Discussion in 'PsychoSocial ME/CFS Research' started by Cheshire, Mar 15, 2019.
Monty Python's "Dead Parrot Sketch" ???
I can't face reading this at present, will need to build up to it.
What is interesting is the timing of this article indicating that the PACE appologists are currently orchestrating a planned and coordinated defence of the indefensible. Also it may be good that they have published it now as it should provide opportunity to demonstrate that they continue to fail to respond to substantive criticism.
This explanation makes ZERO sense. Like, what on earth does this even mean?
No Peter White...
EDIT: the journal received the first manuscript in July 2018, so the bulk of this was probably written, almost a year ago.
ANOTHER EDIT: This journal has Open Peer Review. So normally you can see what reviewers said about the manuscript. There isn't much information though on the website, except for a short comment by one of the reviewers, Francesco Pagnini. It doesn't explain why the PACE-authors had to resubmit their manuscript three times...
Just to quote one of their feeble justifications:
So they admit that they think their recovery rates are better because they 'prefer' their definition. And even more damning, they think their results are right because they fit with 'the literature' and 'our clinical experience'.
That sounds like Wessely's ship analogy of changing course to make sure you reach the correct destination.
And they don't understand that if something is not statistically significant, you can't use the non significant differences to justify anything.
I haven't read it all, but it's not a scientific analysis, it's pages of self justification of the indefensible on very shaky and unscientific grounds.
It means they are in denial about the fact that the findings reported in PACE are entirely consistent with placebo effects. It could also mean that they're just trying to win the debate by producing a lot of bullshit hoping that it will confuse and convince naive readers.
As far as I know, they did this because it was planned in the original protocol. The PACE authors appear to consider this "excessive" because it doesn't give the results they were expecting (the original protocol is very optimistic about finding large treatment effects that would easily withstand such corrections).
and then there were 3........
eta: or 4 if you include the one behind the curtain
It's also revealing that they do not describe in any detail how they defined recovery, presumably because if they did, anyone with some familiary of the scales used would immediately see that there is a problem:
So here's recovery as originally defined (left) versus how it was defined in the publications (right)
A score of 60 on the SF-36 physical function scale is typical for an 80 year old or younger people with severe chronic illness like multiple sclerosis and heart failure.
When the PACE authors write that
I only wonder what that says about the literature and their clinical experience. Also, researchers aren't supposed to fit their results to the previous results or their clinical experience. They're supposed to find out the truth.
We should remember that the definition used by Wilshire is their definition as they published in the protocol. To claim they prefer theirs definition because it fits their expectations does suggest they manipulated the definition until they got the desired result.
Let's just look at their conclusions again:
Strictly speaking, addition of CBT and GET does moderately improve the outcomes they used, whichever analysis you use. Whether that is "safe", depends on how you define "safe" (they don't say how they define it, and only mention it once in the entire rebuttal).
But the conclusions weren't necessarily the problem. Apart from a few niggles (use of difference of means rather than median difference), their analysis isn't necessarily a problem.
It was much more fundamental than that. It was the methods they used. It was the outcome measures themselves (not just the cut-offs). It was their patient selection criteria. It was their prior assumptions about the condition. It was their dismissal of all objective measures. It was the fact that there was no longer any difference at long-term follow-up...
They still haven't addressed any of those things.
Gaslighting is the last recourse of the desperate
3 times is quite usual I think even just for clearer wording, typos and formatting.
Only!? So the scoring method is of no consequence then. Maybe we could modify all manner of scoring systems so that 13% of people are deemed improved at something even when they have not.
I'm not sure that a difference of medians would be valid for the CFQ.
I do think there is a huge statistical naivety around the use of the mean and SD with the sf36 as I see no way that the scores would be equidistant on a definition of physical function. I do think this is an important issue and one that needs to be aired as it seems such a common issue in so much research.
But you are right in terms of the biggest issue being that the outcome measures are not robust as for example sf36 is a measure of perception of physical function not physical function itself therefore you can't test if function or perception has changed and interventions are aimed at changing perception hence the experiment is not worth doing.
The whole point being to prevent subsequent cherry picking of the methodology etc. once you have already run part of a trial and gained insights to how it is going. Especially in a fully unblinded trial.
Separate names with a comma.