Utility of Lung Ultrasound in the Estimation of Extravascular Lung Water in a Pediatric Population—A Prospective Observational Study

Objective Lung ultrasound (LUS) is a promising bedside modality for the estimation of extravascular lung water index (EVLWI), but has not been validated against objective measures in children. This study aimed to investigate the correlation of LUS B-line scoring with EVLWI, thresholds indicating elevated EVLWI, and its outcome following pediatric cardiac surgery. Design Prospective observational study. Setting Cardiothoracic surgical intensive care unit in a tertiary care teaching hospital. Participants Children younger than 12 years undergoing elective complete surgical correction of cyanotic or acyanotic congenital heart disease (Aristotle score ≤9), excluding neonates, those weighing <3.5 kg, and those with thoracic deformities, pulmonary pathology, and hemodynamic instability. Interventions Extravascular lung water index measurement by transpulmonary thermodilution, along with concurrent LUS B-line and Chest-X ray (CXR) scoring. Measurements and Main Results LUS B-line score had a moderate correlation with EVLWI (Pearson’s correlation coefficient 0.57; 95% CI 0.44-0.69). LUS B-line scores showed acceptable discrimination only for higher thresholds of EVLWI (sensitivity 82% and 79%, respectively, for EVLWI >20 mL/kg v sensitivity and specificity 57% and 80% for EVLWI >10 mL/kg). Age, body surface area, vasoactive-inotropic score (VIS), chest X-ray score, and EVLWI but not LUS B-line score were significant predictors for duration of mechanical ventilation in this cohort. Conclusions LUS B-line scoring has limited utility in semiquantitative estimation of EVLWI at lower thresholds of EVLWI in pediatric cardiac surgical patients. It may have better discrimination and acceptable sensitivity and specificity at higher thresholds of EVLWI. Contrasting with multiple reports of clinical utility, these results call for wider evaluation of LUS and its clinical modifiers like age, pathology, and pretest probability in estimation of EVLWI.