Beyond total treatment effects in randomised controlled trials: Baseline measurement of intermediate outcomes needed to reduce.., 2018, Dunn et al

Andy

Retired committee member
One for our more technically minded members I think. They are coming up with different ways to analyse trial results and they apply them to the FINE trial, which is why I've posted it.

Beyond total treatment effects in randomised controlled trials: Baseline measurement of intermediate outcomes needed to reduce confounding in mediation investigations

Abstract
Background:

Random allocation avoids confounding bias when estimating the average treatment effect. For continuous outcomes measured at post-treatment as well as prior to randomisation (baseline), analyses based on (A) post-treatment outcome alone, (B) change scores over the treatment phase or (C) conditioning on baseline values (analysis of covariance) provide unbiased estimators of the average treatment effect. The decision to include baseline values of the clinical outcome in the analysis is based on precision arguments, with analysis of covariance known to be most precise. Investigators increasingly carry out explanatory analyses to decompose total treatment effects into components that are mediated by an intermediate continuous outcome and a non-mediated part. Traditional mediation analysis might be performed based on (A) post-treatment values of the intermediate and clinical outcomes alone, (B) respective change scores or (C) conditioning on baseline measures of both intermediate and clinical outcomes.

Methods:
Using causal diagrams and Monte Carlo simulation, we investigated the performance of the three competing mediation approaches. We considered a data generating model that included three possible confounding processes involving baseline variables: The first two processes modelled baseline measures of the clinical variable or the intermediate variable as common causes of post-treatment measures of these two variables. The third process allowed the two baseline variables themselves to be correlated due to past common causes. We compared the analysis models implied by the competing mediation approaches with this data generating model to hypothesise likely biases in estimators, and tested these in a simulation study. We applied the methods to a randomised trial of pragmatic rehabilitation in patients with chronic fatigue syndrome, which examined the role of limiting activities as a mediator.

Results:
Estimates of causal mediation effects derived by approach (A) will be biased if one of the three processes involving baseline measures of intermediate or clinical outcomes is operating. Necessary assumptions for the change score approach (B) to provide unbiased estimates under either process include the independence of baseline measures and change scores of the intermediate variable. Finally, estimates provided by the analysis of covariance approach (C) were found to be unbiased under all the three processes considered here. When applied to the example, there was evidence of mediation under all methods but the estimate of the indirect effect depended on the approach used with the proportion mediated varying from 57% to 86%.

Conclusion:
Trialists planning mediation analyses should measure baseline values of putative mediators as well as of continuous clinical outcomes. An analysis of covariance approach is recommended to avoid potential biases due to confounding processes involving baseline measures of intermediate or clinical outcomes, and not simply for increased precision.
Open access at http://journals.sagepub.com/doi/10.1177/1740774518760300
 
Saw they thank PACE's Kimberly Goldsmith:

The authors thank Cedric Ginestet, Paul Clarke, Kim Goldsmith, Andrew Pickles and Ian White for their contributions and suggestions, and Alison Wearden and the FINE trial team for their permission to use the FINE data. Trial registration number and register: International Standard Randomised Controlled Trial Number (IRCTN74156610).
 
It would help if they wrote something comprehensible. I cannot understand what they are wanting to say.
Perhaps the BPS brigade are getting fed up with patients, advocates & other non-supportive scientists like yourself understanding their work seemingly better than they do themselves & critiquing it in a 'vexatious' manner. So they decided to make it incomprehensible to anyone but those in the BPS clique in an attempt to make everyone think they're clever than they are and shut up.
 
Perhaps the BPS brigade are getting fed up with patients, advocates & other non-supportive scientists like yourself understanding their work seemingly better than they do themselves & critiquing it in a 'vexatious' manner. So they decided to make it incomprehensible to anyone but those in the BPS clique in an attempt to make everyone think they're clever than they are and shut up.

I think this is an old trick and I think it has more to do with the people within the clique wanting to reassure each other how clever they are rather than being at all bothered with outsiders understanding or not.

But the BPS lot might be right to get fed up with us understanding them better than they do. I suddenly realised yesterday that PACE actually disproves the BPS model in a remarkably elegant way. It was not such bad science after all because it managed to refute its own hypothesis - which is what you are supposed to do.
 
I suddenly realised yesterday that PACE actually disproves the BPS model in a remarkably elegant way. It was not such bad science after all because it managed to refute its own hypothesis - which is what you are supposed to do.

That's the only plus side I've been able to see about PACE. Not only did it demonstrate conclusively that GET and CBT are useless for people with ME/CFS, it also demonstrated that they are useless for the much broader category of patients with Oxford defined 'six months unexplained fatigue', which covers pretty much anyone with so called 'Medically Unexplained Symptoms', since most of those lumped into that category have 'unexplained' fatigue as one of their main symptoms. If the treatments based on the BPS model fail, the model fails too.

And since MUS is the broad category that Sharpe, Chalder et al are now pushing hard to apply the BPS model and treatments to with their roll out of MUS IAPT clinics, it pulls the rug from under the whole thing. Similarly it pulls the rug from under the BPS based approach of the DWP and health insurance industry supported by Wessely, Aylward, White et al.

I have not been at all surprised that these individuals along with the SMC have been so determined to try to block data release and to undermine any paper based on that data that demonstrates their model has failed.

It also explains why they so determinedly bury the null results of the FINE trial, which was run in parallel with PACE to cover the more severely affected, home bound patients to demonstrate (they thought) that the BPS model applied across all severity levels.

In fact what they have inadvertently done is dump the whole branch of psychiatry - psychosomatic medicine which is based on BPS - in the dustbin.

The fact that they are now clinging to ridiculous justifications like the impact factor of the journals they published PACE in, demonstrates how desperate they are.

I'm not sure whether that's what you meant, @Jonathan Edwards. Do elucidate if I've gone off on a different track from what you meant.
 
I'm not sure whether that's what you meant, @Jonathan Edwards. Do elucidate if I've gone off on a different track from what you meant.

Yes, but I found a new argument that makes it clear that PACE actually proves the opposite of the BPS theory, not just fails to corroborate it. Or at least it would if PACE did not, through a similar argument, ensure it did not test the theory on those to whom it was supposed to apply.

I will explain in due course but not just yet.
 
Yes, but I found a new argument that makes it clear that PACE actually proves the opposite of the BPS theory, not just fails to corroborate it. Or at least it would if PACE did not, through a similar argument, ensure it did not test the theory on those to whom it was supposed to apply.

I will explain in due course but not just yet.

I love your tantalising hints...
 
Nope nope nope nope nope nope!

When your entire mediation analysis is based on a non-absolute measure (like CFQ), then there really is no point at all. It's not going to tell you anything.

First rule of stats is to understand your variables, and to understand what they are doing, what they measure, how they behave, are they measuring what you think they are measuring. If this hasn't been done, then anything else is completely pointless and meaningless.
 
Thanks so much @Lucibee. I had a feeling when I read the abstract that they were playing clever clever statistician games without understanding the data they were inputting into their analysis. It's great to have a medical statistician here who can see just what nonsense this is.
 
I confess I'm really not on top of mediation analysis, and their explanation is very poor. But, from what I understand, the underlying assumptions are fairly crucial, and they seem to have made quite a lot of them that are unlikely to be true.
 
Last edited:
Yes, but I found a new argument that makes it clear that PACE actually proves the opposite of the BPS theory, not just fails to corroborate it. Or at least it would if PACE did not, through a similar argument, ensure it did not test the theory on those to whom it was supposed to apply.
It's always seemed to me a bit ironic that the PACE team crow about the success of the trial since, taking their own reported results at face value, they have shown that they can in fact influence behaviour and psychology to the extent of making people think they are better, but that merely thinking it was so had no significant impact on the actual physical health and functional capability of the participants.

Thus changing the mind did not change the body, ME was not reversed using successfully applied BPS treatments, which is not only the opposite of what they set out to prove regarding ME, but also undermines the whole BPS hypothesis.

Will be interested to hear your new argument @Jonathan Edwards
 
It was not such bad science after all because it managed to refute its own hypothesis - which is what you are supposed to do.
Yes, the really bad science was the authors' refusal to acknowledge their trial showed their hypothesis to be wrong, and then basically cheated to make it look like their trial had successfully proved them right. I think down the line the PACE trial will be taught not just as an example of bad science, but as an example of deeply flawed scientists ... so flawed they do not deserve to be called scientists.

Edit: Though as I read further down the thread, I realise I've jumped the gun a bit, as you are looking into something more with PACE :).
 
Last edited:
Thus changing the mind did not change the body, ME was not reversed using successfully applied BPS treatments, which is not only the opposite of what they set out to prove regarding ME, but also undermines the whole BPS hypothesis.

That's the gist but the twist is to rephrase 'changing the mind' in their very own terms so that you can walk round the circular argument in strict bps speak without slipping into any dreadful Cartesian dualism - ahem!
 
Thus changing the mind did not change the body, ME was not reversed using successfully applied BPS treatments, which is not only the opposite of what they set out to prove regarding ME, but also undermines the whole BPS hypothesis.
But of course in their eyes it was only the mind that needed changing, and any proofs to the contrary they suppressed/ignored.
 
That's the gist but the twist is to rephrase 'changing the mind' in their very own terms so that you can walk round the circular argument in strict bps speak without slipping into any dreadful Cartesian dualism - ahem!
I'll play:

The BPS model predicts that any illness with an absence of underlying pathology can be reversed or cured through the therapeutic alteration of the dysfunctional cognitions and behaviours of the patient, which are mooted to be the true cause and/or perpetuating factors of the disability.

ME/CFS is considered to be the example par excellence of this type of health condition, being an illness believed in this model to be characterised by fatigue with no ongoing biological cause, but where "the symptoms and disability of CFS are perpetuated predominantly by dysfunctional illness beliefs and coping behaviours. These beliefs and behaviours interact with the patient's emotional and physiological state and interpersonal situation to form self-perpetuating vicious circles of fatigue and disability... The patient is encouraged to think of the illness as 'real but reversible by his or her own efforts' rather than (as many patients do) as a fixed unalterable disease" (Wessley).

Hence the PACE trial, which sought to test whether the disability associated with ME/CFS could be reversed by means of the standard treatments recommended by the BPS model, specifically CBT and GET...

The results clearly show that the patients' disabling cognitions (and behaviours?) were successfully altered in the direction of health and even, for 20%, recovery (as defined in the trial protocol). However, the levels of physical disability and social exclusion experienced by the patients remained unchanged (as per...the various measures). Thus, correcting the dysfunctional beliefs and other maladaptive psychological aspects associated with the condition did not result in an increase in functional capacity, indicating that the biological element of the illness was unaffected - unimproved - by the psychological and behavioural interventions. The study hypothesis is thus disproven, with significant implications for the underlying BPS model...
 
Back
Top Bottom