At 60 minutes of age, based on predicted probabilities of > 0.50, the sensitivity of the predictive model was 85% and specificity 68%. The parameter estimates of the predictive models are reported.
Conclusions: Age of onset of breathing, administration of chest compressions, and age of onset of seizures were the most important variables predictive of adverse outcome in this study. Although fairly sensitive and specific, these predictive models should be applied with caution. To build more accurate models, a template for the conduct of a large, multicenter prospective study is provided.