PACE Trial follow-up: Here's the table looking at the effects of having CBT or GET after 52 weeks

[elliepeabody]

Sure. So all this data is the score at 2.5 years. Some people (those in the APT/SMC groups) had no sessions of CBT, some in the APT/SMC added a few sessions of CBT, so in total they had either 1-9 sessions or 10 sessions. People in the CBT/GEt groups had 15 sessions through the trial. Some had no more sessions (so 15 in total), some had a few more (so 1-9 +15 =16-24, or 10+15=25)

So to answer your second question In the first half of the graph only: the diamonds are people assigned to the APT arm and the squares are SMC arm. In the second half of the graph only: diamond is people assigned to the GET arm and squares are CBT arm.

I realise this isn't quite clear. I could add a legend but I think it will get a bit cluttered. Or I could average the two points (perhaps weighted by the number of people in each group). Thoughts?

(And yes, if they overlap, they look funny! They vary because, CBT/GET has no real effect )

@Anna Wood
Ahh, thank you for the clarifications. I'm not familiar with interpreting research results so I wasn't sure if the diamonds and squares were representative of a usual way of illustrating data that I wasn't aware of. I like that you have differentiated between SMC, APT, GET and CBT, although I do think the graph would benefit from a explanation of what the diamonds and squares represent.

It works well taking the SF36 PF up to 100 - it puts the results into context.

Sorry if I wasn't clear above but I'm still not sure what the 2.5 years covers. Is this follow up data with a beginning point at the end of the trial period and an end point 2.5 years later?

 

[wdb]

I was curious to see what sort of correlation there was so I added linear trend lines and calculated the correlation coefficients. I used minimum value for CBT/GET sessions as we can't know from the data an accurate average.

Looks to me like 3 out of 4 correlations suggest more CBT/GET leads to worse outcome but would be good if anyone else can verify.

Edit: wondering if the correlations could be calculated more accurately by taking into account the number of participants in each sample

 

[Esther12]

Impressive lack of correlation.

 

[Anna Wood]

Edit: wondering if the correlations could be calculated more accurately by taking into account the number of participants in each sample

Nice - certainly makes it clear what is looks like by eye. Yes, weighting by number of participants would be more accurate...

 

[Anna Wood]

Is this follow up data with a beginning point at the end of the trial period and an end point 2.5 years later?

Ah, sorry, I misunderstood. Yes, beginning at start of trial, end 2.5 years later

 

[Esther12]

I could average the two points (perhaps weighted by the number of people in each group). Thoughts?

Maybe that would be the best, but may not be worth the trouble. Graph already illuminating.

 

[Anna Wood]

Something like this might help clarify it:

Thanks Sean, I missed this yesterday. That's seems like a simple way of doing it.

 

[Anna Wood]

New version - with various additions, thanks Sean and Simon.

 

[Daisymay]

Nice one Dr Abbott !

http://www.meresearch.org.uk/news/ballyhoo-over-pace/

Media ballyhoo over PACE
Posted on 28 Oct 2015
Dr Neil Abbott, ME Research UK

 

[Dolphin]

Nice one Dr Abbott !

http://www.meresearch.org.uk/news/ballyhoo-over-pace/

Media ballyhoo over PACE
Posted on 28 Oct 2015
Dr Neil Abbott, ME Research UK

I haven't highlighted this as I'm disappointed with:

The authors may speculate that patients in the standard medical care or pacing exercise groups were ‘helped’ by additional psychological therapies in the 18 months after the end of the trial, but they don’t know that for certain, and nor do they show data to back up their claim.

Table C in the Appendix does give the information and it doesn't back up the authors' claim.

 

[Keith Laws]

New! Improved! (Bit simpler)
This really is my last word on things, you'll be relieved to hear.



View attachment 13409

Simon - how did you weight the means?

New! Improved! (Bit simpler)
This really is my last word on things, you'll be relieved to hear.



View attachment 13409

 

[Sasha]

Welcome to the forums, @Keith Laws! We love your blog posts. Thank you so much for helping us to get the word out about the problems with PACE.

We've found, sadly, that patients on their own aren't listened to. We desperately need heavyweight scientists like you to take on this bad research.

PACE has damaged lives, and its influence continues to do so every day.

Thanks again for your involvement.

 

[Simon]

Simon - how did you weight the means?

Great to have you on the forum, Keith!

Simply weighted means by group size, hope that was the right way to go. Just shout if you want my spreadsheet with the data (or click on my profile photo and choose "Start a conversation").

added: I take your point that as allocation wasn't randomised it's hard to interpret the data, but it is still the best evidence we have - and seem to me to be, at the very least, grounds for rejecting the author's claim that post-trial CBT/GET had a role in SMC/APT groups catching up at follow up.

 

[Keith Laws]

Welcome to the forums, @Keith Laws! We love your blog posts. Thank you so much for helping us to get the word out about the problems with PACE.

We've found, sadly, that patients on their own aren't listened to. We desperately need heavyweight scientists like you to take on this bad research.

PACE has damaged lives, and its influence continues to do so every day.

Thanks again for your involvement.

Thanks Sasha

 

[Tom Kindlon]

I just posted a quick comment to PubMed Commons on this:

Main finding: no difference between all 4 groups. Data shows improvement for APT or SMC-only groups not due to extra CBT or GET
http://www.ncbi.nlm.nih.gov/pubmed/26521770#cm26521770_15247

Normally I would reference my comments but on this occasion just wanted to write something quickly.