Development And Verification Of The Risk Predict Model Of Bronchopulmonary Dysplasia In Very Low Birth Weight Infants

BackgroundWith the continuous improvement of perinatal medicine and treatment technology for critically ill newborns,the survival rate of very low birth weight infant(VLBWI)and extremely low birth weight infant(ELBWI)has become higher and higher,leading to the increasing incidence of bronchopulmonary dysplasia(BPD).BPD,as a major public health problem,seriously threatens the prognosis and quality of life of children,and brings a heavy burden to families and society.Because the etiological mechanism of BPD is not clear at present,and there is no clinical targeted treatment method,the prevention and control of BPD should focus on early prevention.Therefore,the risk factors for the occurrence of BPD should be analyzed,a risk prediction model should be established,and infants at risk of BPD should be screened out as early as possible to change passive treatment into active prevention.In this study,a retrospective study was conducted to establish a prediction model for the occurrence risk of BPD based on the clinical data easily obtained after admission,so as to provide reference for the early prevention of BPD in children.Objectives1.To describe the incidence of BPD in very low birth weight infants,and to analyze the risk factors for BPD.2.The risk prediction model of VLBW infants was established to predict the occurrence risk of BPD,so as to provide basis for early clinical identification and individual prevention.Methods1.The convenience sampling method was used to retrospectively investigate 611 VLBW infants who were admitted to the NICU of a tertiary grade A hospital in Su Zhou from January 2017 to September 2019.In a 7:3 ratio,they were randomly divided into a modeling group and a verification group using a computer.Application modeling set data to construct a VLBW infants risk prediction model of BPD:according to whether infants BPD modeling data set can be divided into BPD group and non-BPD,VLBW used BPD infants with related factors questionnaire to collect data,including the general clinical data,motherhood,children hospitalization treatment condition and main complications after birth.Excel software was used to input data,establish a database,and SPSS22.0 statistical software was used for statistical analysis.Univariate analysis was used to compare the clinical data of the two groups of children with BPD and those without BPD.Variables with statistical significance in univariate analysis(P < 0.05)were included in binary Logistic regression analysis to establish independent risk factors for the occurrence of BPD(P< 0.05),and a risk prediction model for BPD in VLBW infants was established.2.Hosmer-Lemeshow(H-L)test was performed on the prediction model to verify the fitting effect,with P > 0.05,indicating that the test passed and the fitting effect was good.The area under receiver operating characteristic curve(ROC)was used to evaluate the prediction performance of the model.The data of the validation set were used as independent data to verify and evaluate the established BPD risk prediction model,and the area under the ROC curve was calculated.Results1.Basic information A total of 611 VLBW infants were collected,including 315males(51.6%)and 296 females(48.4%).In a 7:3 ratio,451 patients were randomly divided into the modeling set and 160 patients in the validation set,and the incidence of BPD in the modeling set was 14.9%(63 cases).2.Constructing the risk prediction model of BPD Single factor analysis showed that the P values of sex,childbirth way,small gestational age(SGA),using the alveolar surface active substances(PS),using caffeine/aminophylline,blood transfusion ≥3,neonatal respiratory distress syndrome(NRDS),neonatal asphyxia,sputum culture positive,clinical sepsis,pulmonary hemorrhage,patent ductus arteriosus(PDA),intraventricular hemorrhage(IVH),retinopathy of prematurity(ROP),necrotizing enterocolitis(NEC),gestational age,birth weight,Apgar score,length of hospital stay,duration of invasive mechanical ventilation,continuous positive airway pressure(CPAP),total oxygen use time,parenteral nutrition time and enteral nutrition time were all < 0.05,showing statistically significant differences between BPD children and non-BPD patients.Logistic regression analysis showed that neonatal asphyxia(OR=3.419,95%CI:1.198～9.757),positive sputum culture(OR=3.541,95%CI:1.195～10.497),clinical sepsis(OR=5.351,95%CI:2.05～13.963),NRDS(OR=6.35,95%CI:1.939～20.792),blood transfusion ≥ 3 times(OR=4.285,95%CI:1.588～11.568),PDA(OR=6.266,95%CI: 1.656～23.704),invasive mechanical ventilation time(OR=1.13,95%CI:1.072～1.192),oxygen use time(OR=1.095,95%CI: 1.055～1.136),parenteral nutrition time(OR=1.059,95%CI: 1.002～1.099)were independent risk factors,and 1-minute Apgar score was a protective factor(OR=0.778,95%CI:0.627～0.966).The model was established according to the results of Logistic regression analysis,and the model formula was Z= neonatal asphyxia *1.229+ positive sputum culture *1.265+ clinical sepsis*1.677+NRDS*1.848+blood transfusion(≥3 times)*1.455+PDA*1.835-1 min Apgar score *0.25+ time of invasive mechanical ventilation *0.123+ time of oxygen use*0.09+ time of parenteral nutrition *0.057-8.077.The Hosmer-Lemeshow goodness of fit test showed that the P value was 0.983,the area under the ROC curve was 0.965(95%CI: 0.946,0.983),P < 0.001,the sensitivity was 93.7%,and the specificity was91.3%.3.Validating the risk prediction model of BPDThe data of the validation set(160 cases)were substituted into the above model formula and the ROC curve was drawn.The results showed that the area under the ROC curve was 0.89,the sensitivity was 92.9%,and the specificity was 76%.ConclusionsIn this study,a prediction model with 10 easy-to-evaluate indicators as variables was finally constructed,which has good predictive efficiency and certain clinical value,and can guide medical staff to accurately complete the risk assessment of BPD in VLBW infants,formulate targeted preventive measures,and reduce the incidence of BPD.