妊娠期糖尿病患者并发子痫前期的风险预测列线图模型建立和验证

Establishment and validation of nomogram model for risk prediction of gestational diabetes mellitus complicating preeclampsia

目的建立妊娠期糖尿病(gestational diabetes mellitus,GDM)患者并发子痫前期(preeclampsia,PE)的风险预测列线图模型,同时对其预测效果进行验证。方法选取2021年1~7月在重庆市江津区妇幼保健院产检并分娩的352例GDM患者作为研究对象。根据是否发生PE将患者分为PE组(42例)及无PE组(310例)。收集患者相关临床资料,采用单因素分析患者基本资料、实验室指标等对GDM患者发生PE的影响,采用Logistic回归分析GDM并发PE的危险因素。将确定的危险因素通过R 3.5.3软件,利用rms程序包绘制预测GDM患者发生PE的列线图预测模型,然后利用受试者工作曲线(ROC)下面积对列线图模型预测效能进行检验,采用Hosmer-Lemeshow检验模型的拟合优度。另再纳入2021年8~12月本院产检并分娩的129例GDM患者对模型预测效能进行验证。结果单因素分析结果显示,PE组患者孕前BMI、FPG、HbA1c、TG、TC、UA、Hcy水平及双胎妊娠占比高于无PE组(P<0.05)。Logistic回归分析结果显示孕前BMI、FPG、TG、TC、Hcy均是影响GDM患者发生PE的危险因素(P<0.05)。根据Logistic多因素结果建立预测GDM患者发生PE的列线图模型,并对模型进行验证。采用ROC曲线检验预测模型的区分度,结果显示该预测模型ROC曲线下面积0.823、95%CI(0.810-0.876)、灵敏度为76.38%、特异度为85.42%(P<0.001)。Hosmer-Lemeshow检验得χ^(2)=4.315,P=0.324。129例患者中21例发生PE,风险预测模型预测发生PE者为19例,ROC曲线下面积为0.806、95%CI(0.783-0.869)、敏感度为73.69%和特异度为82.71%。采用Hosmer-Lemeshow检验模型拟合优度,结果显示χ^(2)=5.679,P=0.328,说明该模型拟合优度较好。结论临床中应充分考虑GDM患者孕前BMI、FPG、TG、TC、Hcy因素,综合评估其并发PE的概率,提高临床对GDM患者并发PE的诊断效能。

Objective To establish a nomogram model for risk prediction of preeclampsia(PE)in patients with gestational diabetes mellitus(GDM),and to verify its prediction effect.Methods From January to July 2021,352 patients with GDM who underwent prenatal examination and delivery at Jiangjin District Maternal and Child Health Hospital of Chongqing were selected as the study objects.According to whether PE occurred,the patients were divided into PE group(42 cases)and non-PE group(310 cases).The relevant clinical data of patients were collected,and the influence of basic data and laboratory indicators on the occurrence of PE in GDM patients was analyzed by univariate analysis.The risk factors of GDM complicated with PE were analyzed by Logistic regression.The identified risk factors were incorporated into the R3.5.3 software,and the rms package was used to draw a nomogram prediction model for predicting PE in GDM patients.Then the area under receiver operating curve(ROC)was used to test the prediction efficiency of the nomogram model,and Hosmer-Lemeshow tested the goodness of fit of the model.In addition,129 patients with GDM who were examined and delivered in our hospital from August to December 2021 were included to verify the predictive efficacy of the model.Results Univariate analysis showed that the pre-pregnancy BMI,FPG,HbA1c,TG,TC,UA,Hcy levels,and the proportion of twin pregnancies in PE group were higher than those in non-PE group(P<0.05).Logistic regression analysis showed that pre-pregnancy BMI,FPG,TG,TC and Hcy were all risk factors for PE in GDM patients(P<0.05).A nomogram model was established to predict the occurrence of PE in GDM patients according to the Logistic multi-factor results,and the model was verified.ROC curve was used to test the differentiation of the prediction model.The results showed that the area under ROC curve of the prediction model was 0.823,95%CI(0.810-0.876),sensitivity was 76.38%,and specificity was 85.42%(P<0.001).Hosmer-Lemeshow testχ^(2)=4.315,P=0.324.PE occurred in 21 of 129 patients.The risk prediction model for PE in GDM patients predicted that 19 cases would develop PE,and the area under ROC curve was 0.806,95%CI(0.783-0.869),sensitivity was 73.69%,and specificity was 82.71%.Hosmer-Lemeshow was used to test the goodness of fit of the model,and the results showed thatχ^(2)=5.679 and P=0.328,indicating that the goodness of fit of the model was good.Conclusion Pre-pregnancy BMI,FPG,TG,TC and Hcy should be fully considered to comprehensively evaluate the incidence of PE in GDM patients,so as to improve the clinical diagnosis efficiency of PE in GDM patients,which has high clinical application value.