How Do We Count a Trial and Grade a Compound?
This page sets out the five tests a study must pass before we count it as a completed randomized trial, and what the four grades mean. It also says what none of those numbers can tell you about the vial in your hand.
Last updated
What counts as a completed randomized trial?
A trial counts when it passes five tests. A study that fails one stays off the count, and it can still appear on the record's study list.
- It is in humans. The AOD-9604 record lists a lab-dish study next to its two trials. That row is not in humans, so it does not count.
- It assigns people by chance to the compound or to a comparison. The five small human studies of BPC-157 had no comparison group, so they fail the second test.
- It is completed. The Phase 2 trial of BPC-157 for hamstring strain is still enrolling, so it fails the third test until it finishes.
- Its results are posted on a registry or published in a journal. The three Phase 3 trials listed on the retatrutide record have reported topline results in company announcements, and none is published, so they fail the fourth test for now.
- It tested the same compound the record describes. The randomized trial listed on the TB-500 record tested thymosin beta-4, a different product, so it fails the fifth test.
Why does a count sometimes show a plus sign?
A plus sign means the number is a floor. Where a record counts more trials than it lists, the number shows with a plus, such as 12+. Today, 27 of the 56 records show a plus sign. Semaglutide is one of them: its record counts 12 trials and lists four study rows. The sign comes off when every counted trial is listed with its registry ID, or the count is corrected.
What do the grades mean?
The grade measures human evidence alone. Approval is a separate fact, shown beside the grade.
| Grade | Label | Rule |
|---|---|---|
| Grade A | Large trials | Several completed randomized trials of a health outcome, at the scale that supports an approval |
| Grade B | Some trials | At least one completed randomized trial of a health outcome |
| Grade C | Small studies | Randomized studies that only measured the body's response |
| Grade D | No randomized trial | No completed randomized human trial. Case reports or uncontrolled studies at most |
Not every row fits this table yet. The grades were set before this rule was written down, and only a few records have been re-graded against it since. Each re-grade is listed in the changelog.
Approval takes one of six values: US and EU, US only, EU only, elsewhere, formerly, or none. “Elsewhere” means a regulator outside the US and the EU. Where the record names the country, the page shows it. “None” means the FDA and the EMA have not approved it. Those are the two regulators we check for every record, so we never say “not approved anywhere”.
What does a grade not tell you?
A grade says nothing about any vial for sale. It also says nothing about what you should do. Wherever a trial exists, it tested the product its maker put into that trial. Nobody can tell you from the outside what is in a given vial. So a grade A does not vouch for a vial that carries the same name. A grade D says only that no completed randomized trial has shown what the compound does in people. We leave the decision with you.
How do we word sport status?
Sport status takes one of three values: Banned, Monitored, or Not named. Banned means the compound is on WADA's Prohibited List. Monitored means WADA lists it for monitoring. Not named means it appears on neither list, and under it we print: “Not named on WADA's list. That is not clearance.”
WADA's List has a class, S0, that bans at all times any substance with no current approval from a government health authority for human use. So a compound with no approval may be banned without being named. If you compete, check the current List and Global DRO, the checker the anti-doping agencies run.
Where do the facts come from?
A primary source is the body that owns the fact: the regulator, the registry or the journal. Every source on a record carries the date we checked it, and every record shows the date it was last checked. A source that is not a primary document is marked “Secondary” on the record. Drafts are produced with AI assistance. Where the record is silent, the page says so.
How do changes and corrections work?
Every change to a record is logged on the changelog, with the date, the compound, what changed and the source. A fact change names the old value and the new one. When we get a fact wrong after publication, the Corrections page records what was wrong and what we changed. The original error is not silently deleted. The rest is in the editorial policy.
Ask better questions.
One email on Wednesdays: which records changed and the source behind each change, plus one claim checked. Quiet weeks say so in one line. Four minutes. Free.