Development and validation of a nomogram model for predicting the occurrence of necrotizing enterocolitis in premature infants with late-onset sepsis.

backgroundNecrotizing enterocolitis (NEC), a devastating gastrointestinal disease in preterm infants, is strongly linked to sepsis, and 34-57% of NEC cases develop post-sepsis. However, the risk factors for sepsis-associated NEC are still unclear. Therefore, this study aimed to identify predictive factors associated with the occurrence of NEC in premature infants with late-onset sepsis (LOS) and establish a nomogram for early NEC prediction.

methodsThis single-centre, retrospective cohort study included preterm infants who were diagnosed with LOS and admitted to a tertiary neonatal intensive care unit in China. Patients were classified into either the NEC group (n = 65) or the non-NEC group (n = 127) according to whether they developed NEC after LOS. Least absolute shrinkage and selection operator (LASSO) regression was employed to identify potential predictors from candidate variables, followed by logistic regression to determine independent risk factors. R software was used to establish the nomogram prediction model. Internal validation was performed by bootstrapping 1,000 resamples to assess model stability. Discrimination ability was quantified using the area under the receiver operating characteristic (ROC) curve, whereas the calibration curve and Hosmer-Lemeshow test were used to evaluate the agreement between the predicted and observed probabilities. Clinical utility was further examined through decision curve analysis (DCA).

resultsOne hundred ninety-two preterm infants with LOS were admitted to the hospital, 65 of whom developed NEC. LASSO-Logistic regression analysis revealed three independent risk factors for NEC: red blood cell transfusion (OR = 2.55, 95% CI 1.06-6.13, P = 0.036), an elevated haemoglobin difference (ΔHb) (OR = 1.16, 95% CI 1.10-1.23, P < 0.001), and increased mean platelet volume (MPV) (OR = 3.40, 95% CI 2.15-5.39, P < 0.001). The nomogram prediction model incorporating these variables demonstrated strong discriminative performance, with an area under the ROC curve (AUC) of 0.860. Internal validation by bootstrapping revealed a concordance index (C-index) of 0.862, indicating robust predictive accuracy. The calibration curve and Hosmer-Lemeshow test showed close agreement between the predicted and observed NEC probabilities (P > 0.05), whereas the DCA confirmed the model's practical utility for clinical decision-making. Finally, we developed a freely accessible web-based calculator ( http://106.14.106.176:8080/ ), which dynamically generates individualized NEC risk estimates on the basis of the nomogram, to facilitate clinical implementation.

conclusionsRed blood cell transfusion, the MPV and ΔHb were found to be independent predictors of NEC in premature infants with LOS. A nomogram incorporating these factors demonstrated good discriminative performance. However, the clinical utility of the nomogram requires confirmation through external validation and prospective studies.