An ultrasound-based nomogram for predicting early neonatal necrotizing enterocolitis

To develop and validate an ultrasound-based nomogram prediction model for the early diagnosis of neonatal necrotizing enterocolitis (NEC).

This is a retrospective cohort study. A total of 500 preterm infants admitted to the neonatal intensive care unit (NICU) of our hospital from January 2023 to June 2024 were enrolled. They were stratified by NEC status and randomly assigned to a development cohort and a validation cohort at a ratio of 7:3. According to whether NEC occurred, infants were further categorized into an NEC group and a non-NEC group. General clinical data, including gestational age and birth weight, were collected. Laboratory inflammatory markers included C-reactive protein (CRP), procalcitonin (PCT), and interleukin-6 (IL-6). Ultrasound parameters included bowel wall thickness, intestinal peristalsis, gastric antral area, gastric antral peristalsis, and bowel wall blood perfusion. Logistic regression analysis was performed in the development cohort to identify independent risk factors and construct a nomogram prediction model. Receiver operating characteristic (ROC) curves, calibration curves, and decision curve analysis were used to evaluate model discrimination, calibration, and clinical net benefit in the development cohort. The validation cohort was used to verify model performance.

Univariate analysis in the development cohort showed that gestational age, birth weight, CRP, PCT, IL-6, bowel wall thickness, intestinal peristalsis, gastric antral area, gastric antral peristalsis, and bowel wall blood perfusion differed significantly between groups (P < 0.05). Multivariable logistic regression further identified gestational age, CRP, PCT, IL-6, bowel wall thickness, intestinal peristalsis, bowel wall blood perfusion, gastric antral area, and gastric antral peristalsis as independent factors associated with early NEC (P < 0.05). On this basis, a nomogram prediction model was established. The areas under the ROC curve (AUCs) in the development and validation cohorts were 0.88 and 0.85, respectively, indicating good discriminative ability. The optimal risk cut-off of 0.18 was determined by Youden index solely from the development cohort. At this fixed threshold, sensitivity, specificity, PPV and NPV were 82.4%, 81.2%, 51.4%, and 95.0% for the development cohort; the corresponding performance in the validation cohort was 77.2%, 76.0%, 43.8%, and 92.4%. Goodness-of-fit testing of the calibration curves showed chi-square = 1.235 and P = 0.998 in the development cohort, and chi-square = 1.642 and P = 0.995 in the validation cohort, suggesting favorable agreement between predicted probabilities and observed risks. Decision curve analysis showed that when the risk threshold was greater than 0.05, the nomogram provided significant clinical net benefit.

The nomogram prediction model constructed from inflammatory and multiparametric ultrasound indicators demonstrates favorable predictive performance for the early diagnosis of neonatal NEC. It may provide clinicians with a noninvasive and repeatable auxiliary diagnostic tool, facilitating early intervention for NEC and improving prognosis.