Editorial policy
Where claims come from
Every claim on this site is drawn from peer-reviewed literature, randomised controlled trials, meta-analyses, and observational studies or, where a primary study isn't available, from mechanistic or traditional-use sources that are clearly labelled as such. We don't base claims on marketing material, press releases, or manufacturer websites.
Evidence grades
Every benefit and risk claim carries a letter grade for the strength of the evidence behind it:
- A: Strong evidence
- Multiple randomised controlled trials, or a meta-analysis of RCTs, support the claim. Example: creatine's effect on high-intensity exercise performance, backed by dozens of RCTs and several meta-analyses.
- B: Limited evidence
- At least one randomised controlled trial supports the claim, but the evidence is limited in size, replication, or consistency. Example: a single small RCT suggesting a benefit that hasn't yet been replicated in a larger trial.
- C: Preliminary evidence
- Observational studies, mechanistic reasoning, or animal research support the claim, but it hasn't been confirmed in human RCTs. Example: an in-vitro or animal study showing a plausible mechanism that hasn't been tested in people.
- D: Traditional or anecdotal
- The claim rests on traditional use or anecdotal reports, with no controlled human research behind it. Example: a use documented in traditional herbal medicine with no clinical trials to date.
Citation policy
Every claim on the site cites a specific, verifiable source: a PMID (PubMed identifier) or a DOI, so you can look up the original study yourself rather than take our word for it. An automated script checks every citation against PubMed and Crossref to confirm it resolves to a real, matching record before publication.
How we treat risks
Risks and interactions are held to the same evidentiary standard as benefits: they're graded, cited, and described honestly. We don't downplay a documented risk to make a supplement look more appealing, and we don't inflate a theoretical risk beyond what the evidence supports.
Corrections
When we get something wrong (a misgraded claim, a broken citation, an outdated interaction) we fix it promptly. All site content lives in version control, so every change to a monograph is tracked and auditable; nothing is quietly edited without a record.
AI disclosure
Draft content on this site is AI-assisted: an AI model helps assemble the first draft of a monograph from the underlying sources. Every draft is human-reviewed for accuracy against the cited sources before it is published.