A comparison of clinical variables that predict adverse outcome in term infants with severe respiratory failure randomised to a policy of extracorporeal membrane oxygenation or to conventional neonatal intensive care

Objective: To identify clinical variables predicting adverse outcome in a group of infants with severe respiratory failure who were randomized either to referral for extra-corporeal membrane oxygenation (ECMO) or to conventional neonatal intensive care within the United Kingdom. Methods: Adverse outcome was defined by death or disability by four years of age. Receiver operator characteristic (ROC) plots were constructed for variables with continuous data and relative risk (RR) with 95 % confidence intervals (CI) calculated for binominal data. Results: Of variables measurable at trial entry, congenital diaphragmatic hernia and lower birthweight was also associated with increased mortality and morbidity. Seizures or supplementary oxygen at discharge were markers of disease course, which predicted a poorer outcome amongst survivors. These variables behaved similarly in the two trial groups. Those infants in the ECMO group with an episode of sepsis, established full sucking feeds after 14 days of age or a hospital stay over 30 days were at increased risk of disability. Conclusions: This study has identified clinical variables that predict adverse outcome for term infants with severe respiratory failure. The results may assist clinicians caring for these babies, when counseling their families and in the development of guidelines for neonatal ECMO.