Risk factors and prediction model for necrotizing enterocolitis in preterm infants with gestational age ≤ 32 weeks: a...

To develop a reference tool for necrotizing enterocolitis(NEC) prevention and treatment by constructing a predictive model for NEC risk in preterm infants (≤32 weeks' gestation) in Guangxi, China.

The clinical data of 497 preterm infants with gestational age ≤32 weeks managed at four neonatal care centers in Guangxi between January 2,019 and December 2021 were retrospectively reviewed. The cohort was randomly divided into a training set (for model development) and a test set (for model validation) in an 8:2 ratio. Within the training set, non-NEC infants were randomly selected to match NEC infants at a 1:1 ratio for comparative analysis. Univariate analysis was first performed to compare clinical indicators between the NEC and non-NEC groups and to identify potential predictors. Subsequently, independent risk factors were determined using binary logistic regression analysis, and a nomogram for predicting NEC risk was constructed using R software. Model performance was evaluated using the area under the receiver operating characteristic (ROC) curve, the Hosmer-Lemeshow goodness-of-fit test, and calibration curves.

The incidence of NEC among preterm infants with gestational age ≤32 weeks was 12.27% (61/497). Univariate analysis revealed significant differences between the two groups in gestational age, birth weight, 5-minute Apgar score, presence of neonatal respiratory distress syndrome (NRDS), intrauterine growth restriction (IUGR), sepsis, fungal infection, and the use of both invasive and non-invasive ventilation (P < 0.05). Multivariate logistic regression analysis identified IUGR (OR = 30.586), NRDS (OR = 22.955), sepsis (OR = 36.495), and invasive ventilator use (OR = 1.295) as independent risk factors for NEC (P < 0.05). A higher 5-minute Apgar score was identified as a protective factor (P < 0.05), indicating a decreased risk of NEC with increasing scores. Based on these factors, a nomogram prediction model was constructed using R software. The model demonstrated excellent discriminatory ability, with an area under the ROC curve (AUC) of 0.917 in the training set and 0.906 in the test set. The Hosmer-Lemeshow goodness-of-fit test for the test set (χ ² = 3.761, P = 0.807) indicated no significant difference between predicted and observed probabilities, suggesting good model calibration. The calibration curve approaches the 45-degree line, demonstrating good consistency between the model's predicted values and actual values.

The predictive model developed in this study demonstrates good discriminatory power and calibration, and is effective in assessing the risk of NEC in preterm infants with a gestational age of ≤32 weeks in the Guangxi region. It provides valuable guidance for the early prevention and treatment of NEC in this population.