What the evidence grade means (and what it deliberately does not)

Most compounds, peptides and supplements in this catalog carry an evidence grade (some are deliberately left ungraded — see below). It is computed reproducibly from a documented PubMed-count heuristic — the matching counts of phase 3/4 trial reports, randomised controlled trials, meta-analyses and systematic reviews the National Library of Medicine returns. The counts are reproducible; the query design and the letter thresholds are editorial choices, so it is a transparent measure, not a verdict.

What the letters mean

• S — a large human evidence base: several phase 3 or 4 trial reports plus multiple meta-analyses or systematic reviews.
• A — strong human evidence: a phase 3/4 trial report alongside review-level evidence, or ten or more randomised trial reports.
• B — real human evidence: randomised trial reports, or reviews backed by at least one human trial report.
• C — early human evidence: at least one human trial report and not much beyond it.
• D — no records matched the selected human-trial filters. Observational studies, case reports, or newly-indexed trials may still exist; the trial filters simply did not catch them.
• F — very few PubMed records matched the documented query.

The three things it does NOT mean

It is not a safety score. Testosterone grades S. So does caffeine. The grade counts research, and heavily-researched compounds are often heavily researched *because* they are potent, risky, or both. Read the risks section; that is the part about safety.

It is not an efficacy score. A high grade means the question has been studied a lot, not that the answer was yes. Niacin has a large literature, and large trials found no incremental cardiovascular benefit when it was added to intensive statin therapy in the studied populations. Vitamin E has a large literature and the notable trial finding was harm.

It is not about you. Every grade describes a population and an outcome that some researcher chose. "Strong evidence that correcting a deficiency helps deficient people" is a real finding and tells you nothing about what happens if you are not deficient.

Why some entries have no grade at all

Two different situations produce "not auto-graded", and they are not the same.

Name collisions. Some compounds share a name with unrelated research. P-21 is a research peptide; "p21" is also a famous cell-cycle protein with thousands of papers and dozens of trials behind it. No query separates them cleanly, so publishing a number would be a confident lie. Those entries show no grade, no counts, and no citation list — the papers would be about the wrong molecule.

Composite entries. Some entries cover several drugs at once, or a use the published research was never about. The insulin entry is about misuse in bodybuilding; the literature it matches is about insulin, which is enormous and would grade A. A letter that contradicts the text underneath it is worse than no letter, so those show their citations but withhold the grade.

How to argue with a grade

The counts under each grade are the working. If a letter looks wrong, the answer is the query — every entry links to the live PubMed search, which is always more current than anything we cached. Counts are of published PAPERS, not distinct trials, and several papers can report one trial. That is why the grade is a measure of literature volume and is labelled as one.