O'Dowd-Crawley early intervention study

I'm sure that there's a worthwhile paper in this if anyone managed to piece it together.

 

All that hot air would make a great paper airplane.

 

Or the Hindenberg.

 

I don't understand. Why have they published the dataset but not a full report of the study?

 

At a glance, no. (tho pls check - I do miss things!). Looks like they haven't included the WPAI qs1-4, which includes no hours worked/not worked in past 7 days. They've only included qs 5 and 6. Here's a pdf of WPAI to see why that's important: http://oml.eular.org/sysModules/obxOml/docs/ID_98/WPAI-GH_English_US_V2.pdf
Summary data of WPAI were included in the basic results: http://www.isrctn.com/editorial/retrieveFile/283ff95f-3ce1-48dd-94d3-441b74834d04/23369

 

Someone should tell Ben Goldacre... #AllTrials

 

If it's in the ISRCTN registry, it should at least get picked up by subsequent Cochrane analyses, and NICE guideline committees, one would hope?

 

I'm afraid that my brain cells are slowing down, and I have a couple of areas that I am looking at at the moment. I'll happily look at it when there's a space and my brain is active, but that may be a while yet. Sorry!

 

Because the feasibility study failed to achieve their objectives

The dropout rate for the intervention was 47%, but since most participants spontaneously recovered, and few would have met CFS criteria at any point (ignoring 6+ month requirement) it is impossible to say how large a full scale study would need to be. (it could be 1000+ participants required!)

 

Thanks this seems like a useful observation, even if I think we already knew that a lot of people presenting with possible (broadly defined?) CFS to a GP are going to recover even before the six month mark.

 

Best guess would be that they didn't like the null results. There are no objective measurements, as usual. Frankly this is dereliction of duty from the funding institutions to never require any objective data after decades of the same research in a loop and 2 decades in practice. Seems like the only objective measures planned, hours worked and hours missed, were dropped. As is tradition.

Although looking at the data, I have no idea who these people are since there are 7 with a SF-36 baseline of 95+, who all ended up with 95+ at 6 months. Those have CFQ ranging from 15 to 28. Randomization was pretty bad, the control group had 5/16 of the 95+, leaving 2/20 for the EI group.

A quick calculation, removing missing entries, on SF-36, the only mildly relevant measure:

"Usual" care: start=67 end=84
"Usual" care + EI: start=60 end=77​

On absolute measure, the "intervention" yields a lower SF-36, but the same increase in report of 17. Some are dramatic changes, I see from 20 to 100 and from 10 to 70, but the direction of the various questionnaires have little relation to one another, they are all over the place. And there is a lot of missing data.

Roughly speaking, even by the usual abysmal standards of research by Crawley, this appears too bad to publish. I wonder how much copy and paste is involved in writing every paper, those trials all do the same thing anyway. Complete waste of public funds and academic resources.

 

Thanks a lot of John for his persistence on this (and also to those who first raised this issue long ago).

I wonder if it would be worth a couple of people trying to do a short paper on the results (and maybe the trouble with getting the results out) to get it in the peer reviewed literature? Could also mention the spin around PACE and the difficulty of getting their prespecified results, and then ask Crawley/O'Dowd if they want their names attached?

 

What would the paper say exactly? That there was a high dropout rate and most patients recovered spontaneously leading to insufficient sample size to attempt to detect a difference?