Parenteral nutrition-associated cholestasis and postnatal growth trajectories in preterm infants: A longitudinal...

To evaluate the association between parenteral nutrition-associated cholestasis and postnatal weight and linear growth trajectories to 12 months corrected age, and to assess whether cholestasis mediates the relationship between parenteral nutrition duration and 12-month linear growth in preterm infants.

Retrospective cohort study of a standardized neonatal nutrition and follow-up database including infants born at <32 weeks' gestation. Cholestasis was defined as peak direct bilirubin ≥34.2 μmol/L. Longitudinal weight and length z-scores were analysed using generalized estimating equations with exchangeable correlation, including time, cholestasis, and time × cholestasis interaction, adjusted for gestational age, cumulative parenteral nutrition duration, length of stay, and surgical necrotizing enterocolitis. Poor linear growth at 12 months (length z-score < -2) was evaluated using multivariate logistic regression adjusted for clinically relevant covariates. Mediation analysis used a product-of-coefficients approach with bootstrap confidence intervals.

Among 1249 infants, 107 (8.6%) developed cholestasis. At 12 months corrected age, infants with cholestasis had lower weight (-1.36 vs -0.23, p < 0.001) and length z-scores (-1.95 vs -0.70, p < 0.001). In adjusted models, no difference in birth weight z-score was observed (β = 0.122, p = 0.350), whereas a small adjusted difference in birth length z-score was present (β = 0.339, p = 0.048). The principal adjusted differences in both weight and length trajectories were observed from hospital discharge and persisted through 12 months corrected age (weight β = -0.556, p = 0.007; length β = -0.879, p < 0.001). Cholestasis was not independently associated with poor linear growth (adjusted OR = 1.12, 95% CI 0.54-2.32; p = 0.771). Lower birth weight z-score (adjusted OR = 0.47, 95% CI 0.35-0.65; p < 0.001), male sex (adjusted OR = 1.99, 95% CI 1.21-3.30; p = 0.007), and longer hospital stay (adjusted OR = 1.01 per day, 95% CI 1.00-1.02; p = 0.006) were independently associated with the outcome. Mediation analysis demonstrated no significant indirect effect via cholestasis (adjusted β = -0.013, 95% CI -0.044 to 0.014).

Parenteral nutrition-associated cholestasis was associated with poorer post-discharge growth, with a greater effect on linear growth than weight, but was not independently associated with poor linear growth after adjustment and did not mediate the relationship between cumulative parenteral nutrition duration and later growth.