Moderate and severe bronchopulmonary dysplasia (BPD) is a common pulmonary complication in premature infants, which seriously affects their survival rate and quality of life. This study aimed to describe the mechanical ventilation characteristics and evaluate their prediction performance for the risk of moderate and severe BPD in infants with gestational age <30 weeks and birth weight <1,500 g on postnatal Day 14.
In this retrospective cohort study, 412 infants with gestational age <30 weeks and birth weight <1,500 g were included in the analysis, including 104 infants with moderate and severe BPD and 308 infants without moderate and severe BPD (as controls). LASSO regression was used to optimize variable selection, and Logistic regression was applied to build a predictive model. Nomograms were developed visually using the selected variables. To validate the model, receiver operating characteristic (ROC) curve, calibration plot, and clinical impact curve were used.
From the original 28 variables studied, six predictors, namely birth weight, 5 min apgar score, neonatal respiratory distress syndrome (≥Class II), neonatal pneumonia, duration of invasive mechanical ventilation (IMV) and maximum of FiO2 (fraction of inspiration O2) were identified by LASSO regression analysis. The model constructed using these six predictors and a proven risk factor (gestational age) displayed good prediction performance for moderate and severe BPD, with an area under the ROC of 0.917 (sensitivity = 0.897, specificity = 0.797) in the training set and 0.931 (sensitivity = 0.885, specificity = 0.844) in the validation set, and was well calibrated (
The model included gestational age, birth weight, 5 min apgar score, neonatal respiratory distress syndrome (≥Class II), neonatal pneumonia, duration of IMV and maximum of FiO2 had good prediction performance for predicting moderate and severe BPD in infants with gestational age <30 weeks and birth weight <1,500 g on postnatal Day 14.