These are the rules we hold ourselves to. They are published so you can hold us to them as well.
Sourcing
Every factual claim on this site traces to a named, checkable source. In practice that means peer-reviewed journal articles, systematic reviews, meta-analyses, clinical trial registries, and guidance from national health bodies.
We do not source claims to press releases, university media offices, supplement manufacturers, podcast appearances, social media threads, or other publications’ summaries of research. Where a source is behind a paywall we say so rather than pretending otherwise.
How we rank evidence
We tell you what kind of evidence a conclusion rests on, because the label changes what it is worth:
- Systematic reviews and meta-analyses of randomised trials carry the most weight, provided the underlying trials are sound.
- Randomised controlled trials in humans are strong evidence for the specific population and duration studied, and weaker evidence for anyone else.
- Observational and cohort studies can establish association and generate hypotheses. They cannot establish cause on their own, and we do not let them.
- Mechanistic and animal research explains how something might work. It does not establish that it does work in people.
A finding that has only ever been demonstrated in mice, in cell culture, or in a twelve-person pilot study will be described that way, no matter how striking the result.
Contested science
Where the evidence genuinely conflicts, we present the conflict rather than picking the tidier answer. That means naming the strongest case on each side, identifying what the disagreement actually turns on, and stating what evidence would settle it.
We distinguish between three different situations that are often blurred together: questions where the science is settled and the public conversation has not caught up; questions where researchers genuinely disagree; and questions where the honest answer is that the research has not been done.
Certainty and verdicts
Each claim we examine carries a verdict, and the verdict reflects our confidence rather than our preference. “It’s more complicated than that” means the answer depends on conditions we go on to specify. It is not a way of avoiding a conclusion.
Advertising and editorial independence
This site carries advertising, including programmatic advertising and direct sponsorship. The following rules apply without exception.
- Advertisers have no influence over topic selection, evidence assessment, or verdicts.
- Sponsors do not review content before publication.
- Advertising is visually distinct from editorial content and is never presented as if it were our reporting.
- Where an advertiser sells a product related to a claim we assess, that assessment does not change.
Medical boundaries
We publish general educational information. We do not diagnose, do not prescribe, and do not give individual medical advice. Where a claim touches on medication, dosing, or the management of a diagnosed condition, we say plainly that the decision belongs with a reader’s own clinician.
Corrections and updates
Errors of fact are corrected promptly and the correction is noted on the page rather than made silently. Pages are revisited when significant new evidence appears, and a page that has been substantially revised carries the date of that revision.
If you believe something here is wrong, tell us. Specific objections with a citation attached are the most useful thing a reader can send us.