摘要
目的构建新生儿呼吸窘迫综合征(respiratory distress syndrome,RDS)患儿早期有创通气的预测模型。方法回顾性选择2021年9月至2022年8月徐州医科大学附属医院新生儿重症监护室收治的胎龄≥26周、最终诊断为RDS的新生儿构建预测模型,通过简单随机抽样方法按照7∶3的比例分为训练集和测试集,根据入院后24 h内是否进行有创通气将训练集分为有创通气组和无创通气组,比较两组新生儿的一般情况、实验室检查结果、母孕期及分娩情况,通过在训练集中应用多因素logistic回归筛选最终预测因子进入最终模型,绘制列线图,并结合测试集对最终模型进行区分度、拟合度和临床决策曲线的评价。再前瞻性纳入2023年9~10月徐州医科大学附属医院新生儿科收治的RDS患儿进行模型的外部验证。结果共纳入269例新生儿构建模型,训练集188例,测试集81例,对训练集进行logistic回归分析显示,吸入氧浓度高、动脉血氧分压(partial pressure of oxygen in artery,PaO_(2))/吸入氧浓度(fraction of inspired oxygen,FiO_(2))(P/F指数)低、血钠水平低、白蛋白水平低、谷草转氨酶水平高是RDS患儿早期有创通气的主要风险因素(P<0.05)。纳入相关预测因子后的列线图模型具有较好的区分度(训练集AUC=0.898,测试集AUC=0.835)、拟合度、获益率。外部验证共纳入32例RDS患儿,列线图模型同样具有较好的预测价值(AUC=0.945)。结论高吸入氧浓度、低P/F指数、低血钠水平、低白蛋白水平、高谷草转氨酶水平是RDS患儿早期应用有创通气的主要风险因素,可用于RDS有创通气风险预测模型的构建,该模型具有较好的区分度、拟合度及获益率,在外部验证中同样具有较好的预测价值,值得临床推广。
ObjectiveTo establish a predictive model of early invasive ventilation(IV)for neonatal respiratory distress syndrome(RDS).MethodsFrom September 2021 to August 2022,neonates(gestation age≥26 weeks)diagnosed of RDS and admitted to the NICU of our hospital were retrospectively studied.The data from the neonates were randomly sampled into training set and validation set with a 7:3 ratio.The training set was subgrouped into IV group and non-invasive ventilation(non-IV)group based on the treatment strategy within 24 h of admission.The general condition,lab results,maternal and perinatal data were compared between the two groups.For the training set,multivariate logistic regression analysis was used to identify the predictive factors for the final model.A nomogram was developed and discrimination ability,calibration efficacy and clinical decision curve were evaluated using the validation set.Additionally,the model was prospectively evaluated using data from RDS patients during September and October,2023 as external validation.ResultsA total of 269 neonates were enrolled,including 188 as training set and 81 as validation set.Logistic regression analysis of the training set showed that high inspired oxygen concentration(FiO_(2)),low arterial oxygen partial pressure(PaO_(2))/FiO_(2)ratio(P/F ratio),low levels of serum sodium and albumin and high aspartate aminotransferase(AST)were predictive factors for early IV in RDS patients(P<0.05).The nomogram model incorporating these factors demonstrated good discrimination ability,calibration efficacy and clinical benefit rate.The nomogram model also had good predictive value in external validation using data from 32 new RDS neonates,with an AUC of 0.945.ConclusionsHigh FiO_(2),low P/F ratio,low serum sodium and albumin and high AST are risk factors for early IV in RDS neonates.The prediction model based on these factors has good discrimination ability,calibration efficacy and clinical benefit rate and excel in external validation.
作者
茆宁
薛玉恒
胡湘然
孟孟
刘文强
徐艳
Mao Ning;Xue Yuheng;Hu Xiangran;Meng Meng;Liu Wenqiang;Xu Yan(Department of Neonatology,Hospital of Xuzhou Medical University,Xuzhou 221000,China)
出处
《中华新生儿科杂志（中英文）》
CAS
CSCD
2024年第8期454-460,共7页
Chinese Journal of Neonatology
关键词
新生儿呼吸窘迫综合征
有创通气
列线图
预测模型
Neonatal respiratory distress syndrome
Invasive ventilation
Nomograms
Prediction mode
