Early versus later initiation of parenteral nutrition for very preterm infants: a propensity score-matched observational study.

Residual confounding and survival bias cannot be excluded and justify the need for a randomised controlled trial powered to detect differences in important functional outcomes.