A general thread on the PACE trial!

https://twitter.com/user/status/1054642332887330816

 

https://twitter.com/user/status/1054730672059891712

https://twitter.com/user/status/1054739096000778240

https://twitter.com/user/status/1054807995425677313

 

https://twitter.com/user/status/1054291418674614272


LOL at Sharpe saying 'debate is good'. A few months back he was endorsing the view that debate in the House of Lords was a form of harassment, and claiming that exactly the same thing was now happening with Monaghan's debates.

 

I've not looked at Sharpes twitter feed in ages, but he's really not good a discussing things, is he?

Too many odd things to be worth posting them all, but thought this one was of a bit of interest:

https://twitter.com/user/status/1054359845833490439


What about the Science Media Centre? eg:

http://www.sciencemediacentre.org/chronic-fatigue-syndrome-unravelling-the-controversy/

 

@dave30th @Eagle . Perhaps something that needs some research and communication before the next Westminster debate?

 

Suppression of those who you disagree with is really not.

That is presumably why it is not right to send threatening emails or other forms of communication to people's employers.

 

have been revisiting old tweets of SW:

https://twitter.com/user/status/661824534941736960

https://twitter.com/user/status/661842335085498368

https://twitter.com/user/status/661967018028068868

https://twitter.com/user/status/662246997001379840

https://twitter.com/user/status/662252640949116928

 

The results are reliable, but they show no long-term benefit. Up is down. Blue is hot.

Or in Sharpe's wizened garbled words: "Well that what was what was done"

 

Wessely says

But patients in the control group received far less treatment than the other arms. How can he say that the control group controlled for nonspecific effects? Is my understanding of the term wrong?

 

He is presuming that each group in some way "controls" for the other groups. They even tried to build that into the analysis (in the protocol). Actually, it's more evidence they wanted the APT group to be the "control" group.

But they didn't reproduce "normal practice", because normal practice is to combine CBT and GET.

 

Thanks for posting this revealing tweet by the weasel.

 

I've been trying to get my head round this issue of controls with psychological therapies, and getting confused. As I understand it, a control group should be identical to intervention groups except for the 'active ingredient' in the intervention itself'; any differences in outcomes then have a good probability of being due to the intervention and nothing else.

But what happens when trialling psychological interventions? Where the intervention is itself about motivating people and encouraging better self esteem etc., etc? What sort of control do you need for such a trial? Do you give the control group the same level of attention of non-motivational attention, which sounds a distinctly dubious control to me. This has been bugging me and I'd like to understand it better.

 

The best way around all this is just to drop all the objective measures.

 

There is a Catch22 involved and a very subtle one. The real problem is that the effect may depend on the therapist believing in the treatment. So you need to compare different treatments where the therapists believe equally in their effectiveness. Maybe the comparison of CBT to GET is close to that but the result is no difference so that does notes us either is effective. You really need a bogus treatment that therapists believe in. The Lightning Process gets close to that. And it had an effect on subjective outcomes. So everything is pointing to the idea that it does not really matter what the treatment is as long as the therapist believes in it.

But then you have the problem that the effect in these trials may only occur in trials - where the belief of the therapist induces a trial subject to take on a positively biased role - as a trial subject. So the results tell you nothing about what would happen in ordinary practice.

What people doing trials often forget is that you have to put your wide angled glasses on and think about what happens in real life. Sticking to some rules in a book is no good. In real life all sorts of things affect what happens - and they all have to be controlled for in different ways.

 

One of the major problems with PACE was the complexity of what it was trying to do. It was simultaneously trying to test 3 different therapies and control for all their 'active ingredients' against one another. There is no way to adequately 'control' such a trial. So when MS says it was 'just a trial', he is wrong. When he says that each intervention 'had its own model', he is admitting that it wasn't (and couldn't be) adequately controlled. And when your intervention and outcome measures are so closely linked as to essentially be an education program in how to fill in questionnaires to get the desired result, you definitely can't control it properly.

I'm now beginning to wonder whether they deliberately made it so complex so that if they couldn't understand what was going on, no-one else was likely to either. Oops.

 

I have only managed to find a small portion of the book that Wessely co-authored:
Clinical Trials in Psychiatry. As you would expect, a lot is his usual historical stuff.
But there is this bit which I thought may be of interest:

 

As ever, he is rewriting the evidence to suit, so that folks can say, "Oh, this textbook says it's all OK if I do it like this."
Just no!

Blinding of surgery can be tricky, but there are ways round it. Sham procedures, for example. But he doesn't mention them.

Why would using a blinded evaluator help if you are recording said subjective responses using a paper questionnaire?

The main problem with blinding psychological trials is that the patient absolutely knows what they are getting, and so does the therapist. The assessment is a small factor in comparison to the major biases introduced by the main lack of blinding.

The last bit, he's just hedging.

It's all simply dreadful!

 

I suppose that counts as:

'Two legs are probably OK as long as you have another two in the air - so lets say two legs good then.'

 

I was going to add, that's why 'blinding' and 'placebo control' go together 'like chicken and... another chicken'
(for all you Buffy fans out there).

 

@Sly Saint What was the page number? I might find myself having to quote that bit.