Internally validated model identifies factors associated with parenteral nutrition-associated cholestasis in preterm infants

A retrospective cohort study of 329 preterm infants found that lower birth weight, treated patent ductus arteriosus, higher maximum amino acid dosage and low enteral-feeding volume by day 14 were independently associated with parenteral nutrition-associated cholestasis, according to the published abstract. The model built from these factors had an area under the receiver operating characteristic curve of 0.859 (95% CI 0.806–0.912; p<0.001), with 85.4% sensitivity and 77.1% specificity at its reported optimal cutoff.

The researchers reviewed records from preterm infants admitted to one tertiary neonatal intensive care unit between July 2020 and June 2025. Eligible infants were born at less than 37 weeks’ gestation, weighed less than 2,000 g at birth and required parenteral nutrition for at least 14 consecutive days. Infants diagnosed with necrotizing enterocolitis or who underwent major gastrointestinal surgery were excluded. Of 389 neonates screened, 329 met the inclusion criteria, and 41 (12.5%) developed cholestasis. The investigators analyzed demographic, clinical and nutritional variables, used collinearity diagnostics, and applied multivariable logistic regression to identify variables independently associated with the outcome.

The four variables retained as predictors in the final model were birth weight, treated patent ductus arteriosus, maximum amino acid dosage and enteral-feeding volume on day 14. At the individual thresholds reported in the abstract, birth weight of 1,375 g or less had 69.4% sensitivity and 82.9% specificity, while a maximum amino acid dosage of 4.01 g/kg/day or more had 48.8% sensitivity and 83.0% specificity. The model was evaluated with bootstrap internal validation, subgroup analysis and decision curve analysis. The abstract reports that decision curve analysis found greater net benefit than treat-all and treat-none strategies across threshold probabilities of 0% to 45%, but it does not provide the subgroup results.

These findings describe adjusted associations and model performance rather than evidence that the predictors cause cholestasis or that the model improves clinical outcomes. The study was retrospective, conducted in a single tertiary neonatal intensive care unit and evaluated through internal rather than external validation. It also excluded infants with necrotizing enterocolitis or major gastrointestinal surgery, limiting the population to which the reported performance may apply. With 41 outcomes in the final cohort and no external validation reported, the abstract does not establish performance in other hospitals, larger or more diverse populations, or excluded clinical groups. The available evidence is limited to the abstract, which does not present the full model equation, calibration results, subgroup findings or detailed limitations.

Subject of Research: Development and internal validation of a clinical prediction model for parenteral nutrition-associated cholestasis in preterm infants

Article Title: Development of a clinical prediction model for parenteral nutrition-associated cholestasis in preterm infants: a retrospective cohort study

Article References: Chen, Y.-S., & Chen, S.-M. (2026). Development of a clinical prediction model for parenteral nutrition-associated cholestasis in preterm infants: a retrospective cohort study. BMC Pediatrics. https://doi.org/10.1186/s12887-026-07855-7

Image Credits: AI Generated

DOI: 10.1186/s12887-026-07855-7

Keywords: preterm infants, parenteral nutrition, cholestasis, clinical prediction model, retrospective cohort, neonatal intensive care, birth weight, patent ductus arteriosus, amino acids, enteral feeding, Development, clinical