ABSTRACT
OBJECTIVE: To determine the causal relationship between exposure to early hyperoxemia and death or major disability in infants with hypoxic-ischemic encephalopathy (HIE). STUDY DESIGN: We analyzed data from the Infant Cooling Evaluation (ICE) trial that enrolled newborns ≥35 weeks' gestation with moderate-severe HIE, randomly allocated to hypothermia or normothermia. The primary outcome was death or major sensorineural disability at 2 years. We included infants with arterial pO2 measured within 2 hours of birth. Using a directed acyclic graph, we established that markers of severity of perinatal hypoxia-ischemia and pCO2 were a minimally sufficient set of variables for adjustment in a regression model to estimate the causal relationship between arterial pO2 and death/disability. RESULTS: Among 221 infants, 116 (56%) had arterial pO2 and primary outcome data. The unadjusted analysis revealed a U-shaped relationship between arterial pO2 and death or major disability. Among hyperoxemic infants (pO2 100-500 mmHg) the proportion with death or major disability was 40/58 (0.69), while the proportion in normoxemic infants (pO2 40-99 mmHg) was 20/48 (0.42). In the adjusted model, hyperoxemia increased the risk of death or major disability (adjusted risk ratio 1.61, 95% CI 1.07-2.00, P = .03) in relation to normoxemia. CONCLUSION: Early hyperoxemia increased the risk of death or major disability among infants who had an early arterial pO2 in the ICE trial. Limitations include the possibility of residual confounding and other causal biases. Further work is warranted to confirm this relationship in the era of routine therapeutic hypothermia.