Abstract
Background  Lung ultrasound (LUS) is widely used to diagnose neonatal respiratory diseases. However, to our knowledge, few straightforward method was reported to predict respiratory support need precisely. Our aim is to determine the diagnostic accuracy of a semiquantitative LUS assessent method predicting the need for respiratory support.
Methods  We conducted a prospective diagnostic accuracy study following STARD (Standards for the Reporting of Diagnostic Accuracy Studies) guidelines at a tertiary level academic hospital between 2019 to 2020. 310 late preterm and term infants enrolled. They were delivered in the obstetric department and transferred to a monitoring room to determine whether they need NICU treatment. The LUS assessment was performed for each participant at one of following timings–0.5h, 1h, 2h, 4h, 6h after birth. Reliability was tested by ROC analysis. Surfactant administration and other respiratory support were based on 2019 European guidelines as well as their clinical condition.
Results  74 were confirmed to need respiratory support and 236 were healthy according to a 3-day follow up. Six LUS image patterns can be seen in these infants right after birth. Two ”high-risk” patterns well relate to respiratory support need(area under the curve(AUC) = 0.95; 95% CI, 0.92-0.98, p<0.001). This reliability can be supported by AUC of ”low-risk” patterns(AUC = 0.89, 95%CI, 0.85-0.93, p<0.001). Predictive value of LUS is much greater than that of using respiratory symptoms(e.g.respiratory rate)(AUC of LUS vs AUC of respiratory rate, p<0.01).
Conclusions  LUS can predict respiratory support need and is more reliable than the assessment based on respiratory symptoms.
Key words Lung ultrasound; respiratory support; predictive value; ROC