Graded exercise therapy for ME/CFS is not effective and unsafe. Re-analysis of a Cochrane review (2018) Health Psychology / Vink

This may be me restating the obvious but I am trying to get my head around the issue of deterioration following GET.

If the studies collected by Larun and Vink are taken as showing evidence of efficacy of GET then the reasoning seems to be this:

Physiological illness in ME is perpetuated by a attitude of mind that prevents return to health and this attitude can be improved by GET - as indicated by reports of a more positive attitude to health status. This improvement cannot just be reversal of deconditioning because deconditioning of its own does not produce ME symptoms, although if people with ME are to some extent deconditioned one might expect to see objective reversal of this with GET (I think there is in one study). The studies do not show objective evidence of reduction in disability so improvement in attitude to health status is taken as a valid index of improved physiology.

But if it is accepted that GET can make a physiological change, and attitude to health status is a reliable index of this then it has to be accepted that GET may be followed by a genuine adverse physiological change. Any deterioration in 'attitude to health status' must be taken equally seriously as an index of worsening physiology. If accounts of worsening are to be dismissed as merely a perverse attitude change then there is just as much reason to dismiss the positive attitude changes measured in the studies as not reflecting genuine physiological improvement. You cannot have it both ways.

And if we look at the reports of attitude changes the negative ones are much more plausible than the positive ones. The reports of improvements in trials are very consistent with the sort of positive reporting bias characteristic of trials. But the reports of deterioration often take the form of a person willingly taking part in a treatment hoping to improve on it and then finding instead that they are much worse. The clear impression is that the BPS view is right in that exertion can genuinely impact physiology long term (and who knows maybe they are right that it is the mental exertion that is crucial) but that that can be in a seriously adverse way.

Or to put it simply; if the studies are taken as valid evidence of improvement the reports of serious deterioration must be considered equally valid by the same theory.

 

Do you think that reports of deterioration must be considered valid even if we were to dismiss the reports of improvement as probably due to bias?

I do not think that taking this position means trying to have it both ways. Everyone involved in a trial is biased towards positive results. If patients believed GET would harm them, they would not enroll in the clinical trial, or drop out once they were assigned to a GET arm. The typical report of harm following GET is by a patient that was unaware of risks or underestimated them.

 

It is very hard to prove that a period of treatment caused a subsequent disability. All I think one can argue is that deteriorations after treatment cannot be ignored. In line with standard practice for adverse events they have to be given priority and are not subject to the statistical requirements for showing a benefit.

 

Just as a matter of detail, which does not affect the overall conclusions, was not the attitude of mind originally to be altered by the CBT, the GET being there to overcome the deconditioning?

I have just gone back to the Wessely et al paper Cognitive Behavioural Management of PVFS from 1991. It says that, just under a half of patients refused treatment. So from the start, whilst proclaiming efficacy, they were seemingly only conducting tests on approximately the half who thought there was any prospect of benefitting and then founding their religious movement upon the findings. The "soma" in psychosomatic medicine seems to have more to do with Aldous Huxley than any thing medical.

EDIT Added question mark

 

My understanding is that the GET therapists effectively used CBT to explain to the patients why the GET will work and the CBT therapists told the patients that exercising will get them into a better place. That is to say they were pretty much the same thing.

 

I think that will be correct. It comes down to whether CBT and GET are regarded as separate concepts, as originally planned, or whether when discussing GET we mean some conflation of the two in some imprecise fashion.

 

The snag here is delayed PEM, in my case it can be up to 3 days. So I would ideally need a three day gap. Typically, if I haven't been hit by payback by the third day, I've gotten away with it.

It took a very long time for me to figure that out and so I was unwittingly putting myself in compound PEM. In that state it is almost impossible because every tiny exertion causes payback.

 

What do you mean by this?

We have simply taken the studies from Larun and reanalysed those and I think the reanalysis makes it pretty clear that they do not show evidence of efficacy of the GET

 

I hope you have not misread my text, @Mark Vink. I agree, perhaps I should have said collected by Larun but actually you brought back in a study she excluded I think? I am not in the slightest suggesting that you took the studies as evidence of efficacy. My sentence says 'If the studies... are taken as showing ... efficacy ... then....

I think your analysis is excellent. It is uncannily similar to a referee's report I sent except that yours goes in to much more detail.

 

Hi Dolphin, thank you for all your comments; this part is from Wearden et al. (1998), not from their other paper in the Review; see attached picture

 

@Jonathan Edwards what do you make of the normal VO2max scores in two studies (as mentioned in the review by Mark Vink)? And of some other nearly normal scores, the exact type escapes me at the moment.

It sounds like these studies primarily recruit patients that are barely ill and may not meet internationally used definitions of CFS.

 

Thank you for clarifying that; yes I was sure I misread that that's why I asked. We also included Núñez et al. (2011) which they had excluded.

Last but not least thank you for the great compliment

 

Not caught up with this thread today, but wanted to just jot something down now, even though it is almost certainly repeating what others will have said before.

To me it comes down to two fundamentals regarding the studies that were reviewed, and by inference therefore the review itself, being as it supported those studies' findings:

1. The studies/review shows no proof that people correctly diagnosed with ME improve with exercise.
2. The studies/review shows no proof that people correctly diagnosed with ME do not deteriorate with exercise.

I particularly think it is important to distinguish between 'deterioration' and 'harm', because the impression I get is that BPS researchers somehow manage to avoid categorising deterioration as a form of harm; thereby claiming exercise caused no harm. I think us trying to prove harm might be might more tricky that proving deterioration; though to prove deterioration with statistical significance almost certainly requires reliable objective outcome measures.

Also, when MS bites back against the need for objective measures, reasserting his belief in the adequacy of subjective measures because "that is how the illness is defined", this really tells us how MS and Co totally misunderstand what ME/CFS is and what it is like for people living with it. That may well be how their mind set defines it, but it's an invalid definition, as is their mind set. It's a very physical condition, which has both objective as well as subjective consequences; they erroneously choose to ignore the objective ones.

 

If the word 'deterioration' had been used instead of 'harm' (and it were not EC answering the question), what would the correct answer then have been I wonder?

The inevitable question being: What is the most likely reason these 31% of people dropped out?

 

It could be the first excuse they came up with for this methodological weakness. Remember, they originally planned to objectively measure activity levels in PACE and then dropped that when they realized the results would likely be null. That contradicts the idea that they're dealing with something that's beyond objective testing. So this seems like rationalization for poor methodology.

 

What sort of nonsense is it to say that the best evidence you can ever get is a systematic review? You cannot generalise. It may be true, but any particular systematic review may be worthless. It all depends of the quality of the papers included in the review and their relevance and the objectives of each particular piece of evidence.

 

Maybe every time MS tweets we should ask him via Twitter what people's reasons were. Keep reminding his followers of the dodgy cherry picked data

 

[My bold]


After pondering this outcome measure, I don't think the definition of improvement itself is so daft, but I do think the baseline entry level is; the two are distinct, and it is important to not conflate them. (I hasten to emphasise I'm not suggesting anyone here is - just a general comment!). If you do accept the baseline entry criterion: if your score at baseline for PF is poor, then your score has to improve significantly; if your score for PF at baseline is near normal, then it does not have to improve much at all, which makes sense insofar as an improvement measure goes. But the baseline entry criterion raises the question: what was the condition that improvement was being measured for.

It seems dubious that an entry criterion of virtually normal PF is valid for ME/CFS. It raises the usual issue of whether they were really looking to only recruit people with ME/CFS.

Were any/many participants above a PF score of, say, 20 at baseline?

 

We don’t know. As with most trials, we don’t have the individual data. Though we do have it for the FINE Trial (Wearden et al. 2010) and the PACE Trial (White et al., 2011).

 

Yes, but the Wearden et al. (1998) extract you quote doesn't say whether the threshold of 4 more is based on bimodal scoring (0-14) or Likert scoring (0-42) while your paper says it refers to Likert scoring/0-42.

It doesn't make much sense to use 4 as a threshold for Likert scoring as healthy people would score 14 on average, so everyone in the population basically, would be counted as a fatigue case. So I presume the threshold of 4+ refers to bimodal scoring.