Evaluating the Predictive Value of Umbilical Cord and Infant Blood Gases in Hypoxic-Ischemic Encephalopathy.

Abstract

To evaluate the relationship between umbilical cord blood gases, postnatal blood gases, severity of clinical encephalopathy, and degree of injury on brain magnetic resonance imaging (MRI) in infants with hypoxic-ischemic encephalopathy. This was a retrospective cohort study of 168 infants treated for hypoxic-ischemic encephalopathy with therapeutic hypothermia. Information on demographics, clinical course, umbilical cord blood gases, and postnatal blood gases (capillary, venous, or arterial) were collected. Brain MRIs were blindly graded for severity using a systematic scoring system. The odds of moderate-to-severe brain injury on MRI increased by 45% with each 0.1 mmol/L decrease in postnatal pH (OR = 1.49, 95% CI: 1.02-2.16, P = .038), by 15% with each 1.0 mmol/L decrease in postnatal base deficit (BD) (OR = 1.19, 95% CI: 1.07-1.32, P = .002), and by 22% with each 1.0 mmol/L decrease in postnatal bicarbonate (OR = 1.22, 95% CI: 1.07-1.39, P = .003). Similar associations were observed between postnatal blood gas parameters and moderate-to-severe clinical encephalopathy on examination. Umbilical cord arterial and venous blood gas parameters were not associated with odds of either outcome, except for a nonlinear association with arterial pH in which further decreases among infants at already lower than average pH levels were associated with increased odds. In receiver operating characteristic analyses using postnatal gases, thresholds for predicting moderate-to-severe brain injury on MRI were BD ≤ -11.6 mmol/L and bicarbonate ≤ 13.2 mEq/L, with areas under the curve of 0.78 and 0.76, respectively. For moderate-to-severe clinical encephalopathy, the optimal threshold was BD ≤ -9.9 mmol/L (area under the curve of 0.71). Postnatal blood gas measurements obtained early after delivery are associated with increased odds of moderate-to-severe brain injury and clinical encephalopathy. Umbilical cord gas measurements did not demonstrate predictive power for neonatal outcomes.