Hacker News Reader

Thanks, we really need to bring a bit of balance and data to this topic like this. The hype is certainly real, while the data so far aren't backing this up.

I'm afraid that the public interest in legalisation of psychedelics, motivated by positive experiences in recreational use, leads to a skewed narrative around their health benefits, as a means to open the door for their gradual acceptance and legalisation. This happened to marijuana as well, where health benefits of marijuana were grossly exaggerated to open the door to its legalisation, with medical marijuana as a segway. Marijuana can be harmful and abused, so even though I'm all for legalisation, people should have a balanced view of the benefits and risks to make more informed decisions. Luckily the risk and abuse profile of psychedelics is (much) lower than that of marijuana.

Offered not in the spirit of heartless pedantry but in honest hope that the English language isn’t so egregiously plastic as to allow brand propaganda to take over where etymology left off, I feel the need to point out that a Segway is a rather unusual mode of transport, while a segue is a smooth transition from one theme to another.

The medical marihuana segway, and then this comment, caused some maniacal cackling... thanks

points for couching the correction in such flourid language so it comes across ever so gently

Following that vein — or perhaps that vane, in vain — I hope you don’t mind my also pointing out that it’s terribly “florid”.

Sorry. Truly.

TIL! Nothing to be sorry about, I'd rather know than go around the Internet misspelling it all over the place. :)

This data isn't terribly surprising to me in itself. The promise always seemed much greater for anxiety disorders than for depression, and I honestly believe we'll never find a pharmacological treatment that significantly outperforms SSRIs when depression is defined as broadly as it is in this trial. Depression is really a wide family of disorders that we don't have very good ways of telling apart yet, but likely work quite differently, along with a very fat tail of somatic conditions that are very commonly misdiagnosed as MDD or some other depression diagnosis. This seriously confuses any attempts at comparing efficacies like this.

Anxiety disorders are much better understood and more clearly delineated. And what we see in anxiety land is huge strides being made even just in new forms of therapy. Metacognitive therapy is starting to show real improvement over CBT. Pilot studies with a new form of intensive exposure therapy have shown staggering remission rates of over 70% in severe anxiety disorders like panic disorder(look up the Bergen 4 day treatment). MDMA-assisted therapy for PTSD looks very promising as well.

For depression, dissociatives have looked much more promising than psychedelics for a while now.

Anxiety disorders do seem to be much more tractable than depressive disorders even with traditional treatment approaches and we seem to be making meaningful progress from many directions. The Bergen papers are absolutely wild - I don't think I've ever seen d > 4 in a real psychiatry paper, nor have I seen people continuing to improve post-treatment at anywhere near that rate; obvious caveats apply regarding generalisability and healthcare economics.

https://pubmed.ncbi.nlm.nih.gov/37127598/ https://pubmed.ncbi.nlm.nih.gov/36418989/ https://pubmed.ncbi.nlm.nih.gov/29774005/

I think you're probably right about multiple depressive subtypes that are inadequately differentiated; I think there's a particularly acute problem with the failure to differentiate between short, self-limiting episodes induced by specific life stressors and more durable illness characterised by chronicity and treatment-resistance. The data we've seen regarding elevated inflammatory markers and the associations with severity and chronicity might point towards a meaningful subtype that would benefit from a novel treatment approach, but it's still quite early to say; minocycline has fizzled out as a potential treatment, but the inflammatory pathways are at least worth investigating.

https://pubmed.ncbi.nlm.nih.gov/36103181/

PTSD is a bit of a strange one at the moment, because of the active debate about diagnostic criteria and thresholds; I'm somewhat concerned that an increasingly heterogeneous patient group might muddy the waters in research. I'll elide over the debate regarding cPTSD, because there seems to be far more heat than light. I'm not especially excited about MDMA-assisted psychotherapy for PTSD, but only because we've been seeing comparable effect sizes for specialised psychotherapy without MDMA; again, probably not a breakthrough, but potentially beneficial for a subset of patients.

(Es)ketamine is clearly very useful in the management of acute depressive crises, although I'm somewhat more sceptical about the general applicability for less acute and less severe depressive illness. If I had to wager, I'd say that the next meaningful advance will probably be novel NMDAR antagonists.

> psychedelics are, at best, only a modest improvement over the current state-of-the-art.

Is this measured in terms of efficacy alone, or are you also counting both literal $$ cost, and "cost" in things like side effects? Genuine question - apologies if the tone comes across as harsh at all - you seem like you know this stuff pretty well, and AFAIK those are the other major claims that psychedelics make against pharmaceuticals.

Escitalopram is generic, as are most of the common antidepressant drugs; I'm not sure what the cost is in the US, but in the UK it costs the health service about five cents per dose. Psychedelic therapy is (for the moment) inherently expensive, because established protocols involve a lot of clinical time before, during and after the treatment to protect the safety and wellbeing of the patient. There's simply too much uncertainty to perform a meaningful cost-benefit analysis, but it is plausible that psychedelics could be more cost-effective for more severely and chronically unwell patients who have not responded to other treatments.

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9760680/

We don't yet know enough about the safety profile of psychedelics to draw any firm conclusions regarding side-effects, because of the very small number of patients who have participated in clinical trials. If even a very small proportion of patients treated with psychedelics experience serious harms - and it's very plausible that they could, given the subjective intensity of the experience and the vulnerability of the patient cohort - then that could outweigh the more common but less dangerous side-effects of SSRIs. There are also potential risks that might not be apparent in carefully controlled clinical trials but might materialise in real clinical practice, for example the risk of patients with undiagnosed psychotic traits being prescribed psychedelics.

Even then, escitalopram and SSRIs are not without side effects that make them unsuitable for many, so an alternative that works roughly as well, with different side effects, is still valuable, even if it turns out not to be the magic solution it was once promised to be.

> A subsequent Bayesian reanalysis did show a statistically significant advantage to psilocybin, but this difference was not clinically significant.

What does it mean for something to be statistically significant, but not "clinically significant"?

It's statistically significant because there is a difference in outcome between the group receiving treatment and the one that did not which could not be attributed to chance alone, but the difference isn't clinically significant because it was a very small difference (in statistics, this is sometimes also referred to as the _effect size_)

Wouldn't any statistically significant improvement become clinically significant when applied at a large enough scale, e.g. in global clinical practice?

Imagine someone not hitting a nail with a hammer, striking all over the place, all the time. Now a new hammer comes in, that manages, in a very precise manner, to hit much closer to the nail... but still missing it.

"Clinical significance" is slightly arbitrary and defined in different ways, but it can be taken to mean "an effect that someone could actually notice in practice".

In this trial, the difference between the psilocybin group and the escitalopram group was, on average, two points on the QIDS-SR-16 scale. Those two points might represent someone reporting that their sleep has slightly improved, or that they feel slightly more energetic and slightly more able to concentrate. We can detect that difference in the analysis of the data and we can be somewhat confident that it's a real difference, but it's not the sort of change that most people would be able to distinguish from normal day-to-day variation.

Of course those very subtle differences might aggregate to a meaningful benefit across a whole patient population, but we also have to consider the cost-benefit tradeoff. A tiny benefit might be highly significant if the treatments were very similar in cost, but they aren't - escitalopram is a very cheap generic drug that can be safely prescribed by family doctors and nurse prescribers with minimal follow-up, while psilocybin currently requires a full day in a specialised clinic with significant preparation and follow-up.

https://alnursing.org/wp-content/uploads/2020/03/Depression-...

Does it have to be better than escitalopram? Even if it were at the same level of effectiveness, the lack of side effects and additional insight psilocybin brings are great benefits.

I thought "statistical reanalysis" afterwards was verboten, is there something that makes it ok here?

Not verboten.

Truly dubious researchers will just tinker with the analysis to get the result they want without telling anyone what they've done, which absolutely was not the case here. These researchers followed the accepted best practice of publishing their intended outcomes and analysis methods before starting the trial, which is laudable.

The original paper used standard frequentist methods that are generally accepted by clinical journals, while their follow-up used more esoteric (and arguably more advanced) Bayesian methods. The decision to publish the re-analysis was justified, they were completely transparent about it, it doesn't reflect badly on their professionalism or integrity in any way whatsoever, but it does slightly smack of desperation.

Similarly, some of the authors of that paper subsequently published a paper criticising the validity of the main outcome measure they used, the QIDS-SR-16. The points they make are perfectly reasonable, the paper is informative, it's entirely plausible that they only lost confidence in the instrument because of what they learned from the trial, but it does seem a tiny bit desperate to be making those arguments after the fact.

https://pubmed.ncbi.nlm.nih.gov/37122239/

To be absolutely clear, the team at the Centre for Psychedelic Research at ICL are good guys. They're serious, they do all the right things in terms of research integrity and they clearly care deeply about patients. They worked through a dreadful mire of legal and bureaucratic barriers to get psychedelic research done on a meaningful scale. Nutt and Carhart-Harris and others are absolutely passionate about this area of research, and I just hope that they can find a way through if this disappointing finding proves to be the beginning of a dead-end.