How a compound is graded
Every record answers three questions: what the human evidence shows, what the regulatory status is, and what is actually in the vial. The rules below are applied the same way to an approved drug and to a research chemical.
Evidence tier (per claimed use)
Each use the compound is marketed or claimed for is tiered separately, not only its approved use. A compound with three claimed uses shows three ladders.
| Tier | Definition |
|---|---|
| E1 | Multiple adequately powered RCTs, consistent |
| E2 | At least one adequately powered RCT |
| E3 | Small, underpowered or conflicting RCTs |
| E4 | Uncontrolled or non-randomised human trials |
| E5 | Case reports and case series only |
| E6 | Animal models only |
| E7 | In vitro only |
| E8 | No published evidence of efficacy in any model |
“Adequately powered” means the trial was designed and sized for the outcome reported. Tiers are checked by the build: an E1 claim with fewer than two human RCTs, or an E6 claim with any human trial, fails and is not published.
Safety grade
| Grade | Definition |
|---|---|
| S1 | Large long-term human exposure (post-marketing surveillance) |
| S2 | Substantial controlled-trial safety data |
| S3 | Limited human safety data |
| S4 | Minimal human exposure data |
| S5 | No human safety data |
A documented harm is listed with its citation, in the only red on the site. Absence of data is stated and never presented as safety.
Class
A, Approved: marketing authorisation in the jurisdiction. B, Investigational: in active, sponsored clinical development and not approved there. C, Gray market: neither. Class is assigned per jurisdiction: one compound can be A in one country and C in another.
Verification
- Every journal citation carries a DOI or PMID; a script resolves it against Crossref or PubMed and compares the title. A mismatch blocks publication.
- Every NCT number is checked against the ClinicalTrials.gov API, title for title.
- Every regulatory status cites a primary regulator source (FDA, EMA, MHRA, Health Canada, TGA, PMDA, NMPA, WADA), a verbatim quote and the date it was read. Where no source has been verified the table says so.
- Registered-study counts come from the ClinicalTrials.gov API and are dated.
- Editorial rules are enforced by a linter: no dose amounts, no administration language, no vendors or purchase language, no advice, no testimonials, no benefit verbs without a named model.
Records are researched with AI assistance under a written protocol and then machine-checked; every record names its reviewer and review date. The previous version of PeptideOS failed this standard, and was withdrawn (see the corrections log).
What this resource will not say
No doses, no administration, no vendors, no advice: what we won’t tell you, and why.