Guide: Comment nous évaluons les preuves

A lack of Western documentation is not the same as a lack of effect. When the literature has little on a practice, it means Western science has not measured it yet, not that the practice does nothing. What we can tell you is what the evidence supports, what it counts against, and where no one has looked, so you can decide for yourself.

The four tiers

The tier records how strong the evidence for a claim is, from trial-tested down to a first signal.

Strong Strong. Randomized controlled trials, or a large and consistent body of well-controlled evidence. A Strong claim has trial-level or equivalent support and has held up across studies.

Moderate Moderate. Real supporting trials, either smaller or mixed in their results, and consistent with the mechanism. The direction is reasonably clear. The size of the effect is not settled.

Emerging Emerging. Early and promising: small trials, a strong mechanism, and active research underway. Emerging does not mean no. It means the question is open and being worked on right now.

Preliminary Preliminary. A mechanistic or observational signal, with human outcome data thin or absent. We present the signal and say what is missing, so you can weigh it yourself.

We do not use a "weak" tier. It reads as a verdict on the therapy when what we are describing is the state of the research. Preliminary carries the same information without the editorial.

Grading the claim, not the therapy

A therapy is not Strong or weak as a whole. Heat exposure has good randomized evidence for vascular function and an observational signal for mortality. It has one Strong trial and one null follow-up for depression. Its effect on growth hormone is measurable but minor.

Calling the whole therapy Strong would overstate the weak parts. Calling it Preliminary would understate the strong ones. So each claim gets its own tier, and they sit next to each other on the page.

Every claim lives in a single record with its tier, its effect size, the population it was measured in, its limit and its source. That is why the same fact cannot appear at two different strengths on two different pages: there is only one copy of it.

Strength is not endorsement

A tier tells you how good the evidence is. It does not tell you the news is good. Strong evidence can point at a benefit, at a risk, or at nothing. So every chip carries two things at once: the tier, and the direction of what was found.

A benefit keeps the color of its tier:

• Strong green for Strong,
• Moderate amber for Moderate,
• Emerging blue for Emerging,
• Preliminary sage for Preliminary.

Strong · risk A finding that points at a risk turns red at every tier, so a caution is never shown in a calm color.

Strong · no effect A null result, or one that is mixed, turns grey at every tier. A grey chip never turns green, because green would read as "this works" when good trials looked and found nothing.

Color alone does not carry the meaning. Green is a Strong benefit; amber, blue and sage are benefits at a lower tier. Red is harm and grey is no-effect or mixed, at any strength. Read the tier and the direction together, never the color by itself.

Observational findings

A large, consistent body of observational evidence can reach Strong. A single striking cohort cannot. Cohort studies are how we learn about things that cannot be randomized, and some of the most important findings in this section are observational.

When the research has not looked

Some practices have very little Western literature behind them. That is a fact about the literature, not about the practice.

Research follows funding, and funding follows patents, products and academic careers. A free practice that no one can sell has less reason to be studied than a device that can be. So when we say the evidence is thin, we mean the studies do not exist, and we mark that apart from studies that exist and found nothing. Those are different situations and they deserve different sentences.

The frontier axis

Science-library pages carry a second, separate label: Established, Emerging, Speculative or Debunked. It describes whether the underlying science is real, a different question from whether a therapy built on it works.

That cells hold voltages and that wounds generate electrical fields is Established. That a consumer device based on that idea does anything is a separate claim, graded separately, and usually much lower. Keeping the two axes apart lets us take frontier science seriously, and still grade a product built on it by its own evidence.

Where something has been tested under blinded, replicated conditions and failed, we mark it Debunked and give the specific reason it failed.

What we do not do

We do not chase down every claim other people make. We follow the research and the observations and mark what each one is.

We do not put a product on a page about a condition. Where commerce exists at all, it lives on the page whose subject is that class of product, one click from a reader's symptoms. Some topics carry no product at all.

We are also wrong sometimes. Everything here is researched, cross-referenced and rewritten, and that does not make it free of error. Where you find one, tell us and we will fix it. Our full method is in our editorial policy.

The Chinese medicine lens

Chinese medicine is not graded on these tiers, and it would be a category error to try. It is a clinical system built by observation over two millennia, with its own internal logic, its own diagnostic method, and its own criteria for what counts as knowing something.

When we place it beside a mechanism, we are laying one descriptive language over another because it is illuminating, not because either one validates the other.

We also present what the tradition holds, including where it cautions against something. Chinese medicine does not treat sauna as universally beneficial, and we say so.