Constructing prediction models for the moderate/severe bronchopulmonary dysplasia in extremely preterm infants at different days after birth

Abstract

Objective To construct prediction models for moderate/severe bronchopulmonary dysplasia （BPD） in extremely preterm infants at different days after birth. Methods The preterm infants with BPD at a gestational age of < 28 weeks at the Third Affiliated Hospital of Guangzhou Medical University from September 2017 to December 2019 were retrospectively selected. They were divided into the mild-BPD group and moderate/severe-BPD group. The significant variables screened by using single-factor analysis were subject to risk assessment analysis by multivariate Logistic regression analysis. Early prediction models were established， and the predictive performance was validated and evaluated. Results Ninety extremely preterm infants were included in this study， including 58 cases （64.4%） in the mild-BPD group and 32 cases （35.6%） in the moderate/severe-BPD group. Multiple Logistic regression analysis showed birth weight， maternal pre-pregnancy body mass index （BMI） ≥ 25 kg/m2， 1-minute Apgar score≤7 after birth and pneumonia were the influencing factors for moderate/severe BPD at 7 days after birth. Preeclampsia， maternal pre-pregnancy BMI≥25 kg/m2， pneumonia and hsPDA were the risk factors for moderate/severe BPD at 14 days postnatal age. Preeclampsia， maternal pre-pregnancy BMI≥25 kg/m2， days of invasive mechanical ventilation and hsPDA were the risk factors for moderate/severe BPD at 28 days after birth. The area under the receiver operating characteristic （ROC） curve of the prediction models for 7， 14 and 28 days after birth were 0.864 （95% CI 0.776-0.952）， 0.860 （95% CI 0.774-0.946） and 0.863 （95% CI 0.783-0.944）， respectively. A nomogram was constructed based on high risk factors， the calibrated curve was close to the reference line， and the predicted value was close to the actual value， indicating good calibration of the model. Conclusions The risk factors for moderate/severe BPD in preterm infants vary slightly on 7， 14 and 28 days after birth， mainly the birth weight， maternal pre-pregnancy BMI ≥ 25 kg/m2， 1-minute Apgar score≤7， pneumonia， preeclampsia， hsPDA and days of invasive mechanical ventilation. The predictive models based on different factors at different time points have good predictive value， which can provide guidance for clinical practice.

Keywords

• extremely preterm infant|bronchopulmonary dysplasia|risk factor|prediction model