To identify variables that predict extubation success in extremely preterm infants born <28 weeks gestational age (GA), and to compare outcomes between those who had successful or failed extubation.

A secondary analysis of data from a randomized trial of postextubation respiratory support that included 174 extremely preterm infants. “Extubation success” was defined as not requiring reintubation within 7 days, and “extubation failure” the converse. Predictive variables that were different between groups were included in a multivariable logistic regression model.

Sixty-eight percent of infants were successfully extubated. Compared with those infants who had extubation failure, they had a higher GA and birth weight, were extubated earlier, were more often exposed to prolonged ruptured membranes, more often avoided intubation in the delivery room, had a higher pre-extubation pH, and had lower mean pre-extubation fraction of inspired oxygen and partial pressure of carbon dioxide (PCO₂). Only GA and PCO₂ remained significant in the multivariable analysis (area under a receiver operating characteristic curve = 0.81). Extubation failure was associated with death, bronchopulmonary dysplasia, severe retinopathy of prematurity, patent ductus arteriosus ligation, and longer durations of respiratory support, oxygen supplementation, and hospitalization. When adjusted for allocated treatment in the randomized trial, GA, and birth weight z-score, extubation failure remained associated with death before discharge and prolonged respiratory support and hospitalization.

In extremely preterm infants, higher GA and lower pre-extubation PCO₂ predicted extubation success. Infants in whom extubation failed were more likely to die and have prolonged respiratory support and hospitalization.

Australian New Zealand Clinical Trials Network: ACTRN12610000166077.