Objective:

To develop and internally validate a clinical risk prediction model for estimating the individualized risk of late-pregnancy hypertriglyceridemia (HTG) in women at the time of gestational diabetes mellitus (GDM) diagnosis.

Approach:

• Study Design: Single-center retrospective cohort study including 587 women with GDM.
• Data Collection: Predictor variables included demographic characteristics, glycemic markers from the 75g oral glucose tolerance test (OGTT), and first-trimester lipid profiles.
• Outcome Definition: Late-pregnancy HTG defined as triglyceride level ≥ 2.3 mmol/L.
• Model Development: Used LASSO logistic regression for variable selection and multivariable logistic regression for model building.
• Performance Assessment: Model performance assessed using discrimination, calibration, Brier score, and decision curve analysis.

Key Findings:

• Incidence of late-pregnancy HTG was 32.7% (192/587).
• Final model included five predictors: pre-gravid BMI, fasting plasma glucose at diagnosis, 1-hour post-load glucose at diagnosis, first-trimester triglycerides, and first-trimester HDL-C.
• Model demonstrated good discrimination with an AUC of 0.816 (95% CI: 0.754–0.878) in the internal test set.

Interpretation:

Limitations:

• Study conducted at a single center, limiting generalizability.
• Requires external validation in independent multicenter cohorts before clinical implementation.

Conclusion:

The study developed and validated a prediction model for late-pregnancy HTG in women with GDM using five clinical parameters.