How evidence is graded

Not all evidence is worth the same. This page explains the four-step meter at the top of every compound page, why the wording on this site is cautious, and what the site will not say.

The evidence meter

Each compound page carries a meter with four steps. The filled step is the strongest kind of evidence that exists for the page's main claim, not the average and not the most hopeful.

Anecdotal. User reports, forum posts, and mechanistic reasoning with no controlled study behind it. Included for context, never as proof.
Animal. Studies in rodents or other animals, including cell and tissue work. These show a compound does something in a living system. They do not show it does the same thing in people, and doses do not translate directly.
Small human. Human studies with small numbers, no control group, or no blinding, plus case reports and observational data. Useful for spotting patterns. Easy to fool.
Clinical trials. Controlled human studies, ideally randomized and blinded, published in peer-reviewed journals. The only level at which a claim about people is on solid ground.

A compound with strong animal data and no human trials sits at "Animal," however promising the animal results look. Many compounds with excellent preclinical profiles never reach human trials, or fail when they do. That is the normal fate of a research compound, not an exception.

How confidence is worded

The wording is meant to match the evidence, not to hedge for its own sake. "May support" and "limited evidence suggests" are used where that is the honest state of the research. Where a claim rests on one small study, the page says one small study. Where no human data exists, the page says that in as many words.

Uncertainty on this site is a description of where the science stands, not a disclaimer.

What the site will not claim

That any peptide provides a specific benefit to a reader.
That any compound is safe, including compounds with clinical data.
Anything beyond what the cited study supports. If the study was in mice, the page says mice.
A dose anyone should take. Doses on this site are the doses used in the cited studies, reported as history.

What people say, and how it is rated

Many compound pages have a second tab with reports from people who say they used the compound. These come from public forums, mostly Reddit communities where people describe their own experience. They are here because readers want to know what users say, and because a report of a side effect is worth knowing even when it proves nothing.

Each report is read in full and sorted into one of four outcomes: helped, mixed, no effect, or bad experience. Only first-hand accounts count. Questions, secondhand stories, and posts that promote a seller are left out. Each report is summarized in our own words, with doses and sources removed, and linked to the original thread.

The rating is simple arithmetic on those reports. A report that says it helped counts as one, a mixed report counts as half, and no effect or a bad experience counts as zero. "Mostly positive" means the reports average at least 70 percent, "Leans positive" at least 50, "Mixed" at least 30, and anything lower is "Mostly negative." With fewer than eight reports the page says there are too few to call.

User reports lean positive for reasons that have nothing to do with the compound. People who felt nothing rarely post. People who paid for something want it to work. Injuries heal on their own, and most users change several things at once. A "Mostly positive" rating tells you what users believe, not what the compound does. The evidence meter on the research tab is the measure of that.

Sources

Claims link to the study they come from, almost always on PubMed. Abstracts are read for every cited study; full texts where they are available. A claim that cannot be tied to a source is either marked as a user report or removed.

Updates and corrections

Pages are re-checked when new studies are published or when a reader sends a correction. The "last checked" date under the byline reflects the most recent review. Send corrections to editorial@peptidebreakdown.com.