Statistical analysis
Initially, we utilized univariate analysis to compare maternal characteristics, glucose values during pregnancy, and T2DM outcome stratified according to FPG levels during 1st trimester. FPG levels were assessed in categorical increments of 10mg/dl (under 80, 80-90, 90-100, 100-110, and above 110 mg/dl). Maternal age and BMI were evaluated as continuous and as categorical variables. Advanced maternal age was evaluated with a cut-off of 35 and 40 years, and obesity was evaluated as BMI ≥30 kg/m2. Other glucose levels, gestational age at delivery, and time to follow-up were treated as continuous variables, while hypertension, GDM, neonatal sex, and T2DM were treated as categorical variables. Categorical variables were compared using χ 2 tests and the Mann-Whitney U test was used to test differences for continuous variables. All the tests were 2-tailed, and P < 0.05 was considered statistically significant.
Subsequently, we aimed to identify the optimal FPG threshold during first trimester for predicting the risk of T2DM by employing receiver operating characteristic (ROC) statistical analysis. The Youden index, which maximizes the sum of sensitivity and specificity, determined the best cutoff value. Furthermore, we compared the area under the curve (AUC) for FPG, GDM, and the Youden index in predicting T2DM.
Lastly, given the variations in the follow-up duration for women in our cohort, we calculated Kaplan-Meier hazard curves, which account for these discrepancies. We also applied Cox proportional hazard models to estimate the FPG level and GDM adjusted hazard ratios (aHRs) with 95% confidence intervals (CI) for incident T2DM over the study period. The selection of variables for the model was based on univariate association with T2DM. Notably, we conducted a stratified analysis based on obesity status (BMI ≥ 30 kg/m2), recognizing its potential impact on diabetes risk and clinical approaches for prevention.