Lung ultrasound for early surfactant treatment: Development and validation of a predictive model

Aim To develop and validate a feasible predictive model for early surfactant treatment in very preterm infants (VPI) admitted with respiratory distress syndrome (RDS). Methods Preterm infants less than 32 weeks of gestation with RDS and stabilized with noninvasive ventilation in delivery room were recruited (January 2018-April 2020). Clinical data, chest X-ray (CXR) score, respiratory support, oxygen saturation/fraction of inspired oxygen ratio (SF ratio), lung ultrasound (LUS) score, and diaphragmatic thickening fraction (DTF) were recorded at 60-120 min of life. Oxygen threshold for surfactant administration was fraction of inspired oxygen more than 30%; ultrasound findings were blinded. Logistic regression models using a stepwise selection of variables were developed in the derivation cohort. Coefficients from these models were applied to the validation cohort and a diagnostic performance was calculated. Results A total of 144 VPI with a mean gestational age of 28.7 ± 2.2 weeks were included (94 into the derivation cohort, 50 into the validation cohort); 37 required surfactant treatment (25.7%). Gestational age, SF ratio, LUS score, CXR score, and Silverman score were related to surfactant administration (R2 = .823). Predictors included in the final model for surfactant administration were SF ratio and LUS score (R2 = .783) with an area under the receiver operating characteristic (AUC) = 0.97 (95% confidence interval [CI]: 0.93-1.00) in the derivation cohort and an AUC = 0.95 (95% CI: 0.85-0.99) in the validation cohort. By applying our predictive model, 26 patients (70.2%) would have been treated with surfactant earlier than 2 h of life. Conclusion The predictive model showed a high diagnostic performance and could be of value to optimize early respiratory management in VPI with RDS.