Development of a Clinical Prediction Model for Parenteral Nutrition-Associated Cholestasis in Preterm Infants: A Retrospective Cohort Study

Why It Matters

Early identification of high‑risk infants enables clinicians to adjust parenteral nutrition, potentially reducing cholestasis‑related morbidity and associated healthcare costs.

Key Takeaways

• BW ≤1375 g triples PNAC risk.
• hsPDA increases PNAC odds by 3.4‑fold.
• Max daily amino‑acid ≥4.01 g/kg/day predicts cholestasis.
• Model AUC 0.851 shows strong discrimination.
• Early risk stratification can tailor PN advancement.

Pulse Analysis

Parenteral nutrition‑associated cholestasis remains a leading cause of liver injury in preterm infants, especially those requiring prolonged intravenous feeding. While individual risk factors such as low birth weight and ductal pathology have been noted, clinicians have lacked a unified tool to quantify risk. The growing body of neonatal data and advances in statistical modeling now allow for the creation of predictive algorithms that can be embedded into electronic health records, offering real‑time decision support for NICU teams.

The recent study leveraged a five‑year retrospective dataset from a tertiary NICU, applying rigorous collinearity checks before fitting a multivariable logistic regression. Birth weight ≤1,375 g, the presence of hemodynamically significant patent ductus arteriosus, and a maximum daily amino‑acid provision of ≥4.01 g/kg/day emerged as the strongest independent predictors. With an AUC of 0.851, the model outperforms many single‑parameter approaches, delivering both sensitivity and specificity that are clinically actionable. These thresholds translate directly into bedside criteria: infants below the weight cut‑off or receiving aggressive amino‑acid advancement can be flagged for closer liver function monitoring.

Implementing this risk model could reshape nutritional protocols across neonatal intensive care units. By tempering amino‑acid infusion rates for identified high‑risk infants, clinicians may curb the onset of cholestasis, shorten hospital stays, and lower long‑term hepatic complications. Moreover, the model provides a framework for prospective trials evaluating alternative feeding strategies, such as earlier lipid modulation or adjunctive probiotic therapy. As health systems prioritize value‑based care, tools that preempt costly complications like PNAC will become integral to neonatal quality improvement initiatives.