妊娠期糖尿病四种风险预测模型的外部验证及比较

External Validation and Comparison of Four Risk Prediction Models for Gestational Diabetes Mellitus

目的:外部验证4种妊娠期糖尿病(GDM)患病风险预测模型在临床应用的可行性。方法:纳入本院产检并分娩的478例符合要求的孕妇为外部验证组,采用受试者工作特征曲线下面积(AUC)评价模型的区分度,采用Homser-Lemeshow拟合优度检验和Brier得分衡量模型的校准度,再使用决策曲线分析(DCA)比较4个模型的临床效用。结果:仅有模型2的AUC>0.75,且经Homser-Lemeshow拟合优度检验后P>0.05,4个模型的Brier得分分别为0.243、0.200、0.241和0.240,4个模型的DCA表明使用模型2得到的临床效益最大。结论:与其余3个模型相比,模型2提高了GDM风险预测模型的预测效能。

Objective To externally verify the feasibility of clinical application of 4 models for predicting the risk of developing gestational diabetes.Methods A total of 478 eligible pregnant women who were enrolled in our hospital for prenatal examination and delivery were included as an external validation group.The area under the working characteristic curve(AUC)of the subjects was used to evaluate the differentiation of the model.The Homser-Lemeshow goodness of fit test and Brier score were used to measure the calibration of the model,and then the decision curve analysis(DCA)was used to compare the clinical efficacy of the four models.Results Only model 2 had an AUC>0.75 and P>0.05 after the Homser-Lemeshow goodness-of-fit test.The Brier scores of the four models were 0.243,0.200,0.241 and 0.240,respectively.The DCA of the four models showed that model 2 had the greatest clinical benefit.Conclusion Compared with the other three models,Model 2 improved the prediction efficiency of GDM risk prediction model.