Last updated 2026-08-27
Editorial Policy
The standard is not "has a citation." The standard is whether the cited source directly supports the words on the page.
Evidence review process
- Resolve the exact molecule, fragment, formulation, aliases, and regulatory identity.
- Search regulatory records, trial registries, PubMed-indexed literature, and relevant source lists.
- Extract population, design, route, comparator, endpoint, follow-up, result, and main limitation.
- Separate human outcomes from biomarkers, mechanisms, animal findings, and sponsor-reported topline results.
- Test material claims against the source map and record the review date.
Source standards
Regulatory records and direct human evidence carry the most weight. Trial registrations establish design and status, not efficacy. Systematic reviews help map the file. Animal, cell, and tissue findings remain explicitly indirect. Vendor copy, forums, influencers, and testimonials are not primary efficacy or safety evidence.
Evidence grades
| A | Strong direct human or regulatory evidence. |
|---|---|
| B | Limited or indirect human evidence with material constraints. |
| C | Mainly preclinical, mechanistic, uncontrolled, or otherwise indirect evidence. |
| D | Insufficient, unreliable, mismatched, or contradictory evidence. |
Authorship, review, and automation
A reviewer is identified only when a real person completed a defined review. Automated systems may help locate, extract, compare, draft, and regenerate pages, but they are not sources. Material claims must remain traceable to the underlying paper, registry, label, or regulator record.
Corrections and updates
Substantive errors are corrected in the database and in generated answer, comparison, sitemap, and machine-readable files that inherit the claim. Regulatory status and trial information are date-sensitive and rechecked during substantive reviews.
Commercial independence
App relationships and owned products do not determine evidence grades or safety verdicts.