What a rating covers
Every rating starts with a narrow question. We state the outcome, the people studied, the modality or intervention, the comparison and the follow-up period when the sources provide them. A finding about sleep quality in adults with insomnia, for example, does not automatically apply to healthy adults, disease prevention or longer life.
We also separate a modality from a product. Studies of red light, cold exposure, a supplement ingredient or a testing method do not prove that every device, dose, formulation or service in that category delivers the same result. Product pages still need their own documented specifications, safety information, costs and trade-offs.
What the rating labels mean
The labels describe the support available for the exact claim when we last reviewed the cited sources. They are not medical recommendations, product scores or forecasts of an individual’s result.
| Label | What it means |
|---|---|
| Strong | Several well-conducted studies, often including systematic reviews or consistent trials, support the stated outcome in a similar population and modality. Important limits, harms and uncertainty still belong on the page. |
| Moderate | Useful human evidence supports the stated outcome, but study size, follow-up, consistency, bias risk, population fit or intervention differences leave a meaningful uncertainty. |
| Mixed | Relevant studies point in different directions, or their results do not line up well enough to support one clear conclusion for the stated claim. |
| Weak | The evidence is preliminary, indirect, small, short, uncontrolled or otherwise too limited to support much confidence in the stated claim. |
| Not supported | The sources reviewed do not support the stated claim in the stated population and modality. This label is not a finding that the intervention is proven ineffective. |
How we apply the rubric
We give more weight to human research that measures outcomes people care about, uses a relevant comparison and follows people long enough to answer the claim. We look for systematic reviews, randomized trials, well-described observational studies, guidelines and regulator information when they fit the question.
We reduce confidence when a result comes from a lab marker, a short study, a small or narrow sample, an uncontrolled design, a different dose or device, selective reporting, conflicts that may affect interpretation, or results that other studies do not reproduce. A statistically significant result can still be too small, too indirect or too uncertain to matter to a buyer.
This framework guides new and revised evidence assessments. It does not mean every existing page has already been re-rated under this framework. Each page’s sources and update date show what was reviewed there.
Source dates and limits
We cite the sources used on each evidence page and show the page update date. Research can change, and a source date is not a guarantee that every later study has been captured. We do not treat a publication date, a press release or a marketing claim as a substitute for the underlying evidence.
The ratings do not diagnose, treat or prevent disease. They do not replace a clinician who can weigh your history, medications, symptoms and risk. They also do not settle questions about product quality, manufacturing consistency or personal suitability.
Corrections and updates
If a source is wrong, outdated or missing material context, send the page and source through the contact page. We compare the report with the cited material and newer relevant sources. If a correction changes the claim, rating or conclusion, we update the page and its date.
Our corrections policy explains how reports are checked. Our editorial policy explains who writes the site, how commercial disclosures work and why no rating should be read as medical advice.
How this was written. Last updated . It is general information, not medical advice. Talk to a clinician before acting on anything here if you have a health condition or take medication.
Our editorial policy →