How we research and review

Every page on this site is built from a structured evidence record, not from free-floating opinion. This page explains the workflow, the records we keep, and the limits of what we claim.

What we publish

• Compound profiles follow a fixed 14-section template: a plain-language short answer, what the compound is, why it is being studied, what human studies show, what laboratory and animal studies show, evidence by outcome, how researchers think it works, safety and unanswered questions, study context, status and access by country, testing and documentation, related compounds, common questions, and sources with methodology and changes.
• Study summaries explain one study at a time: what researchers tested, the main findings, what the findings mean in context, the limitations, and the funding and conflicts behind the work.
• Research topic hubs organize questions readers recognize into measured outcomes and the compounds studied for them.
• Comparisons place matched measured outcomes side by side and flag where studies are too different to compare directly.

Editorial workflow

A profile moves through these stages before publication:

1. Draft: content is being written and extracted.
2. Source extraction: study details are recorded in structured records.
3. Scientific review: findings and conclusions are checked against sources.
4. Editorial review: clarity, language rules, and template compliance are checked.
5. Jurisdiction and commercial review: country-specific claims and any commercial content are checked where needed.
6. Published: the page is public and may be indexed.
7. Scheduled reassessment: the page is due for a fresh literature check.

Drafts and unreviewed pages are never indexed. A page still awaiting review says so on the page.

How we read a study

Each study record captures:

• Title, authors, publication year, and identifiers (DOI, PMID, registry numbers)
• Publication status: registered, recruiting, completed without published results, preprint, published, corrected, or retracted
• Study design and classification: human, animal, or in vitro
• Population or species, sample size, intervention, comparator, and duration
• Outcomes measured, effect estimates with uncertainty, and adverse events
• Funding, conflicts, limitations, and the review history

A trial registration is not a results paper. Preprints are labeled as not peer-reviewed. Retracted studies are never cited for efficacy claims.

Evidence conclusions

Each conclusion belongs to a specific combination: compound and formulation, measured outcome, population, and route or context. We do not assign one universal score for whether a compound works.

Each conclusion records:

• Evidence type: human randomized, human observational, animal, or laboratory
• Direction: favorable, no clear difference, unfavorable, mixed, or not assessable
• Magnitude: a numeric estimate with units and comparator, where defensible
• Applicability: who was studied, what formulation and route, and for how long
• Certainty: shown as PROVISIONAL until the rubric is approved by a named qualified reviewer

Certainty ratings: shown as provisional

Evidence tables show the provisional marker on every rating and carry a footnote stating the rubric is pending approval.

The proposed rubric covers risk of bias, directness, consistency, precision, and publication bias. If it adapts an existing grading framework, we will explain the adaptation plainly rather than claim formal compliance.

Corrections and version history

Substantive corrections are published in the page's "Sources, methodology, and changes" section and in the public corrections log. Cosmetic edits never reset a page's substantive review date.

Review cadence

• Regulatory and provider status: monthly checks plus event-triggered review
• Vendor shipping and price observations: rechecked before publication and at least monthly while shown
• High-interest compound evidence: quarterly literature surveillance
• Lower-interest research: at least every six months, adjusted to publication activity

What we do not do

• We give no dosing guidance and no medical advice.
• We recommend no suppliers and no products.
• We run no medical chatbot and no diagnostic quiz.
• We never ask readers for symptoms, medication use, or medical history to browse research.

The public byline on all research content is the Evidence Research Team. We publish no individual names and claim no clinical credentials we do not hold. See Funding for how the site is paid for.