Establishment Of A Risk Model For Gestational Diabetes Mellitus Based On A Cohort Study

Objective:Data from a multicenter prospective cohort study were used to explore the relationship between early pregnancy lifestyle-related factors and gestational diabetes mellitus(GDM),constructing a risk model of gestational diabetes mellitus with early pregnancy-related lifestyle factors as the main independent variable.Methods:The Chinese Pregnant Women Cohort Study Peking Union Medical College(CPWCS PUMC)approved by the Ethics Review Committee of the Scientific Research Department of Peking Union Medical College Hospital and registered at ClinicalTrials.gov(NCT03403543).The research sites are located in 24 hospitals in 15 provinces across China.Since July 25,2017,investigators have begun recruiting pregnant women who came to the hospital for the first-time antenatal care in the first trimester.After the participants agreed to participate in the study and signed the informed consent form,they were instructed to complete the questionnaire.Then,investigators collected the antenatal care data,pregnancy history,disease history and family history.As of January 15,2019,we collected a total of 3875 first trimester questionnaires and pregnancy outcomes information.After removing 229 subjects who did not meet the analysis inclusion criteria,3646 subjects were eventually included.The questionnaire included general demographic data of pregnant women,oral care,smoking and drinking,living and working conditions,physical activity,major household cooking oil,dietary frequency,depression,and sleep disorders.The investigators collected clinical data(pregnancy history,past medical history,combined disease,family history,etc.)from the hospital information system(HIS)and the results of the 75g oral glucose tolerance test.Diagnose GDM according to the diagnostic criteria for gestational diabetes recommended by the International Diabetes and Pregnancy Research Group(IADPSG).All data is uniformly checkek.Statistical analysis was performed by using SAS V9.4(SAS Institute)statistical analysis software.Univariate logistic regression analysis was used to determine the possible risk factors of GDM.Using the training set(randomly extracting 80%of the subject data)to establish a risk model,the model’s dependent variable Y is considered as a GDM score,and the receiver operating characteristic curve(ROC)curve is drawn.The optimal scoring threshold of Y is selected with high sensitivity and a large Yoden index.The prevalence of predicted GDM was calculated using the validation set(the remaining 20%of the study data)and compared with the real situation to obtain the corresponding sensitivity,specificity,positive predictive value,and negative predictive value.Results:(1)Among the 3646 subjects,409(11.22%)are GDM patients,the average age was(28.75±4.28)years old,and 322(9.11%)were older than 35 years old.The average pre-pregnancy BMI was(21.79±3.27)kg/m2,600(16.46%)are overweight pregnant women and 167(4.58%)are obese pregnant women.There were 2,216 primiparas(60.78%),13(0.36%)pregnant women with a history of GDM,and 163(4.47%)pregnant women with first-degree relatives with diabetes;(2)Univariate logistic regression found the following factors related to GDM(P<0.05):age,pre-pregnancy BMI,previous GDM history,cesarean section history,registered permanent residence,working status,family resident population,whether it was willingness to pregnancies,frequency of brushing,frequency of cooking,ventilation conditions of living environment,frequency of medium-strength physical activity,peanut oil intake,frequency of soy milk intake,frequency of nut intake,frequency of night snoring;(3)In the training set,the logistic regression model included age,pre-pregnancy BMI,previous GDM history,family resident population,the frequency of moderate physical activity,frequency of soy milk intake,and frequency of nighttime snoring.The resulting model equation:Logit(PGDM)=Y=0.4679 x age+0.4140 ×pre-pregnancy BMI+1.5732×previous GDM history-0.2761 x family resident population+0.2275×frequency intake of soy milk-0.4427 x frequency of medium-intensity physical activity+0.2751 x frequency of snoring at night.Using Y as the GDM prediction score,the ROC curve was used to select the best score cutoff value of 1.217.The sensitivity of the model was 0.698 and the specificity was 0533.In the verification set,the sensitivity was calculated to be 0.660 and the specificity was 0.578.The verification effect was good.Conclusions:Age,pre-pregnancy BMI,and previous GDM history were associated with GDM.In the first trimester,increasing the number of permanent residents,carefully intaking soy milk,increasing the frequency of moderate-intensity physical activity,and reducing nighttime snoring may reduce the risk of GDM.The GDM risk model constructed with lifestyle-related factors as the main independent variable has certain predictive ability for pregnancy health education and GDM.Primary prevention provides evidence support.