[stefd]

> They use statistical methods to eliminate the confounding variables.

Some of those anyway. You can only account for confounders that you know about a priori.

And given they used Food Frequency Questionnaires, it's mostly wasted efforts imho.