2025: The 2019/24 Cochrane Larun review Exercise Therapy for CFS - including IAG, campaign, petition, comments and articles

[Medfeb]

Of course all of the issues are relevant, but some have wider reaching impacts than others. And some aren't evidence of unscientific practice.

So far, the people involved in the new Cochrane review process who have spoken seem to have focussed on one single issue, claiming that the issue is evidence of unscientific research, when it unequivocally is not. They have ignored both a trial design issue that makes any reported benefits unreliable, and, I think I am right in this, the fact that the finding (that there was no clinically useful benefit) was not properly presented.

You are assuming that there's a focus on one single issue but there's no evidence of that.

Beyond that, I don't have anything further to add at this point.

 

[Hutan]

You are assuming that there's a focus on one single issue but there's no evidence of that.

I'm not sure how you can say that Medfeb. There is evidence of that, and no evidence otherwise.

In Hilda's blog there are paragraphs about how things have changed with the recognition of PEM and the review is constantly referred to as 'outdated' rather than flawed. There is very little talk of other criticisms - a brief mention of 100 or so other faults but a qualification that consideration would not result in action for all of them and indeed that some were contradictory. Anyone reading that blog would come away thinking that the lack of consideration of PEM is the biggest issue with the review, and that it is because it is outdated.

MEAction's piece similarly focuses on the diagnostic criteria issue, as I mentioned upthread.

As far as I know, those are the only public written pieces arising from members of the writing team or IAG or their organisations.

I look forward to hearing what you think about the point about MAGENTA when you have the time. I think that example might show how a trial can have a diagnostic criteria where PEM is not mandatory, but can still produce evidence relevant to ME/CFS. Therefore, the use of a diagnostic criteria that does not require PEM does not make a study scientifically invalid or even irrelevant.

 

[Hutan]

If we want to use Magenta as new evidence of GET not only not working but causing subjective and objective physical and emotional harm (and we absolutely should because it is very powerful evidence) then we have to accept that we can get useful information from trials that don't use our favourite PEM requiring diagnostic criteria.

The psychobehavouralists are the only ones doing GET trials (and even they have mostly have moved on trials in Long covid etc). And they mostly don't look for PEM - they tend to want to minimise the response to exercise as being post-exertional fatigue and post-exertional muscle soreness. So, if we want evidence that GET doesn't work specifically in people who get PEM, I think we will be waiting a long time for good, dedicated, trial evidence.

 

[Utsikt]

But, I would absolutely want any future review of exercise therapy in ME/CFS to consider the MAGENTA study; it does tell us something useful about GET and about the harm that can be done by telling people that they can be better if they put the work in.

FYI, this phrasing makes it seem like you’re saying that the wider inclusion criteria is acceptable because you like the results and want to apply them to a narrower case.

What we want is irrelevant.

 

[Jonathan Edwards]

There are a couple of studies that looked at people referred to specialist centres with a suspicion of ME/CFS and around 50% did not meet ME/CFS criteria (it did not seem like solely a problem of the criteria). The non-ME/CFS group included conditions like depression, sleep disorders, overtraining syndrome, burnout etc.

Interesting data but these 'others' would presumably have been diagnosed as having these other conditions rather than CFS or ME/CFS. I don't think it suggests that people like Peter Denton White were managing many people as 'CFS' or 'CF' who did not in fact have ME/CFS.
They would not have recruited to PACE people they had identified as having depression instead.

 

[Jonathan Edwards]

I completely agree that poor trial design is a huge problem. But for me, patient selection is a critical part of trial design - that's exactly what defines "the group of people you are looking at."

But it doesn't @Medfeb.

If an anti-inflammatory analgesic is studied in large trials of joint pain do we then say that it should not be prescribed for rheumatoid arthritis because the trials did not require symmetrical joint swelling or a raised ESR? No, we use most drugs test on brand categories for all sorts of more restricted subsets.

If PACE had had a positive outcome based on adequate methodology it would have been gerrymandering to argue that it did not apply to ME/CFS because the criteria were not explicitly the same. Exactly the sort of gerrymandering the BPS people are accused of. Their defence of their criteria is entirely reasonable and it is a huge distraction that people are arguing against it.

 

[Utsikt]

The lack of use of a diagnostic criteria including PEM is an argument for there needing to be a new review relevant to ME/CFS. But it's not a reason for saying that the Larun et al review was unscientific. (Although of course there are many other reasons why it was unsound.)

I look forward to hearing what you think about the point about MAGENTA when you have the time. I think that example might show how a trial can have a diagnostic criteria where PEM is not mandatory, but can still produce evidence relevant to ME/CFS. Therefore, the use of a diagnostic criteria that does not require PEM does not make a study scientifically invalid or even irrelevant.

If we want to use Magenta as new evidence of GET not only not working but causing subjective and objective physical and emotional harm (and we absolutely should because it is very powerful evidence) then we have to accept that we can get useful information from trials that don't use our favourite PEM requiring diagnostic criteria.

Here’s another angle that might explain why there’s focus on the diagnostic criteria:

If you study a group that may or may not have PEM, then you can’t deduce that any positive results regarding effect or safety would apply to everyone with PEM. Why? Because PEM acts differently in relation to exertion, compared to non-PEM. So it’s reasonable and plausible to expect that a pure PEM study could yield different results.

MAGENTA is different. They showed that an exertion-based approach was harmful even for non-PEM patients. There is no reason to believe that PEM-patients would respond fundamentally different from the studied group.

No, we use most drugs test on brand categories for all sorts of more restricted subsets.

«This is how it’s usually done» is not a valid argument. That’s what the BPS lobby argues with regards to using subjective outcomes.

 

[Jonathan Edwards]

But the problem is when all of those people, regardless of what they have, are labelled ME/CFS.

Nobody is labelling people who do to conform to ME/CFS criteria as ME/CFS, any more than they are labelling people with joint pain by not RA, RA.

One very unfortunate by product of this sort of debate I see is that if there are indeed a lot of people with 'chronic fatigue syndrome' who do not conform to criteria for ME/CFS, as seems to be suggested, these people no longer have any NICE Guidelines for their care. That to me is pretty much a violation of human rights. Advocated for people with ME/CFS do not have the right to trample over the interests of people who do not!

Diagnostic pigeonholes can cause a lot of harm.

 

[Hutan]

FYI, this phrasing makes it seem like you’re saying that the wider inclusion criteria is acceptable because you like the results and want to apply them to a narrower case.

What we want is irrelevant.

Some people want to ignore all exercise therapy studies that don't have people diagnosed by a criteria that requires PEM. I'm pointing out that if they get what they want, we would have no recent trial evidence that I know of and would have to ignore what I believe is one of the strongest pieces of evidence that GET harms in multiple ways.

I have been consistently arguing that a wider inclusion criteria can tell us something useful, and Magenta is an example.

 

[Jonathan Edwards]

FYI, this phrasing makes it seem like you’re saying that the wider inclusion criteria is acceptable because you like the results and want to apply them to a narrower case.

They apply to the narrower case until proved otherwise by the nature of set theory @Utsikt. I thin you have still got this back to front - maybe because you are unhappy about the implications of getting it the right way around?!!

 

[Trish]

Edit: I haven't been able to read all the discussion.

The PACE trial used the Oxford criteria for CFS, which only requires chronic disabling fatigue, and explicitly includes post viral fatigue, fatigue accompanied (or caused) by mild to moderate anxiety and/or depression, and idiopathic chronic fatigue. Patients were required to report fatigue as their main symptom. If I recall correctly, a surprisingly high percentage were on antidepressant medication. They did subgroup analyses according to some different diagnostic criteria and found no significant difference in outcomes.

The results showed no long term benefit over standard medical care on subjective measures, no benefit on objective tests and not significantly better recovery rate or improvement rate when analysed by their own original protocol.

The researchers could only find small, not clinically significant improvements on subjective questionnaires.

So basically, even when including the broadest possible set of patients with a wide range of possible causes of fatigue, GET doesn't work.

The problem with PACE was not so much indurectness as dishonest researchers who switched outcome measures after the trial had started failed to properly report objective and long term outcomes etc.

The biggest problem that enabled them to claim success was reliance on short term subjective outcomes, not indirectness.

If the researchers had analysed the outcomes honestly, and used the objective and long term data, they and the Cochrane reviewers would conclude that exercise therapy is not an effective treatment for any fatiguing condition, including ME/CFS.
_________

However I also think indirectness can be a problem in some scenarios.

A thought experiment:

Half the participants had fatigue due to mild to moderate depression, and got fitter and less fatigued with GET and antidepressants,
a quarter had short term post vital fatigue and recovered naturally during the year of the trial and got fitter and no longer fatigued,
a quarter had ME/CFS with PEM, some of whom stuck to their GET and got sicker, and the rest realised it was harming them and either dropped out or pretended to be complying.

In that scenario, a GET trial would have been a roaring success and we'd all still be subjected to GET.

So yes, I think indirectness is a potential problem, but as it turned out that problem was outweighed by the one thing the researchers and reviewers have held on to in order to claim effectiveness, and that is not indirectness but reliance on short term subjective outcomes on unblinded trials.

Just that one aspect of unscientific methodology should have made PACE and the Larun review unable to claim any benefits from exercise therapy for anyone with chronic fatigue, including ME/CFS.

The position we're in today with Cochrane is not about indirectness, it's about researchers not accepting their trials of GET failed, and using bad science and political clout to push their lies.

And it's down to editors too weak to stand up to bullies with reputational and/ or financial conflicts of interest, and too lacking in scientific understanding to see that the evidence is too flawed to be used.

 

[ME/CFS Skeptic]

What about the argument that strictly defined ME/CFS with clear PEM would have been more likely to have shown adverse effects in GET/CBT trials. That might overshadow any positive results and explain why the BPS people preferred to focus on broad criteria with unexplained fatigue.

As Jonathan said, if you follow their reasoning about deconditioning and false illness beliefs, the more severe ME/CFS patients would have been the best study subjects who should improve the most. Instead, BPS'ers preferred to focus on patients with unexplained fatigue using the Oxford criteria.

 

[Utsikt]

I thin you have still got this back to front - maybe because you are unhappy about the implications of getting it the right way around?!!

I don’t care what the result of my thinking is. I care about understanding the underlying mechanisms. And I currently don’t understand it.

I stated earlier in the thread that I don’t have an opinion because I recognize that I don’t understand the full picture yet. Besides, the conclusion would still be that GET shouldn’t be recommended on the basis of the studies in the review. Either due to a lack of evidence, or due to evidence that it doesn’t work/is harmful.

So the implication in this case is irrelevant.

 

[Jonathan Edwards]

However I also think indirectness can be a problem in some scenarios.

That's right @Trish, but, as discussed with @Utsikt, examples like this are what I call 'special pleading' and although they may be valid, if we don't know they are, the default probability estimation remains. Basically NICE cannot alter a recommendation on the basis of 'what if we think of this scenario' arguments - although it could certainly raise a caution.

 

[Utsikt]

I just want to be very clear: I am not claiming that indirectness is the main issue with PACE or the Cochrane review.

I’m also not claiming that arguing about indirectness is the strategic approach from the different parties involved.

What I am doing, is tryingy to understand why indirectness is not a fatal flaw to the study design. I want to learn. I want to understand. And I’m not arguing against anyone here because I disagree, but because I don’t understand. And that’s provably a me-problem.

 

[ME/CFS Skeptic]

They would not have recruited to PACE people they had identified as having depression instead.

Not so sure because their concept of ME/CFS might differ and be more inclusive than ours.

I've heard some interesting stories from ME/CFS patients in Belgium. In the early 2000s that country financed CFS centres where patients were treated with GET/CBT. You had to meet CFS criteria (Fukuda) in order to be eligible. Some of the treatments sessions or the initiation was done in group form where patients could meet.

Several ME/CFS patients reported that this was frustrating because many other participants did not have ME/CFS as we would see it today: with PEM, OI and other characteristic symptoms such as light and sound sensitivity.

It included people who were fatigued, depressed, or burned-out, who were much more active and didn't get severe crashes. Many thought their symptoms were stress-related. It also seem to include people who's primary issue was a sleep disorder or pain which in turn also caused fatigue which led them to get a CFS diagnosis.

Later the physicians at the CFS centre admitted they were initially too loose in applying CFS criteria, for example by including too many people with sleep disorders. This was because everything was new for them, etc.

I suspect BPS'ers might have done something similar in their trials to avoid too many ME/CFS patients with clear PEM who might worsen or drop out during treatment.

 

[Jonathan Edwards]

What I am doing, is tryingy to understand why indirectness is not a fatal flaw to the study design. I want to learn. I want to understand. And I’m not arguing against anyone here because I disagree, but because I don’t understand. And that’s provably a me-problem.

I appreciate that. And I think it is useful for us all to realise how hard it can be to see what one might call mathematical necessities - like in set theory. But I think the example of using painkillers for RA on the basis of them working for joint pain is a pretty simple demonstration that 'indirectness' need not be a problem at all.

Sometimes these things are hard to follow and suddenly become clear. I love Roger Penrose's explanation for why 3x4 = 4x3. You might say why should it have to? How can we be sure that 567x439=439x567. As he points out the answer is that if you lay beads out in arrays of 3 by 4 you have by definition laid them out in arrays of 4 by 3 and the same for 567x439.

Extrapolating to subsets is legitimate by default in a probabilistic analysis. Extrapolation to supersets is the problem.

 

[Jonathan Edwards]

Not so sure because their concept of ME/CFS might differ and be more inclusive than ours.

It might indeed, and it sounds as if they had to scrape the barrel to recruit. I am just saying, though, that those figures for people referred with suspected ME/CFS may not be directly relevant if they were actually given different diagnoses from CFS.

 

[Peter T]

There are still no comments appearing on the MEAction blog post on Cochrane’s abandoning the new exercise review and their redating of the old review. I discovered that adding a new comment enabled seeing the old comments in moderation, so added a new comment just now in order to access what I had already posted:

1. PETER TREWHITT
   JANUARY 27, 2025 AT 10:51 PM
   Even if Cochrane now fails to undertake a new review of exercise based intervention in ME/CFS it is imperative that action is taken to prevent more harms arising from their old review. Over sixty patient organisations have co-signed a letter demanding that Larun et al (2019 now misleadingly dated 2024) be withdrawn on the grounds that it failed to address harms, ignored evidence from objective outcome measures but rather used only subjective outcomes in open labelled trials, relied on data using out of date definitions of ME/CFS and the most recent data it includes is from the now discredited PACE study and that it contravenes Cochrane’s own guidelines for their reviews.
   
   More recent analysis by NICE in the UK and the US CDC drew the opposite conclusion that graded exercise therapy should be contra indicated for our condition, even though Cochrane’s slight of hand now has this review dated after the publication of these two accounts.
   
   Further well over 14,000 people have signed a petition (see https://chng.it/XqXkNTMQzH ) demanding this damaging review be withdrawn). The S4ME updates on the petition page provide a devastating record of Cochrane’s failure to engage with the problems of this review and their failure to observe their own complaints procedures.
   
   The discontinued new review processes was beset by delays arising from complaints and pressures from un named supporters of Larun et al, and increasing it seems that the current chaos is the results of pressure from interested parties defending the old flawed review regardless of good science or Cochrane’s own procedures and standards.
   
   Reply
   Your comment is awaiting moderation.
   
   
2. 
   
   PETER TREWHITT
   JANUARY 28, 2025 AT 2:43 AM
   Are we seeing with Cochrane’s current disaster over this ME/CFS exercise review what Brian Hughes termed ‘eminence based’ medicine rather than evidence based medicine or perhaps even ‘science by cronyism’?
   
   Reply
   Your comment is awaiting moderation.
   
   
3. 
   
   PETER TREWHITT
   JANUARY 30, 2025 AT 12:28 AM
   I tried to post a couple of times over the last few days but as they have not appeared it may be that I have done something wrong. Easily done with my brain fog.
   
   I am grateful for the work MEAction are putting it holding Cochrane to account, but am interested in the reasons for limiting requests to an editorial note rather than a full retraction.
   
   Reply
   Your comment is awaiting moderation.

Am posting them here as a record in case they end up getting lost indefinitely. With the Viriology Blog some time ago comments would occasionally get eaten up by the software, though they seem to have now sorted whatever the problem was.

 

[Utsikt]

@Jonathan Edwards I appreciate your patience!

If you say that you study mammals, but only study rats, can you extrapolate those findings to humans because the humans are a subset of mammals?

If yes, why do we need human drug trials?