Reading the evidence
What the standing on each page means, from cell studies to a license, and how to weigh a paper count.
5 min read · 2026-11-16
What the standing on each page means, from cell studies to a license, and how to weigh a paper count.
5 min read · 2026-11-16
Preclinical studies are done in cells and animals. They show that an effect is possible and how it might work. Most compounds that look promising in mice fail in people, so an animal result is a reason to run a trial, not a prediction of its outcome.
Phase 1 trials give a compound to a small number of people to measure safety and how the body handles it. Phase 2 trials, in tens to hundreds of people, look for the dose and the first signs of an effect. Phase 3 trials, in hundreds to thousands, are the ones a regulator weighs. A license means a regulator judged the benefit worth the risk for a stated use, at a stated dose, made to a stated standard.
A randomized controlled trial assigns people to the compound or a comparison by chance, which is what lets a difference be credited to the compound. A case report or an uncontrolled series cannot do that. A meta-analysis pools trials and is only as good as the trials it pools.
Who paid matters too. A sponsor's own trial is not wrong because the sponsor ran it, and most drug trials are run by sponsors, but independent replication is worth more than a second paper from the same group.
Each compound page shows how many papers PubMed holds for it, how many of those are in humans and how many are clinical trials. A large number says a compound has been studied a lot. It does not say the studies found a benefit. A compound with 40 papers and three good trials is better understood than one with 400 papers from cell culture.