-
New Forum Software Installed (Still a few issues but reopening the forum) More details.
A general thread on the PACE trial!
- Thread starter Esther12
- Start date
Two interesting posts @Lucibee .
I'd refer to 20040927 TSC as well, p8/9:
King's (where Wessely and Chalder were/are based) was seen as a CBT centre and also one which 'has a higher portion of referral with mis-diagnoses or previous treatment'.
Snow Leopard
Senior Member (Voting Rights)
It might explain more than we realise. Clinic reputation may very well make a difference, at least in the age of the internet and/or peer support groups.
Yes, that was my thought as well.
And the acknowledgement that KCL is seen as a centre for CBT by patients does suggest that it has been/was getting a biased sample: getting patients who went there looking for CBT. (And equally, odd though it seems to us, patients were going to Barts/QMUL for GET.)
The same could also be said perhaps for the higher numbers of patients there who were screened out. KCL was attracting patients who didn't have 'CFS/ME'.
I also think it's interesting that they were anticipating these problems at KCL before the trial started recruiting. The investigators knew of these tendencies
Barry
Senior Member (Voting Rights)
For a trial such as PACE what is the formal process for randomisation of participants, and how is it recorded and audited? At what stage is it done? Are the processes of recruitment and randomisation going on in parallel? Or can randomisation only be done once all participants have been selected? And is all recruitment halted once randomisation has been carried out. Just trying to understand if any "recruitment recycling" can ever happen, if participants drop out due to not getting the trial arm they hoped for. Presumably that should be impossible.
Snow Leopard
Senior Member (Voting Rights)
Lancet manuscript said:
For a general overview of randomisation techniques, you may want to read this:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3136079/
Last edited:
Barry
Senior Member (Voting Rights)
[My bold]An overview of randomization techniques said:
I cannot claim to understand all the fine detail of this, but ...
- What was different about the first 3 at each centre?
- From the link you supplied, it seems clear that stratified randomisation relies on researchers understanding the implications of potential bias. But that could cut both ways, depending on whether they seek to avoid or induce bias. Can we be confident that their stratification actually did seek to minimise bias? Especially the depressive disorder bit?
Snow Leopard
Senior Member (Voting Rights)
I suspect it may have been as professional a process as the FITNET study offering a cure in the initial blurb...
( Cynical again )
Lucibee
Senior Member (Voting Rights)
This paper provides more info on the randomisation method they used (minimisation): https://www.ncbi.nlm.nih.gov/pmc/articles/PMC556084/
I presume the reason for "the first three patients at each centre being allocated with straightforward randomisation" is just a failure to understand how minimisation works, because that will sort of happen anyway (here, there are three intervention groups vs 'placebo' rather than just one intervention group vs 'placebo'). What they don't mention in the description, but is clear from the TMG/TSC minutes, is that only three centres were operational initially. I presume that Barts I and II are counted together, as Bristol was the seventh centre to be launched (several years into the trial).
Recruitment and randomisation will be an ongoing process. Each clinic will have a maximum capacity, so it will have been as important not to overrecruit as it was not to underrecruit in order to maintain flow through the trial. It must have been a logistical nightmare, particularly when they were also juggling staff recruitment too. That much seems clear from the minutes.
However, one of the key omissions for me from the main trial paper was a table of baseline characteristics that was stratified by treatment centre. I suspect it may show where randomisation may have gone slightly off.
But ultimately, the whole point of randomisation is to ensure balance between the groups in factors that you can't directly control for. If patients already have a notion of which group they want to be in (because it isn't and can't be blinded in any way), then biases will start to creep in. And the issue of recruitment recycling is therefore a valid one. I don't know how they prevented that - the temptation to give pts what they wanted (particularly CBT at Kings) must have been huge.
I presume the reason for "the first three patients at each centre being allocated with straightforward randomisation" is just a failure to understand how minimisation works, because that will sort of happen anyway (here, there are three intervention groups vs 'placebo' rather than just one intervention group vs 'placebo'). What they don't mention in the description, but is clear from the TMG/TSC minutes, is that only three centres were operational initially. I presume that Barts I and II are counted together, as Bristol was the seventh centre to be launched (several years into the trial).
Recruitment and randomisation will be an ongoing process. Each clinic will have a maximum capacity, so it will have been as important not to overrecruit as it was not to underrecruit in order to maintain flow through the trial. It must have been a logistical nightmare, particularly when they were also juggling staff recruitment too. That much seems clear from the minutes.
However, one of the key omissions for me from the main trial paper was a table of baseline characteristics that was stratified by treatment centre. I suspect it may show where randomisation may have gone slightly off.
But ultimately, the whole point of randomisation is to ensure balance between the groups in factors that you can't directly control for. If patients already have a notion of which group they want to be in (because it isn't and can't be blinded in any way), then biases will start to creep in. And the issue of recruitment recycling is therefore a valid one. I don't know how they prevented that - the temptation to give pts what they wanted (particularly CBT at Kings) must have been huge.
rvallee
Senior Member (Voting Rights)
That doesn't sound like patients were properly randomized. How widespread was that problem? 2-3 patients? Or half of them?
Given the newsletter promising a full recovery with glowing testimonies from current participants, I'm curious about what was said to the participants initially. It just sounds a lot like they were primed to believe that they were likely to fully recover, as otherwise why would any patient prefer one treatment over another?
Excerpts from the randomization protocol were published in earlier comments but they have deviated from protocol so many times that I don't see a reason to trust that what is written is actually what happened. The temptation to place participants who were primed to believe in CBT or GET into those groups is just too great when failure would have been catastrophic for everyone's career (especially after having pulled off the NICE guidelines despite having no reliable evidence for their claims). They cheated so much elsewhere, it's hard to believe they didn't cheat there as well.
Last edited:
I wonder if there can be an effect of patients choosing by choosing which center to be referred to (particularly between CBT and GET). In some places it may not be an issue but in London with the Kings and Barts centres offering different treatments would patients get to choose which the GP referred them to?
Barry
Senior Member (Voting Rights)
Even if the patients didn't, there is still potential bias in the doctor making a conscious choice of which they might think "best".
ETA: i.e If that bit of the allocation was not randomised.
Last edited:
ukxmrv
Senior Member (Voting Rights)
and also influenced by the talks given to ME groups in London by Chalder and other PACE people to try and drum up business. If they have a talk to a group and tell them what each centre offers and ask them to get a referral by their GP then they are already influencing patient selection.
The patients who didn't believe them would not have asked their GP for a referral.
Also those who know activity was bad may have avoided barts.
