妊娠糖尿病围生儿不良结局的风险预测模型构建与验证

Construction and validation of risk prediction model for adverse perinatal outcomes of gestational diabetes mellitus

目的 探讨妊娠糖尿病围生儿发生不良结局的影响因素,构建预测模型并进行验证。方法 回顾性分析宿迁市第一人民医院2020年10月至2022年9月收治的330例妊娠糖尿病孕妇的临床资料,按照2∶1的比例将其分为建模组(220例)和验证组(110例),并根据围生儿是否发生不良结局将其分为不良组和良好组。采用多因素Logistic回归性分析法分析围生儿发生不良结局的影响因素,并构建预测列线图模型;采用Bootstrap法以及绘制校准曲线、受试者操作特征(ROC)曲线、决策曲线(DCA)对模型进行验证。结果 经多因素Logistic回归性分析,年龄≥35岁、体质量指数≥25 kg/m^(2)、不良孕产史、糖尿病家族史、妊娠高血压、糖化血红蛋白>5.5%均是妊娠糖尿病围生儿发生不良结局的独立危险因素(P<0.05);将上述影响因素作为预测指标,构建妊娠糖尿病围生儿发生不良结局的风险预测列线图模型,建模组和验证组列线图模型预测校准曲线均接近标准曲线;ROC曲线和DCA曲线显示,列线图模型具有较好的预测效能和临床效益。结论 年龄≥35岁、孕前BMI≥25 kg/m^(2)、高血压、不良孕产史、糖尿病家族史、HbA1c>5.5%均是妊娠糖尿病围生儿发生不良结局的影响因素,据此构建的列线图预测模型可帮助临床早期识别高风险人群,临床实用性高。

Objective To explore the influencing factors of adverse perinatal outcomes in gestational diabetes mellitus,and to establish a prediction model and verify it.Methods The clinical data of 330 pregnant women with gestational diabetes mellitus admitted in the First People's Hospital of Suqian from October 2020 to September 2022 were retrospectively analyzed,and they were divided into modeling group(220 cases)and validation group(110 cases)according to the ratio of 2:1.They were divided into poor group and good group according to whether the perinatal child had adverse outcomes.The influencing factors of perinatal adverse outcomes was analyze by multivariate Logistic regression analysis.The predictive nomograph model was constructed,and the model was validated by Bootstrap method,calibration curves,receiver operating characteristic(ROC)curves,and decision making curves(DCA).Results Multiple Logistic regression analysis showed that age≥35 years old,pre-pregnancy body mass index≥25 kg/m^(2),adverse pregnancy history,family history of diabetes,gesta tional hypertension,and glycosylated hemoglobin>5.5%were all independent risk factors for perinatal adverse outcomes of gestational diabetes mellitus(P<0.05).The above factors were used as predictive indicators to construct a line graph model for predicting the risk of adverse outcomes in perinatal infants with gestational diabetes mellitus.The calibration curves of the nomogram model prediction in the modeling group and the validation group were close to the standard curve.ROC curve and DCA curve showed that the nomogram model had good predictive efficacy and clinical benefit.Conclusion Age≥35 years old,pre-pregnancy body mass index≥25 kg/m^(2),hypertension,adverse pregnancy history,family history of diabetes,and glycosylated hemoglobin>5.5%are all factors affecting the adverse outcome of gestational diabetes mellitus in perinatal infants.The prediction model constructed based on the histogram can help to identify high-risk groups in early clinical stage,and has high clinical applicability.