Development and internal validation of a nomogram for predicting adverse pregnancy outcomes in women with early-onset preeclampsia

Background Early-onset preeclampsia (EOPE) is a severe obstetric complication associated with a high risk of adverse maternal and fetal outcomes. Early identification of patients at high risk of adverse pregnancy outcomes may facilitate timely clinical intervention and improve pregnancy management. This study aimed to develop and internally validate a clinical prediction model for adverse pregnancy outcomes in women with EOPE. Methods This retrospective observational study included 425 women diagnosed with EOPE who were admitted to a tertiary teaching hospital in Southwest China between January 2018 and December 2025. The composite adverse pregnancy outcome was defined as at least one predefined maternal or fetal/neonatal adverse event occurring after admission, excluding baseline pregnancy-related complications used as candidate predictors. Patients were randomly divided into a training cohort ( n = 298) and a validation cohort ( n = 127). Candidate predictors were restricted to variables measured or diagnosed at admission or within the first 24 h after admission, before the occurrence of the composite adverse pregnancy outcome. Candidate predictors were screened using least absolute shrinkage and selection operator (LASSO) regression and further evaluated using multivariable logistic regression. Independent predictors were used to construct a nomogram-based prediction model. Model performance was evaluated in terms of discrimination, calibration, and clinical utility using receiver operating characteristic (ROC) curves, calibration curves, and decision curve analysis (DCA), with internal validation performed in the split-sample validation cohort. Results Seven predictors were independently associated with the composite adverse pregnancy outcome and were incorporated into the nomogram: gestational age at admission, systolic blood pressure, urine protein score, platelet count, aspartate aminotransferase (AST) level, fetal growth restriction, and hemolysis, elevated liver enzymes, and low platelet syndrome (HELLP) syndrome. In clinically interpretable units, lower gestational age at admission, each 10-mmHg increase in systolic blood pressure, each 1-grade increase in urine protein score, each 50 × 10 9 /L decrease in platelet count, each 10-U/L increase in AST, fetal growth restriction, and HELLP syndrome were associated with increased risk. The model showed good discrimination, with an area under the curve (AUC) of 0.859 (95% CI: 0.754–0.897) in the split-sample validation cohort. Calibration curves showed acceptable agreement between predicted and observed risks, and decision curve analysis suggested potential clinical net benefit across clinically relevant threshold probabilities. Conclusion We developed and internally validated a nomogram-based prediction model for adverse pregnancy outcomes in women already diagnosed with EOPE. The model is intended for admission-based risk stratification after EOPE diagnosis, rather than for early prenatal prediction, population screening, or prediction of the initial occurrence of EOPE. The model demonstrated good predictive performance in this single-center cohort and may support clinical risk stratification after external validation. The proposed risk categories are exploratory and should not be used as definitive clinical decision cutoffs without external validation. Independent multicenter validation is required before clinical implementation.