"the less sodium that study subjects excreted in their urine - an excellent measure of prior consumption"

These two statements seem mutually exclusive. If the amount of sodium a subject excretes is an excellent measure of prior consumption (ie, more out is demonstrably linked with more in, and less out = less in), then your hypothesis (less out may mean more retained, and therefore output can't be linked to input) cannot be true.

Of course, it could also be the "excellent measure" which is incorrect - my point is simply that these two are contradictory.