Searching for defining adequate protein intake in ICU patients: is intention to treat analysis the best?

Many clinical trials have failed to demonstrate the superiority of one nutritional protocol over another. This raises questions about the assessment tools used. In particular, intention-to-treat data analysis can mask individual truths, especially in cases of protocol noncompliance. If patients do not always achieve their nutritional goals, an intention-to-treat analysis tends to conclude that the two protocols studied are no different.

This is particularly the case in two recent articles published in The Lancet, where patients failed to achieve the predefined protein target in intensive care patients due to protocol deviation.

I argue here for further analysis, including per-protocol analysis, to avoid the false conclusion that failure to show a difference means there is no difference.