Accurate prediction of hypoxic-ischaemic encephalopathy at delivery: a cohort study

Abstract

Objective: Hypoxic-ischaemic encephalopathy (HIE) is a major acute neurologic manifestation of perinatal asphyxia associated with significant mortality and morbidity. The study aimed to develop a simple, accurate method of predicting HIE at delivery. Methods: Between January 2003 and December 2009, all HIE cases were identified from the 38,404 deliveries at a single tertiary centre. Receiver operating curve (ROC) analysis and multivariate logistic regression assessed the ability of clinical and biochemical assessments to predict HIE. Results: Sixty neonates met the HIE criteria: 39 were moderate-severe HIE. Univariate analyses identified clinical neonatal markers (Apgar scores and neonatal resuscitation level) to be better HIE predictors than biochemical markers (umbilical artery pH, base excess and lactate values). Multivariable models using two to four predictors had areas under ROC curves up to 0.98, sensitivities up to 93% and specificities up to 99%. For moderate-severe HIE, the most effective predictor was neonatal resuscitation level and arterial lactate (ROC 0.98, sensitivity 85%, specificity 99%). Conclusion: The combination of umbilical arterial lactate and neonatal resuscitation level provides a rapid and accurate method of predicting moderate-severe HIE that can identify neonates at birth that may benefit from tertiary care and neuroprotective therapies.