Diaphragm ultrasound as a predictor of extubation success in neonates and infants: A bicentric predictive accuracy...

Invasive mechanical ventilation is a necessary support for patients with critical illnesses who present with severe respiratory failure. However, it carries inherent risks, including potential injury to the lungs and diaphragm atrophy and dysfunction. Recent studies have identified diaphragm ultrasound (DUS) as a promising method for assessing extubation readiness. However, the available data on its effectiveness in predicting successful extubation in infants remain limited.

The objective of this study was to determine the accuracy of DUS in predicting successful extubation in infants.

A bicentric predictive accuracy study was conducted in infants with invasive mechanical ventilation ≥24 h. The following DUS variables were evaluated 5 min before extubation: diaphragm thickness, diaphragmatic excursion (DE), diaphragm thickness fraction (DTF), diaphragmatic contraction velocity (DCV), diaphragmatic relaxation velocity (DRV), and inspiratory time (IT) and expiratory time (ET). A comparison was made between success and failure groups, and the accuracy of DUS variables in predicting success was calculated using principal component analysis (PCA).

The study included 38 neonates and infants, with 50% of extubation failures due to upper airway obstruction. DRV showed a statistically significant difference between the success and failure groups, with a higher value for the extubation failure group (9,68 ± 3,62 vs 17,27 ± 12,28 p < 0.037). Significant correlations were found between DE, DCV, and DRV, showing an inverse association with extubation success in the principal component analysis retained a single and restricted scenario. The area under the curve (AUC) for component 1 was 0.7536, 80% specificity, and 64.29% sensitivity. The restricted bank showed a PC1rest and extubation success (95% confidence interval: 0.223-0.966; p = 0.0388), with AUC = 0.743, sensitivity = 0.750, and specificity = 0.800. In the complete bank, PC1comp was also significant (95% confidence interval: 0.231-0.925; p = 0.0255), with AUC = 0.754, sensitivity = 0.857, and specificity = 0.600.

Diaphragm velocities, especially DRV, may be an outcome measure to use in predicting weaning in infants. Component 1 (DE, DCV, and DRV) in the restricted and complete bank demonstrated good performance in distinguishing success extubation outcome in neonates and infants and showed potential as predictive metrics, suggesting its value in enhancing predictive models.