How this index is built

Every record here comes from a public source, carries a link back to it, and is rebuilt daily. That's the whole promise. This page explains where the data comes from, what we do to it, and where it's weakest — because a reference that won't name its own gaps isn't one.

Sources

Clinical trial studies and sites come from ClinicalTrials.gov, pulled through its public API and filtered to interventions involving psilocybin, MDMA, ketamine, esketamine, LSD, ibogaine, ayahuasca, DMT, 5-MeO-DMT, mescaline and salvinorin A. Licensed centers come from state registers: the Oregon Health Authority's licensee list under ORS 475A, and Colorado's Natural Medicine Division register, which is where the NMHC numbers and approval statuses originate. Clinic and retreat records draw on operator websites and public directory listings, then get checked against those registers where one exists.

What we add

Three things. Geocoding, so records land on a map: 4,550 trial sites and 2,206 clinics carry coordinates, resolved to city or region level rather than street address. Substance tagging, parsed from intervention names and center descriptions into a controlled list. And the legal basis field, which is the one piece of genuine editorial work here: the statute, license or exemption a facility operates under, attached to the record and traceable to its source. Of the 172 retreat records that carry one, 49 are the operator's own citation; 121 are the statute or scheduling status we established for that country and substance, where the operator publishes nothing specific; and 2 carry a line we have not yet been able to label. Those are different strengths of evidence and we do not print them as the same thing.

Where it's weak

Retreat data is the softest. Only 172 of 214 retreat records carry a stated legal basis, because many operators simply don't publish one, and an absent basis on a record means we couldn't find a claim, not that none exists. Directory sources also skew toward operators who market in English, so coverage of Latin America and Southeast Asia understates what's there.

Geocoding resolves to city centroids for most trial sites, so a dot marks a city rather than a building. Study status reflects what sponsors last reported to the registry, and sponsors are inconsistent about updating: a study marked recruiting may have quietly stopped. Treat the registry link as the authority, not us.

Clinic records skew heavily toward ketamine (1,942 of 2,023 US clinics), which reflects the actual market rather than a sampling artifact, but it does mean "psychedelic clinic" counts read higher than most people expect.

Corrections

If a record is wrong, the source link is the fastest way to show us. Operators asking for changes should point at a public source we can verify rather than sending copy, since the whole value of this thing is that every field traces back to something checkable.

More on what we're building and why: about the index.