Editorial and evidence-grading policy

This encyclopedia is a free reference on dietary supplements and nutrients. Each page summarizes what human research has tested, in which form and at what dose, with a link to every study we rely on. It is educational information, not medical advice, and nothing on it is intended to diagnose, treat, cure or prevent any disease.

Evidence grades

We grade each compound and outcome pair (for example, magnesium and sleep quality), never a product or a brand. A grade describes how strong and consistent the human evidence is, not how large the effect is. When the evidence comes from a different form than the page is about, the row says so in its form column.

• Grade AAt least two concordant randomized controlled trials (RCTs), or a meta-analysis of RCTs, with a combined total of 300 or more participants.Example: A meta-analysis pooling 12 RCTs (n = 1,100) that finds a consistent effect on the same outcome.
• Grade BAt least one adequately powered RCT, or a consistent set of smaller RCTs.Example: One well-run RCT with 200 participants and a pre-registered primary outcome, with no larger trial contradicting it.
• Grade CSmall or short RCTs, or trials with mixed results.Example: Two 8-week trials of 40–60 people each, one positive and one null.
• Grade DA single small trial, or uncontrolled human data (open-label, before-and-after).Example: One open-label study of 20 people with no placebo group.
• No effectAdequately powered RCTs that found no effect on the outcome.Example: A 300-person RCT with no difference from placebo on its primary outcome.

Animal and cell studies never earn a grade and never appear in an evidence table. We mention them only when explaining a proposed mechanism, and we label them as such. Grades are re-checked every time a page is reviewed.

How to read an evidence row

Each row reports one outcome from human trials:

• Effect direction: ↑Improved, ↓Worsened, ↔No effect compared with the control group, or ?Mixed when trials disagree.
• Grade: the A–D scale above.
• Best evidence: study design, number of participants analysed (n), publication year, population, the form tested, the studied dose and the duration.
• PMID: the PubMed ID, linked, so anyone can read the abstract.

Doses are shown as studied. For minerals we give the elemental amount (for example, milligrams of magnesium rather than milligrams of magnesium glycinate) and explain the arithmetic, because labels and marketing do not always use the same basis.

Source hierarchy

1. NIH Office of Dietary Supplements fact sheets, Cochrane reviews, and meta-analyses or systematic reviews of RCTs.
2. Individual randomized controlled trials.
3. Observational human studies (context only; they cannot show cause and effect).
4. Mechanistic, animal and cell studies (mechanism sections only, labelled).

Interactions are listed only when a source documents them: an NIH fact sheet, FDA labeling or safety communication, NCCIH, or a published trial, pharmacokinetic study or case report. We do not add theoretical interactions from mechanism alone. When we find none, we say “none found in our sources”, never “safe”. Intake references (RDA and upper limits) link to the NIH ODS tables rather than copying them.

Claims and dosing language

We follow the FTC’s standard of competent and reliable scientific evidence for health claims. Pages describe what studies found; they do not say a supplement cures, treats or prevents a disease, and disease names appear only when describing who was studied. Dose sections use fixed wording (“doses studied in human trials ranged from…”) and never tell readers what to take. An automated linter blocks imperative dosing words and disease-claim wording before a page can be published.

Who checks these pages

Pages are written by the Kymata Health editorial team. Before a page is published, a separate review pass tries to find what is wrong with it: it re-checks every PubMed ID against the PubMed record and its abstract, confirms each number against its source, and checks the wording rules above. Before any page is opened to search engines, the publisher audits it on a preview build.

We do not currently have a credentialed medical reviewer (such as a pharmacist or registered dietitian). Our pages therefore never say “medically reviewed” or display a reviewer credential. If we add a credentialed reviewer, their name, credentials and license details will appear on each page they review, and their review will be logged in that page’s update history.

How we use AI

AI tools help us search the literature, draft summaries and run checks. They do not have the last word: every study cited on a page is matched to its PubMed record, every evidence row is checked against the study’s abstract by a separate review pass, and pages that fail those checks are not published.

Affiliate links and rankings

Compound pages carry no product cards, prices, rankings or affiliate links. Kymata Health earns money through affiliate links on separate product pages, which are clearly marked and list only brands we have an affiliate relationship with. Commission never affects what a compound page says or how evidence is graded. See our affiliate disclosure and site editorial policy.

Corrections and updates

Every page shows when it was last reviewed and keeps a “What changed” log with the date, the section, a plain description of the change and what prompted it (new research, a reader report, an NIH fact-sheet update). If you spot an error, email contact@kymatalabs.com. We aim to fix confirmed errors within 5 business days and log the fix on the page.

← Back to the Supplement & Nutrient Encyclopedia