Updated
The grade is counted, not judged
Every grade on this site comes out of a rule applied to public registers. No one weighs the quality of a trial by hand, because that would make the grade an opinion, and an opinion is what the rest of this subject already has plenty of. Here is the rule in full.
| Grade | Awarded when |
|---|---|
| Repeated trials | Two or more randomised trials in people, or a randomised trial plus a posted result. |
| A first trial | One randomised trial in people, or one registered trial that has posted a result. |
| Being tested now | A trial is registered and has not reported. |
| Laboratory work | Published work exists, but no trial in people for this use. |
| Traditional use | No trial and no indexed study of this use. |
How a trial gets attached to a use
ClinicalTrials.gov is searched for each species' scientific name and every synonym, one term at a time. A study is kept only if a synonym appears in its title, its summary or the name or description of an intervention — a passing mention elsewhere is not enough. Each kept study is then filed under a use if the register's own condition words, its title or its primary outcome measures contain one of that use's terms, which are printed at the top of every use page so you can see exactly what was matched. The published literature is read the same way out of Europe PMC, filtered to its own publication types for randomised trials, systematic reviews and meta-analyses.
What this rule cannot see
Five things, and they matter.
Whether a trial was any good
A tiny unblinded trial and a large rigorous one both count as one trial. The table shows how many people were in each, which is the single most useful thing to look at, but the grade does not weigh it.
Which way a result went
A grade says something was measured, not that it worked. A trial that found nothing counts the same as one that found an effect — follow the link and read it.
Trials that were never registered
Older work, and work published in journals outside these indexes, is invisible here. Registration only became the norm in the mid-2000s, so a long Japanese or Chinese record can be under-counted.
The difference between a drug and a capsule
Where a trial tested a purified fraction, the dose column says so in the protocol's own words. The grade cannot make that distinction for you, so read the dose.
Publication bias
Trials that find nothing are published less often and less fully. Every count on this site is therefore tilted towards the positive, and that tilt cannot be measured from here.
Who wrote this
Boruch Merkur, editor, The Tree of Life Books. Not a clinician. The judgement exercised here is about what to count and how to say it plainly; the figures themselves are not mine and each links to its source.