A meta-analysis of randomized, double-blind, placebo-controlled trials for the effects of garlic on serum lipid profiles.