Edzard Ernst: Quackery is on the rise, and the placebo effect is part of the problem

So that all other things are as equal as possible, it's basically A/B testing. If it's not done right it doesn't give you reliable results and it's rarely perfect anyway, you can only reduce uncertainty, never eliminate it. There are always individual biological differences that can't be accounted for. The average values for the average human body only hold for other aggregates, not very well on individuals. That's why size is usually what makes this "effect" disappear, they almost never survive regression to the mean, and the few that do are either too trivial or random to be useful.

But there has never been any significant and reliable objective changes observed to such an effect. The effects that are hyped are always subjective, about "helping". Observing "a change" in something doesn't necessarily mean anything. Or argued in some way that accomplishes the same, like how cortisol is often used as a proxy for stress, even though cortisol does a crap ton of things. That just falls in all other things not being equal.

The same principle is commonly used in economics, where the problems are the same: you can't ever account for all factors going into why people make some decisions over others. But it's understood to be a limitation, at least by most. That's why economics has moved away from holding their early sacred models of homo economicus, the rational actor who optimizes utility based on perfect information, to understand that they are just that: models.

 

That is the claim, but here are the results from one of the cited studies:

https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0049477

This was an attempted replication of the initial study that claimed a result for four re-expositions.

But it's worse -they didn't actually directly measure cytokine levels. They isolated PBMCs, stimulated them in cell-culture and then tried to measure a proxy of IL-2 levels - in this study is was via flow cytometry, in the prior study it was via gene expression (mRNA).

The prior study also measured several neuroendocrine parameters (epinephrine, norepinephrine, and cortisol concentrations) and found no effect.

A later study of renal transplant patients failed to replicate the effect - they found no evidence of a conditioned effect on IL-2 using similar methodology - exposure to CSa, stimulating PBMCs in culture and then measuring gene expression. They also found no effect on epinephrine, norepinephrine, and cortisol concentrations.
https://pubmed.ncbi.nlm.nih.gov/29610294/ (note this study didn't have a control group, so the other claims about T-cells are low-quality evidence)

A study also utilising cyclosporine A found no evidence of a (patient reported) nocebo effect.
https://www.sciencedirect.com/science/article/pii/S0149291818304594

But there is some potential evidence for reduction in cytokines due to conditioned effects of endorphins in the brain:
https://www.nature.com/articles/s41380-021-01365-x

This is the kind of study I would be looking at if I was a placebo researcher, rather than relying on stimulation of PBMCs outside of the body to try and measure an effect.

 

While I'm at it...

What they said:

The reality:



Day 5 was the placebo, Group 1 was conditioned with insulin, Group 2 glucose, Group 3 saline. # means P<0.1 - they're really scraping the barrel with that one, given the graph doesn't look particularly impressive. - there was an immediate reduction in insulin on the 5th day in all groups, contrary to the claim of the "science" show above.

I'm really frustrated with the frequent claims of "evidence" that isn't evidence at all.

 

Indeed, just scary that the conclusion and abstract didn't underline absolutely with the side of common sense and history in their discussion of that 'non-finding'.

 

135day rule starts to sound even more important for charlatan-filtering.

Could we come up with some catchy storytelling-style analogies for this one? I'm thinking there is material in the sector of hairdressers to cover a few of these effects

 

If it is the same one that was on the TV programme (trust me I'm a docor I think) then the phrase 'conditioned response' is something I mentally debate about whether it has either been stretched as a term or there is something else more strictly accurate.

Given the drink was in fact medicine the first many times (rather than someone being just told it was) with eventually the v unusual concoction being drunk without the medicine then I'm not sure where the 'mind' or 'behaviour' features if it is direct.

It's not like a trained response in the same way when you've administered a painkiller in the drink (and never had that exact drink before or without said painkiller), as it is to ask someone to do a task and getting an electric shock or treat.

now I'm thinking about that fun one when I was a kid of eating say an orange sweet that tastes of liquorice, or decaf coffee when you normally have caffeinated. Closer to eating a berry or substance the body is trained/innaely programmed (in some way) to expel from the system due to being poisonous or something? I find it interesting to consider the words for.

In the example in the link it's all more direct with no emotions or 'thinking' involved but proprioception, perception, senses then something which activates neurotransmitters or neuron-like cells or other actual effects in the body (some of which might actually come from the drink I don't know) etc. The drug itself which has to break down and do its various things before eliciting that effect is what is the 'instruction' for the first however many times. It isn't immediate - which is normally the rule of effective conditioning, and it sort of isn't placebo because of the order in which it is done.

Anyway one day I might pick up my thinking on this one to finish it through my brain tends to wear out around the same difficult point.

 

Even accepting the premise that there is something to this placebo effect, by definition it would be the least effective treatment used in medical care, since every treatment first has to prove it does better. You can take the lest effective treatments out there, ones with dubious efficacy claims, and every last one of them still outperforms "placebo". Even in the very best case, it is so trivial an effect that you have to torture data in order to get it to confess to anything. This is after decades of people obsessing over it, trying every possible variation to make it systematic.

But following the logic of this effect appearing in clinical trials, if it existed, supplements would work pretty well. If this white lab coat effect worked, all physicians would have to do is prescribe them, give the ritual substance. They wouldn't be miraculous, but they would work, better than a control with sugar pills anyway. It's pretty well-established that they don't work, although given the quality of assessment, it's more of a shmaybe they don't work, since the normal process is not good enough to validate trivial effects. However, in the very best case, those effects would be trivial, and thus not worth any effort even in the very best case, since it would still underperform against literally every single other medical treatment.

And that's in addition with the uncomfortable truth that this "effect" seems to get lower and lower with time. The same with CBT. Not that it is actually getting lower, it's simply disappearing by way of being exposed to more scrutiny and regression to the mean. All of this only makes sense in the light that it's simply an illusion, just as convincing as the celestial spheres, and just as wrong and self-centered.

This is a completely different thing than actively trying to exploit this effect by manipulating people into thinking positive things and doing more than they think they can, or whatever, which is the basis for the whole CBT for MUS/whatever. This is not what is found in the "evidence" for placebo, since they were not trying to convince people to downplay or ignore their symptoms or anything like that. What they found is simply that by not being able to account for everything that can affect an outcome, they can't account for everything that does affect the outcome. This isn't especially hard, it just has the ego-lowering truth that medicine still has far more work ahead of it than behind it.

The basis for this biopsychosocial approach is the plac/nocebo effect. Without accepting the premise of this effect being "one of the most powerful interventions we have", in the words of Simon Wessely, the entire paradigm simply does not make sense, it is derived from a belief that we are experiencing a nocebo effect and so it's natural that the solution is a placebo.

And frankly at this point I think it's the main reason people are pretending that Santa's still real, at least when talking to the children. Too much has been invested in this illusion that it's simply become impossible to actually voice seeing the emperor's naked butt.

 

Sounds like that pseudoskeptic Ben Goldacre.

I guess a placebo could change heart rate or blood pressure but not actual disease.

 

Perhaps Ernst is part of the problem, too, if he is not willing to look at one of the issues within conventional medicine that drive people towards quacks — all the more as he is well equipped to do so —.

https://twitter.com/user/status/1555893683697979392


https://twitter.com/user/status/1555908130093113344


https://twitter.com/user/status/1555911226370277377

 

This seems to be getting close to a field of expertise in identifying, as poor quality, T-shirts that come in a package saying 'poor-quality T-shirt'.

 

The refusal to even say anything on the issue is so damn bizarre. You'd almost think people are getting death threats if they ever dare talk against psychological pseudoscience, or engage on the issue of chronic illness. I freaking hate the secrecy over this. It's like we're secretly an untouchable caste, acted out openly but cowardly never admitted to.

That's how you know an issue is 100% politics. Skeptical, but only if it's popular. A profession unable to face uncomfortable truths, going strong for decades and not about to let go. This is exactly who you want dealing with health issues! Not.

 

Didn't I just tell you that I will stay in my area of expertise?

Well, actually, no you did not. What you said was that

I prefer to stay in my area of expertise.

Which is rather different. If he can be condescending, I can be pedantic.

 

If you think about people going into a trial optimistic that this could be 'the cure' for whatever horrendous thing they have, and told the fact that if they don't get x% and miss it by just 1% then it won't get to round 2 etc then I see it as 'trial effect' rather than placebo anyway. You'd probably do absolutely everything right, try your hardest on what might be impossible to sustain longer term not have that drink or sausage roll, take medication at exactly the right time etc, report if you've done some unusual activity that would affect your condition (dragged off to a wedding somewhere far away, taking 5 buses to get to a hospital appt then be left sitting on a hard chair in bright lights etc), others around you might have hope or you might even be in a situation where you are taken away from exertions and are being paid for doing the trial, others might be more diligent about their job than usual and inc laypersons treat you as they should etc.

That's why you have to subtract it in objective trials, even when it is objective measures and that is why it has to exist as a blinded control. I can't see how it could transform any of these, but given blood pressure and heart rate are snapshots and despite the hype it is small differences in 'placebo' that relate to people not having as hard a time as normal and the 'unexpected stuff' that we all know set illnesses back often leading to drop-out

The more and more I think about some BPS research the more I wonder whether they take a cohort and 'lose' the most ill in the drop-out data, miraculously ending with a 'more well' cohort at the end - I'd like to look back at the methodology of some. I'm not sure when the treatment actually harms people whether ironically blinding or even subjective would have made a difference without forcing for all drop-outs to be accounted for in data, properly (and the question of how we think that should be is a long, big thing - when you look at how many in some treatments) because the worse the treatment for people the more you had to be 'weller' to survive it e.g. GET vs not being forced to do GET might mean the GET group gets more 'whittled' (survival of the fittest)

 

Ernst is now retired but he used to be professor of complementary medicine at the University of Exeter.