| Research PurposeThe risk factors of bronchopulmonary dysplasia(BPD)in premature infants were discussed through prospective studies of clinical data,and a predictive model was constructed by combining pulmonary ultrasound examination to analyze its predictive value for the occurrence of BPD.Research methodsTo be admitted to our department between October 1st,2019 and June 30 st,2020,the gestational age less than 32 weeks should be admitted within 24 hours after birth.The preterm infants with chromosomal abnormalities,severe malformations,genetic metabolic diseases,death before 36 weeks of gestational age or automatic discharge,and surgical diseases during hospitalization were excluded as subjects.Maternal pregnancy,general conditions of the child,basic diseases,blood transfusion,fluid inflow and outflow in the first week after birth,and respiratory support were recorded to analyze risk factors for BPD.All newborn babies were examined by Lung Ultrasound(LUS)one week after birth and were recorded anonymously,graded by another doctor who had experience in Lung ultrasound.Divide into BPD group and non-BPD group according to whether BPD finally occurs,analyze the data of the two groups with multi-factor logistic regression analysis method,and build a predictive model,and use Receiver Operating Characteristic(ROC)curve to predict the performance of the model to evaluate.Research results1.A total of 88 newborns were included in this study,including 30(34.09%)females and 58(65.9%)males,among whom 34 patients eventually developed BPD,with an incidence of 38.6%.2.Univariate analysis showed that: There were statistically significant differences in Gestational age,birth weight,Neonatal Respiratory Distress Syndrome(NRDS),Score for Neonatal Acute Physiology-Ⅱ(SNAP-Ⅱ),birth asphyxia rate,invasive ventilation rate,high concentration of oxygen uptake rate,rate of multiple transfusions,LUS Score between the BPD group and the non-BPD roup.The multivariate logistic regression analysis using the above factors as independent variables obtained: Small gestational age at birth,high SNAP-Ⅱ score and high LUS score are independent risk factors.3.Build prediction models: Combined predictor= 29.08 + SNAP-Ⅱ × 0.151-Gestational age ×1.108 + LUS score×0.272,to calculate the incidence of BPD.The Area Under Curve(AUC)was 0.949,the sensitivity(94.1%)and specificity(90.7%)were higher than the other three indexes separately,and the 95% confidence interval was0.904～0.995.Kappa test was conducted for the consistency between the prediction results of the prediction model and the final diagnosis results,and the Kappa value was0.858>0.75.Research conclusionSmall gestational age,high SNAP-Ⅱ score,after 1 week LUS score after birth have a higher risk of subsequent development of BPD;The prediction model constructed including gestational age at birth,SNAP-Ⅱ,and LUS score can better predict the occurrence of BPD. |
