基于孕前危险因素的妊娠期糖尿病风险预测模型的构建及效能验证  

Construction and efficiency validation of a risk predictive model for gestational diabetes mellitus based on preconception risk factors

作　　者：晋凤珍 刘小利[1] 杨元沛 孙俊杰[1] 李睿越 JIN Fengzhen;LIU Xiaoli;YANG Yuanpei;SUN Junjie;LI Ruiyue(Department of Obstetrics and Gynecology,Chongqing Health Center for Women and Children,Women and Children's Hospital of Chongqing Medical University,Chongqing 401147,China)

机构地区：[1]重庆市妇幼保健院,重庆医科大学附属妇女儿童医院妇产科,重庆401147

出　　处：《山东医药》2022年第32期6-10,共5页Shandong Medical Journal

基　　金：重庆医科大学附属妇女儿童医院青年培育项目(2020YJQN08)。

摘　　要：目的构建基于孕前危险因素的妊娠期糖尿病(GDM)风险预测模型并验证其预测效能。方法收集1762例孕妇的临床资料,根据口服葡萄糖耐量试验结果,将孕妇分为GDM组792例和糖耐量正常组970例。比较两组孕妇年龄、孕次、孕前空腹血糖、孕前血压、孕前1年体质量增加量、孕前体质量指数(BMI)、孕前饮食及生活方式等。采用Logistic回归法分析影响GDM发生的孕前危险因素,构建基于孕前危险因素的GDM风险预测模型。采用受试者工作特征(ROC)曲线检验模型的预测效能。结果孕前年龄≥35岁(X_(1))、孕前舒张压≥80 mmHg(X_(2))、孕前空腹血糖≥5.1 mmol/L(X_(3))、孕前BMI≥24 kg/m^(2)(X_(5))、孕前1年体质量增加≥10 kg(X_(6))、家庭年收入≥30万元(X_(8))、孕次≥3次(X_(9))、孕前每日主食摄入量≥300 g(X_(10))、孕前喜食甜食(X_(11))、糖尿病家族史(X_(15))是影响GDM发生的孕前危险因素,孕前BMI<18.5 kg/m^(2)(X_(4))、文化程度高(X_(7))、孕前每日运动持续时间≥1 h(X_(12))、孕前高运动强度(X_(13))、孕前睡眠质量良好(X_(14))是保护性因素。根据变量回归系数,获得孕前预测GDM发生风险的模型方程:P_(GDM)=1/{1+exP[-(-0.083+1.060X_(1)+0.771X_(2)+1.022X_(3)-0.574X_(4)+0.567X_(5)+0.898X_(6)-0.720X_(7)+0.773X_(8)+0.869X_(9)+0.868X_(10)+1.827X_(11)-0.452X_(12)-0.942X_(13)-1.206X_(14)+0.662X_(15))]}。应用ROC曲线检验模型预测效能,ROC曲线下面积为0.854(95%CI 0.836~0.872)、灵敏度为0.708、特异度为0.856,Hosmer-Lemeshow拟合优度检验P=0.683。结论基于孕前危险因素的GDM风险预测模型具有较好的预测效能,有助于孕前筛查GDM高危人群。Objective To construct a risk predictive model for gestational diabetes mellitus(GDM)based on preconception risk factors and to validate its efficiency.Methods The clinical data of 1762 early pregnant women were collected,and the pregnant women were divided into the GDM group(n=792)and normal glucose tolerance(NGT)group(n=970)according to the results of oral glucose tolerance test(OGTT).We compared the age before pregnancy,diastolic blood pressure before pregnancy,FPG before pregnancy,weight gain in 1 year before pregnancy,BMI before pregnancy,diet and lifestyle before pregnancy and other factors.Logistic regression analysis was used to determine the progestational risk factors of GDM.A risk prediction model for GDM based on progestational risk factors was constructed.The efficiency of the model was validated by receiver operation characteristic(ROC)curve.Results Age≥35 years(X_(1)),diastolic blood pressure≥80 mmHg(X_(2)),fasting plasma glucose(FPG)≥5.1 mmol/L(X_(3)),BMI≥24 kg/m^(2)(X_(5)),weight gain≥10 kg in 1 year before pregnancy(X_(6)),family annual income≥￥300000(X_(8)),pregnancy≥3 times(X_(9)),daily staple food intake≥300 g(X_(10)),fond of sweets(X_(11)),and family history of diabetes(X_(15))were independent preconception risk factor for GDM;BMI<18.5 kg/m^(2)(X_(4)),master degree or above(X_(7)),daily sport duration≥1 hour(X_(12)),high-intensity sport(X_(13)),and good sleep quality(X_(14))were obvious protective factors(P<0.05 or P<0.01).According to the logistic regression analysis results,the equation of preconception risk prediction model for GDM were obtained:P_(GDM)=1/{1+exP[-(-0.083+1.060X_(1)+0.771X_(2)+1.022X_(3)-0.574X_(4)+0.567X_(5)+0.898X_(6)-0.720X_(7)+0.773X_(8)+0.869X_(9)+0.868X_(10)+1.827X_(11)-0.452X_(12)-0.942X_(13)-1.206X_(14)+0.662X_(15))]}.We used the ROC curve to validate its efficiency.The area under the ROC curve(AUC),sensitivity and specificity of the risk prediction model were 0.854(95%CI:0.836–0.872),0.708,and 0.856,respectively;and the Hosmer-Lemeshow test s

关 键 词：妊娠期糖尿病 妊娠期糖尿病风险预测模型 孕前保健 

分 类 号：R714.25[医药卫生—妇产科学]