Establishment and validation of a prediction model for small vulnerable newborns: a retrospective study.

Background: The concept of small vulnerable newborns has been proposed, including preterm birth, low birth weight, and small for gestational age, leading causes of perinatal mortality. We aimed to identify high-risk factors for small vulnerable newborns and develop a predictive model through a retrospective analysis. Methods: We collected clinical data from pregnant women who met inclusion criteria between January 2015 and December 2023 and divided them into training and validation cohorts. We used univariate analysis and mean decreases in the Gini index to screen for potential risk factors. We applied the least absolute shrinkage and selection operator regression to select final predictors and construct a nomogram. We assessed model performance using receiver operating characteristic curves, calibration curves, and clinical decision analysis, with internal validation via 10-fold cross-validation and temporal internal validation. Results: Among 129 554 women, 13 801 (10.66%) had small vulnerable newborn, with the incidence increasing from 2015 (10.15%) to 2023 (11.61%). Key risk factors included multiple pregnancies (odds ratio (OR) = 37.2), pre-pregnancy body mass index (BMI) of <18.5 (OR = 8.61) and ≥25 kg/m Conclusions: The incidence of small, vulnerable newborns rose notably during 2021-2023. The developed model, incorporating age, pre-pregnancy BMI, multiple pregnancies, hypertensive disorders of pregnancy, and placental disorders, is designed to be applied in the third trimester and enables risk identification, facilitating targeted interventions to reduce neonatal mortality and complications. Registration: Chinese Clinical Trial Registry, ChiCTR2400093923.