Design: Secondary analysis of the PEPaNIC randomised trial
Population: 1,328 critically ill children (newborn–17 years) not on renal replacement
Primary outcome: New infection and duration of PICU dependency, by urea-to-creatinine ratio
Effect: Higher first-week ratio: infection aOR 1.75 (1.35–2.27); 90-day mortality aOR 1.43 (1.03–1.99)

The PEPaNIC trial earlier showed that starting parenteral nutrition early in critically ill children — rather than waiting a week while enteral feeding is built up — increased infections and delayed recovery. This secondary analysis asks why.

In 1,328 children not on renal replacement, early parenteral nutrition raised urea and the urea-to-creatinine ratio over the first 11 days. A higher first-week urea-to-creatinine ratio independently tracked with more new infections (adjusted odds ratio 1.75 per 30 units), slower discharge from PICU and hospital, and higher 90-day mortality (adjusted odds ratio 1.43), and statistically accounted for part of the harm.

The reading is that the amino-acid load of early parenteral nutrition can exceed what a sick child can metabolise, and that urea may flag it. The practical point stands with or without the mechanism: do not rush to supplement feeding with parenteral nutrition in the first week of paediatric critical illness, and watch a climbing urea as a possible warning.

Early parenteral nutrition raised urea and the urea-to-creatinine ratio in critically ill children.
A higher first-week ratio tracked with more infections, slower recovery and higher 90-day mortality.
The rise statistically explained part of the harm from early parenteral nutrition.
Avoid rushing parenteral nutrition in the first week of paediatric critical illness.

Why it matters

It gives a mechanism, and a cheap marker, for the established harm of early parenteral feeding in critically ill children.

The statistics, in plain English

These are associations within a randomised trial: the urea rise correlates with worse outcomes and partly explains the known harm, but urea as a monitoring tool still needs testing prospectively.