Does "Psychedelic-Assisted Therapy" Have Supporting Evidence?
An argument.
Welcome to The Frontier Psychiatrists. This is a media empire—not just a newsletter—dedicated to writing, podcasting, and even video content about human health. Today, I’m tackling one of my favorite topics, in writing, mainly to save myself time in verbal debates. I’m doing the work to evaluate the evidence…for a perplexingly popular treatment modality, given, I will argue, a paucity of evidence.
Many of us have heard about psychedelic-assisted therapy.
And the way humans work, as far as I can tell, is if we've heard a thing enough, we assume it to be true. Taking an idea out of our heads? Difficult.
Here is what ChatGPT says when you ask it the question,
“What data exists to demonstrate the need for psychedelic treatments to include psychotherapy? Is there any? How strong is this evidence?”
It replied (along with many, many more words, when using the “deep research” setting):
“It is important to note that no modern trial has deemed it ethical to give powerful psychedelics entirely without support, so there is no RCT of (for example) psilocybin with zero therapy sessions for direct comparison. The closest analogs are comparisons of different levels or types of therapy, or using therapy versus some minimal control condition. Interpretations must therefore consider that support of some kind (even if just a few sessions or an integration meeting) is always present in clinical research with psychedelics. Within those ethical boundaries, the emerging consensus is that therapy is a critical component for safety and for helping patients integrate the often intense psychedelic experiences.”
Your author disagrees with the last statement— based on the first!—but the training data for LLMs is biased enough to generate the above as part of its answer, presumably on the way back from the metaverse version of Burning Man for Large Language Models experimenting with radical self-sufficiency?
An Argument, Most Disagreeable
Let's start, to make my overarching point, with an absurd idea.
“High-five assisted blood transfusions.”
If, at the beginning of the use of blood transfusions, it was asserted, by enough people, that “high-fives” were necessary to make the blood transfusion work, and every study undertaken of blood transfusions included a robust high-five at the beginning, one could be forgiven for assuming that “the high-five” was an integral part of the blood transfusion magic in helping those suffer massive physical trauma survive. If I were to assert to you that high-fives were necessary today, after having seen enough blood transfusion data absent a high-five, you would be justly skeptical. To be clear, blood transfusions are lifesaving in major trauma.
Regarding treatment for acute post-traumatic bleeding, literature research has considered standardized massive transfusion protocols as a key component to obtain better survival and a low rate of complications and organ failure.1
However, if they had always been administered together, we would be worried about detangling them. Somebody who offered a blood transfusion without a high-five might even be understood as inappropriate or risky. It would take a clinical trial, where individuals were randomized to either blood transfusion with or without a robust high-five, to tease apart the difference. In fact, we can all agree that it would be hard to blind high-five assisted blood transfusions.
You'd have to come up with an elaborate protocol in which the surgeon and the anesthesiologist were placed in some anechoic chamber, wearing virtual reality glasses and fancy headphones, such that they were unaware of whether a high five had been completed or not, prior to the trauma surgery and blood transfusion. We might even give up on the entire process of blinding and accept the common wisdom that high-fives are part of the process. There might even be some who encourage us to high-five more regularly, or high-five for 12 sessions before, as part of the appropriate set and setting of the trauma surgery, and then to have subsequent post-blood transfusion high-fives in our postoperative check-ups. And then those interventions would also need the elaborate blinding protocol to detangle what, I hope we can all agree, is a completely farcical part of a blood transfusion’s claims to efficacy?
The above may sound stupid—because it is. I would argue that's what's happened with psychedelic-assisted psychotherapy. I'm not anti-therapy, I want to make that clear.
One of the problems with the people doing psychedelic therapy, or at least advocating it, is that they lack practical experience in the design of psychotherapy trials that didn't involve psychedelics. I've run clinical trials in psychotherapy—they're hard, the one trial I was a principal investigator on didn't meet recruitment goals, which is what happens when you try to run a clinical trial with no funding. That being said, I'm not the only person in the world who has conducted a psychotherapy study. Other experts have as well, and could be consulted for new trials.
Psychotherapy studies, as a brief aside, are more complicated than biological interventions in one important regard—it's hard to know if the clinical trial participant is getting the intervention you think they're getting. In a drug trial, you don’t have to worry if the person's gonna get a different drug than the one you prescribed. They're getting a placebo, or the study drug. No one is slipping them Advil instead. However, in a psychotherapy trial, you need to evaluate the psychotherapy session with a validated adherence scale, and then audit those sessions for adherence of the therapist to the therapy model. The MDMA – AT trials that MAPS ran did have this adherence rating as part of the study design, but nobody bothered to ask whether the therapy to which they were being asked to be adherent was any good in the first place.
To make a point, one could imagine a high-five adherence scale that rated the quality of the high-fives with the blood transfusions:
“Please rate on a scale of one to five whether there was a loud cracking sound at the completion of the high five?”
1-2-3-4-5
“Were the high-five participants looking each other in the eye during the high five? Please rate eye contact as 1 (for staring in the opposite direction) to 5 (deep, prolonged eye contact).”
1-2-3-4-5
And so on!
There are plenty of ways to make this seem more scientific, but none of them answer the question: is this an appropriate adherence scale to a plausibly effective therapy model? One would need experience in other adherence scales, for related investigations of other therapies, in order to have any sense of whether the adherence manual in question made sense.
I haven't met every psychotherapy researcher in the world, not by a long shot, nor have I met every psychedelic researcher, but I have met a bunch of each kind of individual, and I've never seen them in the same room.
All of which is to say that to study psychotherapy, you need to have a therapy manual that outlines what the therapy will consist of, and a way to measure whether the therapy that you think you're studying is what was delivered. This doesn't mean that the therapy is any good, or even therapeutic. You need to know that an apple is an apple, and an orange is an orange, so that if you're running an apple study, you can be sure that it's because of the apples. If everybody's sitting around having orange juice in your apple RCT, you can't be sure it's because of the crisp apple flavor that people’s blood sugar has been raised!
Drug studies and device studies rarely have this problem! The one exception for this, for the clinical trials in the audience, is the question of adherence to the drugs by the study subjects.2
Returning to the problem of psychedelic-assisted therapy being without any evidence whatsoever. I’m the senior author of the largest scale review of all the data on the topic as of 2024, but to save you 136 or so pages of Volume 31, Issue 2 of the American Journal of Theraputics, here is a quick— and updated—summary:
MDMA: Never Studied Absent Therapy
MDMA has had several trials of MDMA plus “assisted therapy” using the MAPS therapy manual, of which I’ve drafted critical reviews previously:
Now, Everyone Is Interested in Drug Development
Should MDMA-AT Be Saved, Part III
Saving MDMA (and other psychedelic therapies): Part VI
And one prescient podcast appearance before the FDA decision:
The summary of all of that: even if you believe the now retracted data from a number of the MAPS clinical trials, one of the things they never did was study the MDMA absent therapy. The MAPS trials studied, in theory, MDMA plus “AT” and “AT” alone, but never established that “AT” alone was an effective therapy for PTSD. The studies lacked an arm for “assay sensitivity”—this is a third trial arm designed to demonstrate that an already established therapy, in the recruited population, outperformed the placebo in the studied condition. Typically, this design is employed to address situations where both the active agent and the placebo exhibit similar improvement, and the study sponsor wants to suss out if that was because the trial failed (and thus it’s worth doing another one) or if the intervention failed compared to placebo and also compared to treatment as usual.
In the MAPS MDMA trials, this could include an established treatment like Prolonged Exposure Psychotherapy (or venlafaxine), assuming the blind could be preserved.
MAPS went all in on studying a combination of their unproven therapy plus MDMA, making it impossible to determine if the psychotherapy helped or the MDMA helped or the combination helped or…well, depending on your perspective about the trial design, if none of it helped much more than expectation effects would explain. A review of every controlled trial is available here.
Psilocybin: “Psychological Support” And Its Discontents.
There have been some excellent trial designs in the evaluation of psilocybin with “psychological support.”
At this point in the article, I asked the question: What does anyone actually mean when they say psychotherapy? Psychotherapy does not mean a nice person sitting next to you while you have a tough experience—that is psychological support. We have nurses in ICU settings doing that kind of work routinely. We have direct care staff, and mental health technicians, on inpatient psychiatric units, who have spent time on constant observation with individuals who require it, due to the risk to themselves or others, which is also not considered psychotherapy.
The only individuals who can legally administer psychotherapy as an intervention are generally licensed therapists, and these licenses typically fall into one of two categories: master’s level or PhD level, with the occasional MD added for good measure. One of the important things to know about psychotherapists, other than MDs, is that none of them have actual medical training. To be clear, when I say, “No Actual Medical training,” I mostly mean they don't know what to do in a medical emergency.
If your heart stops, unless your therapist has learned basic life support, they'll have no more idea than any other member of the public. Psychotherapists are trained to be good at psychotherapy, not medical crises. Although some of them are allowed to make a mental health diagnosis, their education does not include the comprehensive differential diagnoses and management around general medical conditions that would include: cardiac arrest, delirium, acute intoxication, seizure, serotonin syndrome, heart arrhythmia, you name it—a lot can go wrong with a human body. Most physicians, to be honest, really wouldn't know what to do all by themselves either, because so much of managing medical emergencies is a team sport. We work closely with nurses, our colleagues in other medical specialties, respiratory technicians, pharmacists, and others. Without this collaboration, our patients would never have the outcomes that modern medicine makes possible. However, the very bad things that could happen mid drug exposure could include these medical emergencies, and we'd rather not have psychologists offering empathy mid-stroke.
When individuals take a psychedelic compound, they generally have an altered state of consciousness. The risk medical treatment is trying to manage is only partially around the healing nature of the experience for the psychiatric condition we are targeting. Still, it's probably also about the person not jumping out the window, having a heart attack, attacking someone else in the room, or any number of other things that can happen when people are acutely intoxicated.
And yes, when people take a psychedelic compound, they are often acutely intoxicated.
What is Intoxication, anyway?
The Cleveland Clinic offers the following helpful guide to intoxication (which can happen beyond alcohol). It can include:
Aggression.
Agitation.
Anxiety.
Decreased levels of consciousness, like drowsiness or lethargy.
Euphoria.
Impaired judgment.
Inattention or difficulty focusing.
Increased energy or hyperactivity.
Mood swings (mood lability).
Risk-taking behaviors.
Physical symptoms of intoxication may include:
Balance and coordination issues.
Body temperature changes (hyperthermia or hypothermia).
Flushed face.
Heart rate changes, like tachycardia.
Slurred speech (dysarthria).
Now, when we know what's causing it, like for example, in anesthesia, we don't say someone's high. To be more accurate, we referred to their state of altered consciousness as “under anesthesia.” This is how most ketamine has been administered for most of the time, for example. Anesthesiologists, I’ll note, aren't generally also experts in psychotherapy mid-sedation. They usually wait until the patient returns to a normative state of consciousness to discuss what happened.