Early lung ultrasound affords little to the prediction of bronchopulmonary dysplasia

Objective: To test the hypothesis that lung ultrasound (LUS) performed in the first week of life would predict bronchopulmonary dysplasia (BPD). Secondary outcomes included the utility of LUS in predicting interim respiratory interventions. Design: A prospective observational cohort study in preterm infants born <28 weeks' gestation in the single tertiary statewide neonatal intensive care unit in Western Australia. Methods: A rigorous protocol for LUS acquisition on day 1, day 3-4, day 7, day 28 and 36 weeks' postmenstrual age (PMA) was implemented with blinded analysis using a modified, previously validated LUS score. BPD was defined by both recent National Institute of Child Health and Human Development categorical criteria and a continuous physiological variable using a modified Shift test. Results: Of the 100 infants studies, primary outcome data were available for the 96 infants, surviving to 36 weeks' PMA. In a univariate logistic regression analysis, LUS on days 3-4 and day 7 accurately predicted BPD (day 3-4 OR (95% CI)=1.54 (1.03 to 2.42), p=0.044; day 7 OR (95% CI)=1.66 (1.07 to 2.70), p=0.031). The predictive value of LUS was insignificant in a multivariate model in which gestational age was the dominant predictor. LUS accurately predicted interim respiratory outcomes including surfactant administration, duration of intubation and extubation to non-invasive support at 48 hours. Conclusions: LUS in the first week of life predicted BPD. However, LUS offers little additive accuracy to current gestational age-based models. Trial registration number: ACTRN12617000208303.