We give each health claim one of four grades based on the human research only. Strong means several good-quality trials agree. Moderate means more than one small placebo-controlled trial points the same way. Weak means mixed, very small or uncontrolled studies. No human evidence means only cell or animal studies.
On this page
What do the four grades mean?
| Grade | What it means | What it would take to move up |
|---|---|---|
| Strong | Several good-quality human trials, run independently, agree | Already the top grade |
| Moderate | More than one small placebo-controlled trial points the same way, or one good-quality trial that has not been repeated | A large independent trial confirming the result |
| Weak | Human studies are mixed, very small, short, or have no proper placebo group | Consistent results from placebo-controlled trials |
| None | Only cell or animal studies; no human trial supports the claim | Any well-run human trial |
The dots beside each grade fill up with it: three for strong, two for moderate, one for weak and none for no human evidence.
What do we look at in each study?
- Was there a placebo group, and were people blinded? Without one, expectation alone can explain many results, especially for mood, sleep and energy.
- How many people took part? Small trials can find large effects by chance and miss real small ones.
- What was measured? A change on one of a dozen tests is weaker than a change on the main outcome the trial was designed for.
- How long did it last? Single-dose studies say little about taking a supplement for months.
- Who funded it? We note when the maker of the product paid for or ran the trial.
- Has it been repeated? One trial, however good, is a starting point.
Why do we ignore animal studies when grading?
Because most things that work in rats do not work in people, at least not at the doses people take. Animal studies tell us what is worth testing. They are not evidence that a supplement works in humans, so a claim supported only by them gets the “no human evidence” grade, however many animal studies there are.
First published 3 November 2026. Spotted an error? Tell us and we'll correct it. Read our editorial policy.