Diagnostic utility of lung ultrasound in preterm neonates with respiratory distress at a tertiary neonatal intensive care unit in South Africa

Background. Accumulating data suggest the potential for lung ultrasound (LUS) to diagnose pathology in preterm babies, but there are no published data from South Africa (SA). Objectives. To describe LUS diagnoses, respiratory pathology and outcomes in preterm babies receiving non-invasive respiratory support in a tertiary SA neonatal unit, and to compare LUS with clinical diagnoses and surfactant administration. Methods. We conducted a prospective, observational study of babies at 27 - 34 weeks’ gestation, with birthweight ≥800 g, receiving non-invasive respiratory support and who had LUS at age ≤3 hours. Surfactant was administered at a fraction of inspired oxygen (FiO2) of 0.35 - 0.45 and was not influenced by LUS findings. Results. Fifty-one neonates were included, of whom 16% received surfactant, all with respiratory distress syndrome (RDS) as their only clinical diagnosis, compared with multiple diagnoses in the non-surfactant group, including RDS (93%), transient tachypnoea of the newborn (TTN) (16%) and pneumonia (14%). Lung ultrasound indicated less RDS in the non-surfactant group than the surfactant group (42% v. 88%; p=0.02), and more TTN (61% v. 13%; p=0.01). The LUS score (LUSS) predicted surfactant administration (area under the curve 0.8 (95% confidence interval 0.67 - 0.94)). A LUSS of 7 had the best combined sensitivity (75%) and specificity (72%) but a low positive-predictive value (33%). A LUSS of 8 identified 8 (19%; n=8/43) additional babies for surfactant who did not need treatment. Conclusion. LUS suggested more diagnoses than clinical assessment, particularly TTN, but did not accurately predict surfactant administration at an FiO2 of 0.35 - 0.45.