妊娠糖尿病不良结局的影响因素与预测模型的构建及验证

lnfluencing factors construction and verification of adverse outcome prediction model for gestational diabetes mellitus

目的分析妊娠糖尿病不良结局的影响因素,根据影响因素创建预测模型。方法选取2021年3月至2022年3月入同济大学附属第一妇婴保健院产科门诊产前检查并住院分娩的诊断为妊娠糖尿病的孕产妇370例,根据妊娠糖尿病患者孕期或者产后是否发生不良结局,将建模的对象分为两组,正常结局组(n=228)和不良结局组(n=142)。采用单因素分析和多因素Logistic回归分析妊娠糖尿病不良结局的危险因素,并建立预测模型。绘制受试者工作特征(ROC)曲线对预测妊娠糖尿病患者不良结局风险预测模型进行预测效能评价,以敏感度和特异度、总正确率验证模型的实际应用效能。结果单因素分析和多因素Logistic回归分析结果显示,年龄≥35岁、妊娠史≥2次、孕前体重指数>28 kg/m^(2)、有糖尿病家族史、分娩方式为剖宫产、流产次数2次以上和糖耐量试验(OGTT)血糖3项异常指标是影响妊娠糖尿病患者发生不良结局的危险因素,并据此构建预测模型。列线图模型可以较好地预测妊娠糖尿病患者发生的不良结局,曲线下面积为0.938(95%CI:0.897~0.975),敏感度为0.873,特异度为0.792,表明列线图模型具有较好的预测准确度;Hosmer Lemeshow检验显示χ^(2)=4.178,P=0.146,说明模型校准度良好。模型的校准图显示不良结局的概率与实际观察到的概率之间存在较好的一致性。外部验证165例妊娠糖尿病患者发生不良结局为61例。该风险预测模型预测验证组发生预后不良的敏感度为89.27%,特异度为92.48%。模型预测总正确率为90.91%。结论年龄≥35岁、妊娠史≥2次、孕前体重指数>28 kg/m^(2)、有糖尿病家族史、分娩方式为剖宫产、流产次数2次以上和OGTT血糖3项异常是妊娠糖尿病不良结局的影响因素,据此构建的列线图预测模型对妊娠糖尿病不良结局具有较高的预测效能和临床应用性,对妊娠糖尿病不良结局的识别具有一定的潜在意义。

Objective To analyze the influencing factors of adverse outcomes of gestational diabetes mellitus and establish a prediction model according to the influencing factors and its verification.Methods A total of 370 pregnant and lying in women who were admitted to the obstetric clinic of the First Maternal and Child Health Hospital affiliated to Tongji University from March 2021 to March 2022 and were diagnosed as gestational diabetes were selected for prenatal examination and hospital delivery.According to whether the gestational diabetes patients had adverse outcomes during pregnancy or postpartum,the objects of modeling were divided into two groups:normal outcome group(n=228)and adverse outcome group(n=142).The risk factors of adverse outcomes of gestational diabetes mellitus were analyzed by using Univariate analysis and multivariate Logistic regression,and the prediction model was established.The predictive efficacy of the predictive model for predicting the risk of adverse outcomes in patients with gestational diabetes mellitus was to evaluated by drawing receiver operating characteristic(ROC)curve,and the practical application efficacy of the model was verified by sensitivity,specificity and total accuracy.Results Univariate analysis and multivariate Logistic regression analysis showed that age≥35 years old,pregnancy history≥2 times,pre-pregnancy body mass index>28 kg/m^(2),family history of diabetes,delivery mode of cesarean section,frequency of abortion more than 2 times and OGTT blood glucose were the risk factors for adverse outcomes in gestational diabetes patients.And construct the prediction model.The neomorph model could well predict the adverse outcomes of gestational diabetes mellitus patients.The area under the curve was 0.938(95%CI:0.897-0.975),sensitivity was 0.873,and specificity was 0.792,indicating that the neomorph model had good prediction accuracy.Hosmer Lemeshow test showed χ^(2)=4.178,P=0.146,indicating that the model was well calibrated.The calibration plot of the model shows a good agreement between the probability of adverse outcomes and the actual observed probability.External verification of 165 patients with gestational diabetes mellitus had adverse outcomes in 61 cases.The sensitivity and specificity of the risk prediction model to predict poor prognosis in the validation group were 89.27%and 92.48%.The total accuracy of the model was 90.91%.Conclusion Age≥35 years old,gestational history≥2 times,pre-pregnancy body mass index>28 kg/m^(2),family history of diabetes,delivery mode of cesarean section,frequency of abortion more than 2 times and abnormal OGTT blood glucose were the influential factors for adverse outcomes of gestational diabetes.The nomogram prediction model has high predictive efficiency and clinical applicability for the adverse outcomes of gestational diabetes mellitus,and has certain potential significance for the identification of adverse outcomes of gestational diabetes mellitus.