Construct A Risk Prediction Model Of GDM Based On The Routine Obstetric Examination Information In The First Trimester Of Pregnancy

[Objective]Analyzing the correlation between the information of routine obstetric examinations in the first trimester of pregnancy and the incidence of gestational diabetes mellitus,to explore a model to predict the risk of GDM in the first trimester based on the information of routine obstetric examination,assess the predictive value of the model for the risk of GDM,and provide a reference for establishing early screening criteria of gestational diabetes.[Methods]Through the electronic case system and the inspection system to collect the information of pregnant women who gave birth in the obstetric department of Shenzhen Second People’s Hospital from January 1,2019 to December 31,2020.This retrospective study conducted on 923 pregnant women(135 with GDM and 788controls),and the information includes(1)demographic characteristics such as age and educational attainment;(2)clinical features such as gravidity,parity,in vitro fertilization-embryo transfer(IVF-ET),family history of diabetes,height,pregestational weight,pre-gestational body mass index(BMI);(3)biomarkers before14 gestational weeks such as fasting plasma glucose(FPG),glycated hemoglobin A1c(Hb A1c),pregnancy related plasma protein-A(PAPP-A)and pregnancy related plasma protein-A multiple of the median(PAPP-A Mo M)value.We used SPSS 25.0 and Empower Stats 3.0 for statistical analysis to estimate the coefficients of each risk factor,mutually-adjusted odds ratio(OR)assigned for GDM,and construct a risk prediction model of GDM by multiple Logistic regression analysis.We used Hosmer?Lemeshow test,calibration curve,receiver operator characteristics(ROC)curve and decision curve analysis to assess the value of the model,and we explored the best cutoff of the model.Then bootstrap was used to finish the internal verification.[Results]1.Between the GDM group and the controls,age,gravidity,IVF-ET,family history of diabetes,pre-gestational weight,pre-gestational BMI,FPG,Hb A1 c,PAPPA and PAPP-A Mo M value before 14 gestational weeks were statistically different.2.Comparing the pregnancy outcomes between the GDM group and the controls,the incidence of preterm birth and the ways of delivery including both induction of labor and cesarean in the GDM group was higher than those in the controls.3.The results of univariate Logistic regression analysis suggested that age,gravidity,IVF-ET,family history of diabetes mellitus,pre-gestational weight,pregestational BMI,FPG,Hb A1 c,PAPP-A and PAPP-A Mo M value before 14 gestational weeks were associated with GDM.4.A risk prediction model was constructed by multiple Logistic regression analysis which included age,educational attainment,parity,family history of diabetes mellitus,pre-gestational BMI,FPG,Hb A1 c,PAPP-A Mo M value before 14 gestational weeks.In this prediction model,educational attainment,parity and PAPP-A Mo M value before 14 gestational weeks were the protective factors,and the rest were independent risk factors.The area under the curve(AUC)of the risk prediction model was 0.741(95% CI: 0.695 to 0.793).We explored the best cutoff of the model was 0.147 with an accuracy of 0.608,a sensitivity of 0.793,and a specificity of 0.576.HosmerLemeshow test and calibration curve showed good consistency of the model.At the same time,DCA results showed positive clinical benefits,and the internal verification results suggested that the model demonstrated good predictive efficacy in the population.[Conclusion]1.In the information of routine obstetric examinations in the first trimester of pregnancy,Age,educational attainment,parity,family history of diabetes mellitus,pregestational BMI,FPG,Hb A1 c and PAPP-A Mo M value before 14 gestational weeks were correlated with the incidence of GDM.2.It is feasible to construct a risk prediction model of GDM based on the information of routine obstetric examinations in the first trimester of pregnancy.3.This risk prediction model of GDM,which including age,educational attainment,parity,family history of diabetes mellitus,pre-gestational BMI,FPG,Hb A1 c and PAPP-A Mo M value before 14 gestational weeks information of pregnant women in the first trimester of pregnancy can predict the risk of GDM and has good predictive efficacy.