01 · Define the exact claim
We preserve the original wording, identify implied universals, and state the population, intervention or exposure, comparator, and outcome when the claim permits it. A broad slogan is not silently converted into a narrow clinical proposition.
02 · Search the evidence hierarchy
We prioritize systematic reviews, public-health agencies, regulator decisions, clinical guidelines, and peer-reviewed human studies. Mechanistic, animal, and laboratory studies can explain plausibility but do not, by themselves, prove human outcomes.
03 · Assign grade and confidence separately
| Supported | Relevant higher-quality evidence converges on the claim as worded. |
|---|---|
| Mixed | Credible evidence or outcomes conflict, or benefits and harms differ materially by context. |
| Unsupported | The claim conflicts with credible evidence or goes materially beyond what it establishes. |
| Unverified | Reliable evidence is too sparse, indirect, or immature for a defensible conclusion. |
Confidence reflects evidence quality and directness. It is not a probability that an individual will experience an outcome.
04 · Editorial and risk review
Health claims receive a second-source check and a medical-risk language review before publication. Legal claims must name the jurisdiction and date. Commercial relationships never determine a grade.
05 · Date, monitor, correct
Every record carries a review date. High-risk or fast-changing claims are checked at least quarterly; other records at least annually. Material changes trigger a correction note and newsletter notice.