The first question to ask is whether the study is observational in nature or a randomized controlled trial, possibly double-blind & placebo-controlled. The former is only good for developing hypotheses or educated guesses, although there are times when the best we'll ever get is multiple observational studies, all pointing towards the same result. Ideally, however, we'd like to base our recommendations upon the latter which gives us evidence of cause & effect. This is the gold standard of proof.
And if you want 24 karat proof, look for reduction in all-cause mortality. After all, we're all going to die from something, right? It's kind of like playing Whac-a-Mole (apologies to PETA ) whereby decreasing one disease-specific mortality (whacking a mole) allows an increase in some other disease process (another mole pops up). What you really want is a decrease in all-cause mortality (where no moles are left standing). Stay tuned for tomorrow's post . . .