Development and validation of an early prediction model for intestinal failure-associated liver disease after enterostomy for necrotizing enterocolitis
In brief
One in four infants developed liver disease after surgery for necrotizing enterocolitis
Among 388 infants who underwent enterostomy for necrotizing enterocolitis, 26.5% developed intestinal failure-associated liver disease. A six-factor model using routine perioperative data reached an area under the curve of 0.805 in a validation group, but it has not been tested externally; whether it can reliably guide monitoring and nutrition decisions remains unknown.
- Journal
- Pediatric research (Q1)
- Published
- 11 November 2026
- Study design
- Retrospective cohort
- Evidence level
- Level 3, Low (CEBM 3b)
- Authors
- Mengying Cui, Pengfei Chen, Jinfeng Hou, Yi Wang, Zhenhua Guo, Wei Liu, et al.
- PMID
- 43015330
- DOI
- 10.1038/s41390-026-05606-z
Why clinicians should know about it
- Picked for Pediatric Surgery (top studies of the week, 15 November 2026): NEC enterostomy cohort; predicts postoperative intestinal failure-associated liver disease
Abstract
OBJECTIVE: Infants undergoing enterostomy for necrotizing enterocolitis (NEC) are at high risk of intestinal failure-associated liver disease (IFALD). Early identification of high-risk infants remains challenging. This study aimed to develop and validate an early prediction model for IFALD using routinely available perioperative variables. STUDY DESIGN: This retrospective study included 388 infants who underwent enterostomy for NEC between November 2019 and November 2024. Clinical, nutritional, and laboratory variables were collected. Boruta and least absolute shrinkage and selection operator (LASSO) regression were used for feature selection. Logistic regression, LASSO regression, random forest, naïve Bayes, and partial least squares (PLS) models were compared using discrimination, classification metrics, calibration, and decision curve analysis. Shapley additive explanations (SHAP) were used to interpret the final model. RESULTS: Among 388 infants, 103 (26.5%) developed IFALD. The training cohort comprised 273 infants (73 IFALD and 200 non-IFALD), and the validation cohort comprised 115 infants (30 IFALD and 85 non-IFALD). Infants with IFALD had lower gestational age, lower birth weight, shorter residual bowel length, delayed enteral nutrition recovery, and longer parenteral nutrition (PN) dependence. The random forest model showed the best validation performance, with an area under the receiver operating characteristic curve of 0.805 (95% confidence interval, 0.719-0.891), sensitivity of 0.700, specificity of 0.765, and accuracy of 0.748. SHAP analysis identified birth weight, residual bowel length, age at admission, preoperative systemic immune-inflammation index, postoperative platelet-to-lymphocyte ratio, and central venous catheter infection as the most important predictors. CONCLUSIONS: Infants with enterostomy after NEC surgery often need prolonged PN, which increases the risk of IFALD. This study built an early prediction model using routine perioperative data. The random forest model showed good discrimination and high sensitivity, and may help clinicians identify infants who need closer liver monitoring, careful PN management, and infection prevention. IMPACT: 1. An early six-variable model predicts intestinal failure-associated liver disease after enterostomy for neonatal necrotizing enterocolitis. 2. The model uses birth weight, residual bowel length, age at admission, preoperative systemic immune-inflammation index, postoperative platelet-to-lymphocyte ratio, and early central venous catheter infection. 3. Random forest performed best among five evaluated models, with a validation area under the curve of 0.805. 4. This study adds a direct comparison of conventional and machine-learning approaches using routinely available perioperative variables in a comparatively large cohort. 5. After external validation, the model could support early risk stratification, closer liver monitoring, individualized parenteral nutrition management, and catheter-infection prevention.
Abstract as published, via PubMed.
For healthcare professionals. The summary is generated by AI from the published abstract, and the evidence level is assigned automatically from the study design on the Oxford CEBM hierarchy. Neither is medical advice. Read the full paper before changing practice.