The Risk Factors And Risk Assessment Model Of Gestational Diabetes Mellitus

Object1.Analyze the risk factors of GDM by conducting a cohort study for pregnant women from Zhangjiagang, as our objectives.2.Establish a GDM risk assessment model and validate the model effects to prevent and cure GDM in a better way.MethodsThe study population comprised all pregnant women(n=2927) who established inspection card and did regular prenatal check at 13 hospitals in Zhangjiagang. The questionnaire consists of 3 parts :basic information of the objectives, situation before pregnancy,situation early in pregnancy.Using Epidata 3.1 to establish a database and entry the data, SPSS19.0 to conduct statistical analysis, P <0.05 indicates significant difference. Using multiple Logistic regression analysis to find risk factors of GDM then constructed a model to estimate the probability of development of GDM. The statistical methods we used is univariable and multivariable analyses, Receiver operating characteristic curve(ROC) and Hosmer-lemeshow goodness-of-fit test were used to evaluate the model. We use Youden’s index to determine the threshold value, then randomly selected 10% of the population to test and evaluate the model results.Results1. General information2927 questionnaires were included in this study, eliminate the serious lack of information and suspect investigations, we got 2769 effective questionnaires, the effective rate is 94.6%, after ruling out 10 diabetic pregnancy(FPG≥7mmol/L), finally, we got 2759 questionnaires.Comparing the two groups, it has significant difference between age and BMI before pregnancy, no significant difference in other characteristics.2. Risk factors of GDMThe univariate analysis showed that age, overweight before pregnancy, obesity before pregnancy, diastolic blood pressure≥80mm Hg, PCOS, first-degree relatives history of diabetes, gravida, para, the history of macrosomia, abortion times, GDM history, menstrual cycle≥35, FPG were related with GDM; then we conduct multivariate Logistic regression analysis with the significant factors, the results showed that age, overweight before pregnancy, obesity before pregnancy, diastolic blood pressure≥80mm Hg, first-degree relatives history of diabetes, GDM history, FPG were risk factors of GDM; Protective factors were pregnancy underweight, high height.3. Logistic regressing modelSeven factors were added into the regression model. The model of the risk of the GDM was:Logit P=-4.92+0.044X1-0.028X2-0.309X51+0.362X52+0.652X53+0.326X7+0.501X111+2.894X20 +1.300X29S=-4.92+0.044*age-0.028*height-0.309*pregnancy underweigt+0.362* pregnancy overweight +0.652* obesity +0.326* diastolic blood pressure≥80+0.501* First-degree relatives history of diabetes +2.894* GDM history +1.300* FPGThe area under the ROC curve was 0.722, Hosmer-Lemeshow goodness-of-fit test P was 0.71. When the cumulative score reached-1.79, Youden’s index became the largest of 0.33. We use SPSS19.0 extract 10% random sample to verify the model, the sensitivity was 73.1%, and the specificity was 54%.Conclusions1. Age, overweight before pregnancy, obesity before pregnancy, diastolic blood pressure≥80mm Hg, first-degree relatives history of diabetes, GDM history, FPG were risk factors of GDM; Protective factors were pregnancy underweight, high height.2. A simple and efficient regression model was built.