Assessing the diagnostic accuracy of lung ultrasound in determining invasive ventilation needs in neonates on non-invasive ventilation: An observational study from a tertiary NICU in India

Effective management of neonatal respiratory distress requires timely recognition of when to transition from non-invasive to invasive ventilation. Although the lung ultrasound score (LUS) is useful in evaluating disease severity and predicting the need for surfactants, its efficacy in identifying neonates requiring invasive ventilation has only been explored in a few studies. This study aims to assess the accuracy of LUS in determining the need for invasive ventilation in neonates on non-invasive ventilation (NIV) support. From July 2021 to June 2023, we conducted a prospective study on 192 consecutively admitted neonates with respiratory distress needing NIV within 24 h of birth at our NICU in Hyderabad, India. The primary objective was the diagnostic accuracy of LUS in determining the need for invasive ventilation within 72 h of initiating NIV. We calculated LUS using the scoring system of Brat et al. (JAMA Pediatr 169:e151797, [ 10 ]). Treating physicians’ assessments of the need for invasive ventilation served as the reference standard for evaluating LUS effectiveness. Out of 192 studied neonates, 31 (16.1%) required invasive ventilation. The median LUS was 5 (IQR: 2–8) for those on NIV and 10 (IQR: 7–12) for those needing invasive ventilation. The LUS had a strong discriminative ability for invasive ventilation with an AUC (area under the curve) of 0.825 (CI: 0.75–0.86, p = 0.0001). An LUS > 7 had 77.4% sensitivity (95% CI: 58.9–90.8%), 75.1% specificity (95% CI: 67.8–81.7%), 37.5% positive predictive value (PPV) (95% CI: 30.15–45.5%), 94.5% negative predictive value (NPV) (95% CI: 89.9–97.1%), 3.1 positive likelihood ratio (PLR) (95% CI: 2.2–4.3), 0.3 negative likelihood ratio (NLR) (95% CI: 0.15–0.58), and 75.5% overall accuracy (95% CI: 68.8–81.4%) for identifying invasive ventilation needs. In contrast, SAS, with a cutoff point greater than 5, has an AUC of 0.67. It demonstrates 62.5% sensitivity, 61.9% specificity, 24.7% PPV, 89.2% NPV, and an overall diagnostic accuracy of 61.9%. The DeLong test confirms the significance of this difference (AUC difference: 0.142, p = 0.04), underscoring LUS’s greater reliability for NIV failure. Conclusion: This study underscores the diagnostic accuracy of the LUS cutoff of > 7 in determining invasive ventilation needs during the initial 72 h of NIV. Importantly, while lower LUS values typically rule out the need for ventilation, higher values, though indicative, are not definitive. What is known? • The effectiveness of lung ultrasound in evaluating disease severity and the need for surfactants in neonates with respiratory distress is well established. However, traditional indicators for transitioning from non-invasive to invasive ventilation, like respiratory distress and oxygen levels, have limitations, underscoring the need for reliable, non-invasive assessment tools. What is new? • This study reveals that a LUS over 7 accurately discriminates between neonates requiring invasive ventilation and those who do not. Furthermore, the lung ultrasound score outperformed the Silverman Andersen score for NIV failure in our population.