Chawla, S., Natarajan, G., Gantz, M. G., Shankaran, S., & Carlo, W. A. (2018). Reply: Markers of successful extubation in extremely preterm infants, and morbidity after failed extubation. Journal of Pediatrics, 194, 263-264. DOI: 10.1016/j.jpeds.2017.11.017
To the Editor:
We appreciate the opportunity to respond to the comments by Shalish and Sant'Anna regarding our publication in which the factors associated with failed elective extubation among extremely premature infants were described. The study was a secondary analysis of the Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network's “Surfactant, Positive Pressure, and Oxygenation Randomized Trial” that included extremely preterm infants born at 240/7 to 276/7 weeks of gestation.1
We agree that an accurate and reliable prediction tool for extubation readiness among premature infants is needed. We acknowledge that previous small, mostly single-center studies have evaluated bedside tests and complex technology, such as a “spontaneous breathing trial” (SBT), minute ventilation, heart rate variability, and respiratory variability for the prediction of extubation readiness.2-6 Although the SBT was noted to have a good positive predictive value (88%), the negative predictive value for successful extubation was low (63%).4 The optimal duration of the SBT and criteria to define failure of SBT remain uncertain. Complex machine learning algorithms for autonomic function remain in the research setting and require laborious analysis. Although we agree that many of the variables assessed in our recent study are routinely used by clinicians, data in our study were derived from a large, randomized trial with clear criteria for extubation and re-intubation. The study was able to objectively identify the variables that are independently associated with extubation success and, therefore, are of immediate clinical usefulness.
There is no consensus on the definition of extubation success for premature infants. There has been wide variation in the definitions used to define extubation failure, ranging from 72 hours to 7 days.2-5 However, these criteria are empiric and are not based on strong evidence. A meta-analysis showed that, in studies of infants with a median birth weight of 1000 g or less, reintubation rates steadily increased as the window of observation increased, without apparent plateau (P = .001).6 Extending the duration to define extubation failure beyond 5 days would certainly include more infants in the failed group. However, this would also increase the risk of including infants who fail extubation owing to new and unforeseen complications, such as necrotizing enterocolitis or infection that may not be related to the respiratory disease. In the current study, we noted that the majority of infants (75%) in the failed extubation group were reintubated within 2 days, with a clear decrease in the rate of reintubation between 2 and 5 days. So, we believe that a duration of 5 days to define extubation success is consistent with the literature and reasonable.
We agree with Shalish and Sant'Anna regarding the independent effect of duration of mechanical ventilation on neonatal morbidities. We included both gestational age and postmenstrual age at the time of the initial elective extubation in the logistic regression analysis. The duration of mechanical ventilation before an extubation attempt was reflected in the postmenstrual age, because the majority of infants were intubated shortly after birth. The duration of mechanical ventilation after the extubation attempt is a confounding variable that is associated with the outcome (failed and successful extubation) as well as certain morbidities such as bronchopulmonary dysplasia, duration of hospital stay, and duration of oxygen support. Extubation failure has been associated independently with increased mortality, longer duration of hospitalization, and more days on oxygen and ventilatory support.4,7
We agree with Shalish and Sant'Anna that a more complex algorithm using sophisticated, machine learning tools will be the future for prediction of extubation readiness. This tool would allow the clinician to extubate infants as early as possible while minimizing the risk of extubation failure to reduce the duration of mechanical ventilation and improve outcomes of premature infants.