Development and Internal Validation of a Mid-Pregnancy Nomogram to Predict Large-for-Gestational-Age Infants in Women With Hyperglycemia in Pregnancy

Background: Women with hyperglycemia in pregnancy (HIP) have an increased risk of delivering large-for-gestational-age (LGA) infants, a condition associated with both short- and long-term adverse outcomes for mothers and offspring. The identification of women with HIP at high risk of LGA using routinely available clinical indicators could help optimize antenatal management. Methods: This retrospective cohort study included 675 women diagnosed with HIP who delivered at Anhui Maternal and Child Health Hospital in Hefei, China, between January 2017 and December 2019. Maternal demographic characteristics and biochemical parameters measured at 25–29+6 weeks of gestation were extracted from medical records. LGA was defined as birthweight above the 90th percentile for gestational age and sex, according to the INTERGROWTH-21st international newborn standards. Candidate predictors were selected based on clinical relevance and correlation analysis. A multivariable logistic regression model with stepwise selection was used to develop a prediction model for LGA, followed by the construction of a nomogram. Internal validation was performed using 1000 bootstrap resamples. Discrimination, calibration, and clinical utility were assessed using the area under the receiver operating characteristic curve (AUC), Brier score, Hosmer-Lemeshow goodness-of-fit test, calibration plots, and decision curve analysis (DCA). Results: Among the 675 women with HIP, 113 (16.7%) delivered LGA infants. The final multivariable model retained 8 predictors: maternal height, serum albumin, lactate dehydrogenase, triglycerides, total cholesterol, blood urea nitrogen, homocysteine, and fasting plasma glucose. The model showed acceptable discrimination, with an AUC of 0.7333 (95% confidence interval [CI]: 0.6856–0.7810) in the original dataset and an optimism-corrected AUC of 0.7132 after 1000 bootstrap resamples. Calibration was acceptable (Brier score 0.1260; Hosmer-Lemeshow p = 0.0644), and DCA indicated a net clinical benefit of using the model across a wide range of threshold probabilities. Conclusions: We developed and internally validated a simple mid-pregnancy prediction model to estimate the risk of LGA in infants born to women with HIP. This nomogram, based on routinely available clinical and biochemical indicators, may help clinicians identify high-risk women in mid-pregnancy, thus facilitating individualized antenatal management. However, external validation in independent populations is required before broader clinical implementation.