Establishment of a risk prediction model for large-for-gestational-age infants among Chinese women with gestational diabetes mellitus

Abstract

Infants classified as large for gestational age (LGA) are often born to mothers with gestational diabetes mellitus (GDM). This study aimed to develop a prediction model to estimate the risk of LGA infants with GDM mothers. This retrospective study included 791 singletons of mothers with GDM delivered at our hospital between June 2018 and May 2020. Data was collected from the hospital’s electronic information system. According to whether LGA occurred, participants were divided into two groups to analyze the related factors affecting LGA. Pregnant women were randomly divided into two groups in a 7:3 ratios to generate and validate the model. To optimize the selection of variables, the Least Absolute Shrinkage and Selection Operator (LASSO) regression analysis was employed. A predictive model was subsequently constructed using multivariable logistic regression, incorporating predictors identified through LASSO regression. A nomogram was devised based on the selected variables for visual representation. The predictive model’s performance was evaluated using the area under the receiver operating characteristic (ROC) curve (AUC) to assess discrimination, and calibration plots to assess calibration accuracy. Furthermore, decision curve analysis (DCA) was utilized to evaluate the clinical applicability of the models. Logistic regression analysis identified prepregnancy BMI, gestational weight gain (GWG), the 0-hour oral glucose tolerance test (OGTT0h), and parity as independent risk factors for LGA infants. The model demonstrated an area under the curve (AUC) of 0.777 in the training set and 0.744 in the validation set. The DCA illustrated that the nomogram exhibited superior net benefit within the validation cohort when the threshold probabilities were situated between 5% and 55%. Prepregnancy BMI, GWG, OGTT0h, and parity into the risk nomogram increased its usefulness for predicting LGA risk in patients with GDM.