Written by Austin Mao. Educational only, not medical advice.
What gets into our index
Studies in our research index are peer-reviewed, run as randomized controlled trials or open-label trials, have a sample size of at least 10 participants, and are published in an indexed journal. We review and update the index as new research is published.
Every entry names its authors, year and journal, and links the DOI, so you can read the original paper instead of taking my summary on trust.
Primary outcome first
Every trial picks one main question before it starts. That is the primary outcome, and it is the fairest test of the idea. Results on the other questions are interesting, but they are weaker evidence, especially when nobody corrected for testing many outcomes at once.
A good example is the 2021 trial of psilocybin against escitalopram for depression (Carhart-Harris et al., NEJM, n=59). On the primary measure, the two did not differ significantly (p=0.17). Several secondary outcomes favored psilocybin, but they were not corrected for multiple comparisons. Both statements are true, and I give you both.
Blinding is harder than it sounds
In psychedelic studies, participants often work out whether they received the active drug. Functional unblinding exceeded 90% in many psilocybin, LSD and ayahuasca trials, and only 29.5% of 112 randomized trials assessed blinding at all (Orsini et al., JAMA Psychiatry 2026, doi:10.1001/jamapsychiatry.2026.0255). FDA's 2024 decision on MDMA-assisted therapy turned partly on blinding and data integrity.
That is why I report effect sizes, sample sizes and limits alongside every finding. When a participant knows what they took, expectation can carry part of the result.
Who paid for the result
When a figure comes from a company press release or a trial registry instead of a peer-reviewed paper, we label it sponsor-reported. You will see that label on the COMPASS phase 3 numbers. Sponsor-reported means the company announced the result, and a journal has not yet put it through peer review.
Mice are not people
Some of the most quoted findings come from animal work. In mice, a single psilocybin dose raised dendritic spine density and size by about 10%, persisting at one month. That is a laboratory finding. No study has measured a plasticity window in humans, and a BDNF increase has not been shown in humans. When a mechanism comes from animal work, we say so on the page.
Who was in the trial
The psychedelic trials cited on this site studied clinical populations in research settings. They did not evaluate a Ceremonia retreat, and no cited trial tested how to structure preparation, ceremony or integration. A result from people with life-threatening cancer, or from treatment-resistant depression, tells you about those people. It does not tell you what will happen for you.
Research is one thing. Our practice is another.
How we prepare, hold ceremony and support integration is our own practice. It draws on the research, and it is not a research finding. You can read how we work on our method page. Our method
When participants can guess their group
Blinding is intended to keep expectations from systematically changing how a trial is experienced or assessed. With a psychoactive drug, participants or clinicians may correctly guess whether an active dose was given. That can affect self-reports and the relationship around a session, even in a trial described as double-blind.
A careful reader can ask whether researchers measured these guesses, what comparison was used, and whether the people assessing outcomes were separate from those providing support. No single design answers every question. A trial registration tells you what investigators planned; the published results show what they actually measured and reported.
Review of evidence and trial limitations · Preparation guide · Integration guide