Comparison of lung ultrasonography and chest radiography in detecting neonatal respiratory distress as a composite clinical condition: a prospective cross-sectional diagnostic accuracy study at Muhimbili National Hospital

Abstract

Neonatal respiratory distress (RD) is a leading cause of neonatal morbidity and mortality. While chest radiography (CXR) is the traditional modality of diagnosis, lung ultrasound (LUS) has emerged as a bedside, radiation-free alternative for evaluating neonatal respiratory conditions. To determine the diagnostic performance of lung ultrasonography (LUS) compared to chest radiography (CXR) in detecting neonatal respiratory distress (RD) as a composite clinical condition at Muhimbili National Hospital (MNH). A prospective cross-sectional diagnostic accuracy study was conducted at Muhimbili National Hospital (MNH) between June and December 2022. A total of 179 neonates with clinical signs of respiratory distress were consecutively enrolled. All neonates underwent both LUS and CXR within 6 h. Demographic and clinical data were collected using a structured tool. Diagnostic performance was assessed for overall respiratory distress rather than individual disease etiologies. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and diagnostic accuracy (DA) were calculated using a composite clinico-radiological reference standard. Agreement was assessed using Cohen’s kappa with 95% confidence intervals and significance set at p < 0.05. Among 179 neonates (mean gestational age: 32 ± 4 weeks; mean birth weight: 1,769 ± 856 g), LUS demonstrated high diagnostic performance in this setting, with higher sensitivity (95.1% vs. 90.3%), specificity (94.7% vs. 89.5%), positive predictive value (PPV) (96.1% vs. 92.1%), negative predictive value (NPV) (93.5% vs. 87.2%), and diagnostic accuracy (95% vs. 90%) compared to chest radiography (CXR). We observed almost perfect agreement between lung ultrasonography and the reference standard (κ = 0.898) and substantial agreement between lung ultrasonography and chest radiography (κ = 0.806). Lung ultrasonography demonstrated high diagnostic performance in this setting for detecting neonatal respiratory distress as a composite clinical condition. However, interpretation should consider the use of a composite reference standard that included chest radiography, which may have introduced incorporation bias. Additionally, the composite outcome limits interpretation for individual respiratory conditions. Despite these limitations, LUS appears to be a useful adjunct imaging modality in resource-limited settings, where access to radiography may be constrained.