Development And Validation Of A Model For Predicting Bronchopulmonary Dysplasia In Very-low-birth-weight Infants

Background and objectivesBronchopulmonary dysplasia(BPD)is a common chronic lung syndrome in newborns,especially a higher incidence in very low birth weight infants.With the development of neonatal intensive care and respiratory support treatment and the high incidence of premature infants,the survival rate of them has increased significantly,while the prevalence of BPD raised.Studies have shown that BPD is a syndrome caused by complex factors,and there still no effective drugs or methods in treatment.Once it occurs,the life quality of newborn would be seriously affected.Therefore,how to prevent the occurrence of BPD has become a hot point in recent study.Our study aims to establish a prediction model for early identification of the risk of BPD based on the early clinical characteristics and biochemical indicators of VLBW infants,so as to improve clinical treatment and effectively reduce the incidence of BPD.MethodsIn this single-center retrospective study performed from January 1st 2017 to December 31st 2020,we reviewed infants who admitted to the neonatal intensive care unit department Ⅱ immediately.And their gestational ages were below 32 weeks,birth weight were below 1500g.We divided premature infants admitted to the hospital before May 2019 into the training set,and the rest were included in the validation set.The outcome of the study was the occurrence of BPD or death.According to the outcome of the study,they were divided into non-BPD group and BPD/death group,and single factor analysis was performed to compare the basic clinical characteristics between two groups.Significant risk factors of univariate analysis in the training set were included in LASSO regression to screen predictors,logistic regression used to construct a predictive model,and the probability of BPD or death in individual patients could be visually predicted through nomogram.Calculate the C index to evaluate the discrimination,calibration curve and the Hosmer-Lemeshow test were used to evaluate the calibration in training set and validation set respectively.Results1.From January 2017 to December 2020,499 very low birth weight infants whose gestational age was below 32 weeks were admitted to the First Affiliated Hospital of Zhengzhou University.Five patients were diagnosed with congenital diseases and metabolic disorders.In addition,59 patients were excluded because of incomplete medical records.A total of 435 preterm infants were included in this study,and 20 infants died before the BPD criteria could be evaluated.We divided 362 infants into a training set and 73 into a validation set according to the admission data(May 2019).2.Among the 362 children in the training set,195 children were diagnosed with BPD,and 19 cases died.Univariate analysis found that 14 variables were significantly different(all P<0.05),including gestational age,birth weight,delivery route,1 minute and 5 minutes Apgar scores,NT-proBNP levels on the 3rd and 7th days,the duration of mechanical ventilation in the first week,the usage of pulmonary surfactant,neonatal asphyxia,heart failure,respiratory failure,pneumonia,pulmonary hemorrhage,meconium staining and gestational hypertension.3.After LASSO regression analysis,gestational age,the time of mechanical ventilation and the level of NT-proBNP on the 7th day after birth were finally selected.Logistic regression was used to construct a logarithmic equation and draw a nomogram.We ploted the ROC curve calculating the AUC was 0.867,and the best cut-off was 0.476,which the specificity and sensitivity was 75.7%and 83.2%respectively.The bootstrap method(n=1000)is used for internal verification.The C index of the model was calculated as 0.853(95%confidence interval:0.851-0.854)and 0.855(95%confidence interval:0.77-0.94)respectively in the training set and the validation set.The calibration curve and Hosmer-Lemeshow test(training set:P=0.40,validation set P=0.48)showed good calibration.Conclusions1.Gestational age,the time of mechanical ventilation and the level of NT-proBNP on the 7th day were the risk factors of BPD or death.2.In this study,we constructed a well-performing nomogram based on risk factors to assess the risk of BPD in VLBW infants under 32 weeks of gestational age at 7 days postnatal with a specificity of 75.7%and a sensitivity of 83.2%.