Ibogaine for PTSD: One Cohort, Six Papers

This is now the most talked-about thing about ibogaine, and it moved fast: from almost nothing in 2020 to a Nature Medicine paper, congressional interest and state funding. The underlying evidence has not grown at the same rate.

Start with the registry

The simplest check anyone can run is to look at what has been registered.

Searching the trial registry for ibogaine returns nine studies. For PTSD and brain injury, this is the complete picture.

Not one person has ever been randomised to ibogaine for PTSD or traumatic brain injury. The only interventional study ever registered was withdrawn having enrolled nobody, and it was neither randomised nor blinded in the first place. It reappeared as an observational study run by a private research company.

The study everyone means

Thirty male Special Operations veterans with a history of traumatic brain injury received ibogaine together with intravenous magnesium at a clinic in Mexico. The results, at four to five days and again at one month, are among the largest ever reported for any psychiatric intervention.

Cognitive testing improved too, across processing speed, executive function and memory, with no decline in any domain. That matters, because neuropsychological tests are harder to move by expectation than a rating scale is.

There were no serious adverse events. Every participant had transient ataxia and tremor, resolving within a day.

These are real measurements and they deserve to be taken seriously. What follows is not a debunking. It is what has to sit beside them.

Who these thirty people were

The design fact that governs everything is in the paper’s own methods.

Participants had independently scheduled themselves for treatment at a clinic in Mexico, after being approved for a grant from a veterans’ non-profit. The researchers played no part in administering anything. The consent form states that the study team is not supporting, facilitating or condoning ibogaine use.

So this is a cohort of people who had already decided to go, already been funded to go, and already been assigned a coach.

The cohort was also narrow in ways that matter for anyone reading it as a prediction about themselves. All thirty were men. Twenty-six were white. Mean estimated IQ was 114. Twenty-eight of the thirty had mild traumatic brain injury, not moderate or severe.

The instruments are not in the registration

This one is checkable in a single click and it is the sharpest appraisal point available.

The registry entry for the study lists one primary outcome and one secondary outcome, both of them the disability score, at two timepoints. That is all.

CAPS-5, MADRS and HAM-A — the PTSD, depression and anxiety measures that produced every headline effect size — are described in the paper as prespecified secondary outcomes. They do not appear in the public registration. The registry entry still lists enrolment as an estimate rather than an actual figure, and no results have been posted.

Registering outcomes in advance exists to stop a study from choosing, after the fact, which of its measures to lead with. Here the measures that led are the ones that were not registered.

The same thirty people, six times

This is the part most likely to mislead someone reading around the subject.

Since the original paper, five more have been published on the same cohort: resting EEG, functional brain imaging, structural brain imaging and brain-age estimation, a qualitative study of the subjective experience, and a study relating the intensity of the mystical experience to symptom improvement.

None of them is a replication. They are six analyses of thirty people, and each lands as its own headline.

The imaging teams are candid about their own work: the functional imaging is a whole-brain exploratory analysis, and the structural authors note that the scans they used are sensitive to non-structural change. One of the papers took the unusual step of having its conflicted senior author recuse himself from analysis, with two unconflicted investigators supervising instead.

Who funded and who benefits

The site applies this to every study it leans on, and it applies here.

The paper’s competing-interests statement declares that two authors are inventors on a patent application covering the safety of this protocol, and two on a second covering ibogaine for disorders of brain ageing. Three authors are shareholders in the company that operates the clinic where the treatment took place, and are listed with that company as their affiliation. One founded a company that sources and converts ibogaine precursors.

All of it is disclosed. None of it makes the measurements wrong. It does mean the most-cited favourable result in this field was not produced by disinterested parties, and that the clinic which treated the participants also employed some of the people who wrote it up.

The scale of that patent position can be checked rather than inferred. The university named in the paper’s affiliations appears as applicant on fifteen patent documents carrying an iboga alkaloid on the front page of the WIPO register, and an independent claims-level study places one of its filings, on iboga alkaloid mixtures, among the applications with the most decisions still outstanding. The iboga patent record sets out the queries.

What the rest of the literature adds

Three other human sources exist, and the pattern in them is instructive.

Eighty-six veterans treated at a Mexican clinic were followed prospectively for six months. Improvements were real and much smaller: effect sizes between 0.27 and 0.41 at one month, against the Stanford cohort’s figures above 2. This is a cohort nearly three times larger, treated similarly, and it reports effects five to ten times smaller. It also gave 5-MeO-DMT after the ibogaine, so no ibogaine-specific effect can be separated.

Fifty-one people answered a retrospective survey in which they rated their functioning before and after, both from memory, in one sitting. It reports the largest effect sizes in the entire literature. Retrospective before-and-after recall is the most inflation-prone measurement in psychiatry, and this is a clean demonstration of why.

Three people appear in a case series of iboga microdosing after brain injury, alongside weekly psychotherapy. Its authors’ own conclusion is unusually well-calibrated: the findings do not establish causality or any iboga-specific effect.

What a careful reader should take from this

The most recent independent review of thirty years of ibogaine research puts it plainly: observational data suggest ibogaine may alleviate symptoms of opioid use disorder, PTSD or polysubstance dependence, and these findings remain exploratory, with most positive efficacy data coming from uncontrolled, open-label or retrospective studies at high risk of bias.

That is not a reason to dismiss the veteran results. Something happened to those men and the cognitive testing is hard to wave away.

It is a reason to be precise about what is known. Thirty self-selected people, one month, one team, six papers, three unregistered instruments, and nobody randomised. Set against what the treatment can do to a heart, that is the whole basis on which a decision would be made.

What the trials found covers the wider registry picture, and why there is no success rate explains what a real outcome figure would require. Ibogaine for veterans covers the practical side, what the route to treatment actually is and who is paying for it, and the Stanford study as a public event covers how one cohort has been reported as a sequence of findings.