Clinical prediction of post-extubation radiological changes of the chest in newborn infants

Objective: Whether a chest radiograph should be performed routinely in all infants after extubation, or selectively only in those with clinical deterioration, is a controversy in neonatal unit practice. This study tested the hypothesis that most cases of post-extubation radiological deterioration in the lungs could be detected by clinical assessment. Methods: A chest radiograph was performed at 8 h post-extubation in 100 episodes of extubation in 85 newborn infants ventilated for a variety of lung diseases. Each infant was assessed at the same time by a neonatologist blinded to the radiological findings, to determine whether a chest radiograph would have been requested based on clinical judgement. The infants were continuously monitored for their respiratory and oxygenation status before and after extubation. Results: Compared to the pre-extubation chest radiographs, 23 of the 100 post-extubation chest radiographs showed either deterioration of the pre-existing lung pathologies or appearance of significant new pathologies. The clinicians' assessment failed to detect most of the deterioration, with a sensitivity of only 21.7%. Systematic analysis of the infants' clinical parameters showed that the development of significant intercostal/subcostal retraction, and an increase in inspired oxygen concentration by ≥ 7% after extubation, were the best predictors of post-extubation radiological deterioration (sensitivity 82.6%, specificity 62.3%, positive predictive value 39.6%, and negative predictive value 92.3%). Serial blood gas in contrast had little predictive value. Conclusion: We conclude that most cases of radiological deterioration of the lungs after extubation are clinically predictable, provided the correct clinical criteria are used.