Development and Validation of a Multidimensional Indicator-Based Risk Prediction Model for Gestational Diabetes Mellitus: A Nested Case-Control Study

Read the full article See related articles

Listed in

This article is not in any list yet, why not save it to one of your lists.
Log in to save this article

Abstract

Background: Gestational diabetes mellitus (GDM) could contribute to significant health risks in both mothers and their offspring. Therefore, this study aims to construct a prediction model to identify women at elevated risk for GDM in early pregnancy. Methods: This study was a nested case-control study. 346 participants were randomly allocated to the training set (n=242) and the validation set (n=104) at a ratio of 7:3. Sociodemo-graphic characteristics, clinical indicators, and lifestyle behaviors of all participants were obtained at 8–13+6 weeks of gestation. GDM was confirmed through the 75-g oral glucose tolerance test (OGTT). The least absolute shrinkage and selection operator (LASSO) regression was applied to select the most significant factors among candidate variables. We further established a GDM risk prediction model based on the risk factors chosen by the LASSO. The model's calibration, discrimination, and clinical use were assessed using the calibration analysis, area under the receiver operating characteristic (ROC) curve and decision curve analysis (DCA). Finally, we presented the model with a nomogram. Results: In the study, the prevalence of GDM in the training and validation sets were 24.8% and 26.0%, respectively ( P =0.93). In the training set, we developed a simple GDM risk prediction model by using family history of diabetes, pre-pregnancy body mass index (BMI), progesterone, aspartate transaminase (AST), activated partial thromboplastin time (APTT), and triglyceride to high-density lipoprotein cholesterol (TG/HDL-C). Among them, family history of diabetes, higher pre-pregnancy BMI, progesterone, AST, and TG/HDL-C levels were associated with increased GDM risk, while higher APTT level was associated with decreased GDM risk. The calibration curve indicated satisfactory accuracy. The ROC curve demonstrated excellent discrimination, with the area under the curve (AUC) of 0.85 (95% confidence interval [CI], 0.80-0.91) and 0.73 (95%CI, 0.62-0.83) for the training and validation set, respectively. The DCA curve demonstrated high net benefit. Furthermore, internal validation with excellent performance demonstrated the generalizability of the model. Conclusions: The present study developed a model with excellent performance for predicting GDM. Furthermore, a nomogram was constructed to visualize the model. Therefore, this model can serve as an effective GDM prediction tool.

Article activity feed