PACE-gate ……. More wrongdoings? – PARTICIPANT NUMBERS seemingly just DON’T ADD UP

[A.B.]

To this end, FOI requests have been made by different parties to each of the participating centres (i.e. the Trusts involved) for their participant numbers and these responses also strongly suggest that the figures do not add up.

I see what you're saying now.

May I suggest making this the central piece of information in future discussions? ie. that participant numbers as told by the treatment centers don't match what is reported in published articles on PACE.

(or people whose concentration runs out after a few paragraphs can easily misunderstand, like I did)

 

[lilpink]

I see what you're saying now.

May I suggest making this the central piece of information in future discussions? ie. that participant numbers as told by the treatment centers don't match what is reported in published articles on PACE.

Apologies, I thought it was clear.

 

[Barry53]

I'd be curious for your input.

No, I was wrong. The info in that back of my mind was simply projected information in the protocol, back in 2006.

 

[Barry53]

From online sources – The Edinburgh cohort was about 100 – http://www.meassociation.org.uk/201...he-body-edinburgh-evening-news-15-april-2011/

To have any argument worth arguing and not risk getting shot down in flames, the data sources have to be unimpeachable. The above worries (especially as journalism and accuracy tend to be like oil and water) me where it says:-

It involved major scientists, universities from across the UK, including Edinburgh, and 600 patients – 100 of whom were based in the Capital.

• Round numbers, like 100, always make me suspicious for things like this, especially when ...
• They use a round number of 600 which we know was not right.

Unless we can rely on the numbers going in, any result coming out will be unreliable.

The value of "around 70" sounds more valid.

 

[Barry53]

I'm sorry but I don't think this goes anywhere.

The value of 70 is "around 70" so let's say it is 71.

The value of 100 is dubious I think, especially as the other number is 600. Let's say it was 102.

You could then have:-

135 + 63 + 71 + 102 + 135 +135 = 641.

Although these will not be the actual figures, suspect the truth is along those lines.

Edit: I realise others have got there before me.

 

[Barry53]

Hi, might have found the table you are meaning with the number of trial participants to centres. If you click the link ( http://www.sciencedirect.com/science/article/pii/S0022399914001883 ) then go to the top of Table B. It's from the adverse events paper. If you add up the participants experiencing less than 4 and more than 4 adverse events then that should give you how many participants were at each centre. Kx

So this gives:-

110 + 108 + 135 + 113 + 111 + 63 = 640

Edit: Corrected my silly mistake!

 

[Deepwater]

Perhaps Peter White was just too preoccupied at the time working with Esther Crawley on this NOD paper https://academic.oup.com/qjmed/article-lookup/doi/10.1093/qjmed/hct061 ?

I've often wondered if patients were dropping out of the trial because it was making them worse, being erased from the data and replaced. Does anyone know how they handled drop-outs?

 

[lilpink]

No, I was wrong. The info in that back of my mind was simply projected information in the protocol, back in 2006.

No worries. Thanks for looking.

 

[lilpink]

To have any argument worth arguing and not risk getting shot down in flames, the data sources have to be unimpeachable. The above worries (especially as journalism and accuracy tend to be like oil and water) me where it says:-

If 'unimpeachable' means FOIA request info from NHS Trusts..then it should be kosher.

 

[lilpink]

The slight discrepancy of the 642 figure is presumably a reporting error in one of the numbers.

No. I have exact numbers which match entirely (bar one) with the Table provided earlier in this thread. There are no 'ifs' and 'buts'. It's not a question of fuzzy numbers at all. It's a question for the PACE PI's to explain missing data sets.

 

[Jenny TipsforME]

I've found this hard to follow too, though very intriguing. So there's information to suggest that the total participants supplied from the treatment centre sites under FOIA doesn't add up to the numbers quoted in PACE publications? How many people are missing/wrongly placed?

 

[lilpink]

How many people are missing/wrongly placed?

It appears to be a substantial number but at this point in time I'm afraid I will have to refrain from making any more comments as work is now progressing behind the scenes. Indebted to @kirstenthomson2000 for adding a very important piece of the jigsaw.

 

[Jenny TipsforME]

Even if it turns out not to be this I do wonder if there's something like this to be found. QMUL spent an awful lot of money to cover over null results :thumbdown:

 

[RogerBlack]

QMUL spent an awful lot of money to cover over null results :thumbdown:

I really doubt it.
If it was clear fraud they were covering up, the researchers would have been thrown under the bus.
Actual flat out making up data is a whole different class from 'disagreements over interpretation of data'.

 

[user9876]

No. I have exact numbers which match entirely (bar one) with the Table provided earlier in this thread. There are no 'ifs' and 'buts'. It's not a question of fuzzy numbers at all. It's a question for the PACE PI's to explain missing data sets.

Can you give the numbers if it is an FoI then the data is considered in the public domain.

It may be worth looking at each center and what treatments they carried out. I think Barts is GET and Kings CBT. I assume the SMC and APT were split between them but not sure. I wonder if one figure is off (say from Barts) if that is a figure for those in an active treatment group rather than SMC.

 

[user9876]

I really doubt it.
If it was clear fraud they were covering up, the researchers would have been thrown under the bus.
Actual flat out making up data is a whole different class from 'disagreements over interpretation of data'.

For clear fraud to be found someone would have to look. I thing that QMUL didn't bother to look and nor did the MRC. But I don't think there was fraud - the results are so poor which is why they had to do such blatant spinning. If there was any made up figures they could have avoided this. I suspect that White and co think they are being honest and believe they are justified in manipulating results partly because they believe and partly because they think everyone does it.

 

[Barry53]

Total 640 (off by 1 from the 641)

But one dropped out, withdrawing their consent. So the released data has 640 participants, presumably because they did not have permission to use that other person's data.

 

[Barry53]

There was one participant that withdrew consent late if I remember right and that might be the source of this error.

Pretty sure that is going to be it. The FOI-released data is for 640 participants.

 

[Barry53]

Nope ...the data shouldn't lie. Bear in mind this is all based on data either they have presented in published papers or Trusts have delivered courtesy of FOIA requests. Trust me... there isn't a simple answer to this which doesn't make them look dodgy in one way or another.

In that case I suggest being extremely careful that the FOI responses you get reveal all the participants handled at each centre, over the whole time period, because the recruitment was phased I believe? Would be a devil if a "vexatious response" to a FOI request misled you, by not asking exactly the right FOI question, and them exploiting that to be selective about the data they give you.

 

[Barry53]

Apologies, I thought it was clear.

Nah ... I was confused too