Define the claim before searching
“Raises NAD”, “improves walking speed” and “extends lifespan” are three separate claims. Each requires its own evidence.
Prioritise human outcome evidence
Randomised trials and systematic reviews receive the most weight when the intervention and outcome match the claim. Observational findings are useful but cannot prove causation.
Keep biomarkers in their lane
A biomarker may demonstrate biological activity. It only supports health benefit when the marker is validated and the change is meaningfully linked to outcomes.
Label animal lifespan work explicitly
Animal experiments can be unusually informative for ageing biology. Translation to humans still depends on dose, metabolism, disease context and clinical trials.
Record safety uncertainty separately
Short-term tolerability in 40 people does not establish multi-year safety in millions. Known medicine interactions and vulnerable groups remain visible.
Revise rather than defend
Dossiers carry a review date. Larger trials, credible adverse-event signals and high-quality null results can lower as well as raise a rating.
The five-point scale
- Strong: multiple relevant human studies or a robust evidence base with reasonably consistent outcomes.
- Moderate: useful human evidence with important limitations, inconsistency or population dependence.
- Limited: a small or mixed evidence base that may justify cautious interest.
- Preliminary: early human signals, surrogate endpoints or evidence that does not yet support the public claim.
- None: no meaningful evidence in that specific domain.
Conflicts and commercial evidence
Industry-funded work is not discarded automatically. Funding, proprietary formulations, author interests, selective endpoints and independent replication affect confidence and are stated where material.