How every number is made

Each compound gets an evidence label (human trial evidence, limited human data, preclinical research, mechanistic only, or not yet assessed) and, where the studies have been read in full, a letter grade. Grades come from reading the cited studies: the model (cells, animals, humans), design, size and whether results were replicated. They never come from study counts or popularity.

Approved drugs are graded on their regulatory label and pivotal trials. A compound approved nowhere can never be graded as established in humans, and its page lists human studies without amounts.

Read the full evidence standard.

PubMed counts use a title/abstract search for each compound's name and verified synonyms. Human studies are those indexed with the MeSH term Humans; animal-only studies are indexed Animals and not Humans; randomized trials and reviews use publication types. ClinicalTrials.gov counts list studies naming the compound as an intervention, with the highest phase reached. FAERS counts are spontaneous reports, which do not establish cause. NIH RePORTER counts show funded interest, not evidence.

Names shared with other substances can inflate counts, and indexing lags publication. A new study enters a compound page only if the compound name or a verified alias is in its title or abstract, and every addition is checked before it is shown. Report method.

The COA checker reads each certificate twice and applies a published rubric. Grading starts at A- and the lowest cap that applies wins. Nothing scores above A- because certificates are not yet confirmed with the issuing lab.

• Identity: mass-spectrometry masses must match the compound (within 0.1%, minimum 1.5 Da). No identity test or a wrong mass is an F.
• Purity: HPLC purity printed, ideally detected at 200–230 nm. Below 95% caps at C+; no figure caps at C.
• Measured amount: a measured mg per vial. Missing caps at C; outside 90–110% of the label caps at B.
• Endotoxin: required for injectables; missing caps at D.
• Laboratory: a named independent lab; none caps at D; seller-run testing is an F.
• Paperwork: lot number (missing caps at B), test date, and every result inside its own printed limit (any failure is an F).

A grade describes the document, not the material in any vial.

Amiravo sells nothing. Where a compound is available from our referral partner, the page shows the partner's size, list price before any discount, and whether it is out of stock. Prices and stock are read from the partner's public product pages every four hours; its sale calendar, codes and buy-one-get-one items every hour, and specials are shown only while the partner's own sale window is open. Amiravo does not test, compare or vouch for any seller's products, and a listed price is not a recommendation to buy.

When the partner lists a compound Amiravo has no page for, a page is added automatically from primary sources only: identifiers from PubChem, the PubMed abstracts it cites, and ClinicalTrials.gov and Drugs@FDA searches. Each sentence is checked against the abstract it cites, and the page carries no amounts. Compounds approved as drugs, in late-stage trials, or with too little literature are held for an editor instead. Automatic pages are marked as new entries until an editor has reviewed them.

News items and regulatory trackers are written from the primary document (Federal Register, FDA, HHS, court or trial registry), with secondary reporting cited and attributed. Each is fact-checked against its sources before publication and shows a last-verified date. Corrections are logged on the public change log.

Content is partly AI-assisted and human-reviewed. Every meaningful change to a compound page is recorded with its date, before and after, and source. Anyone can report an error; confirmed errors are corrected and logged.