The Evidence Hierarchy Explained

A meta-analysis combines results from multiple randomized controlled trials. It's the strongest evidence because it pools data from many studies, increasing statistical power and reducing the chance that one flawed study misleads. When a meta-analysis shows something works, pay attention.

Participants randomly assigned to treatment or placebo. Neither knows which they got (double-blind). This design controls for placebo effect and selection bias. A well-designed RCT in humans is solid evidence. Multiple RCTs showing the same thing is very solid.

Cohort and case-control studies observe what happens without intervention. They can show associations but can't prove causation. "People who take X have better Y" doesn't mean X causes better Y. Maybe healthier people take X. Useful for generating hypotheses, not confirming them.

Most supplement claims start with rat studies. Problem: rats aren't humans. Doses are often massively higher relative to body weight. Physiology differs. An effect in rats is interesting but doesn't guarantee human relevance. Many compounds that worked in animals failed in humans.

Studies in isolated cells can show mechanisms but don't account for digestion, absorption, metabolism, or systemic effects. Killing cancer cells in a dish is easy. Many things do it, including bleach. That doesn't make them cancer treatments. In vitro is mechanistic exploration, not clinical evidence.

Watch for: small sample sizes (under 30 participants), short durations (under 4 weeks for chronic effects), industry funding without independent replication, surrogate endpoints instead of real outcomes, cherry-picked outcomes from larger studies. These don't invalidate findings but should increase skepticism.