Cochrane ME/CFS GET review temporarily withdrawn

ladycatlover said:
Chrome. I wonder if they only allow computers in UK to access the full paper for free?
Click to expand...

I could access it and just downloaded the whole paper.

There seems to be different levels of access, based on (among other things) country and wether each nation have a "national provision", Norway and UK amongst them:

Free online access for residents through a national funded licence; see countries with funded provision table below.
Click to expand...

There are more, different ways to access, tables in the link. (sorry, muddled brain)

https://www.cochranelibrary.com/help/access
 
inox said:
(sorry, muddled brain)
Click to expand...

You and me both! ;) We have a wasp problem at home that is doing my head in, so in an over-vigilant state all the time. Hopefully they will disappear soon as we have colder weather forecast for end of the week. This is on top of journey home on Thursday in husband's uncomfortable (to me) car. I really should have asked him to come home and swap cars for drive home for me. Oh, and Saturday he went on the People's Vote march in London, so I had to deal with the wasps on my own - fortunately only one appeared in bedroom that day! Sorry, very off topic!
 
Lucibee said:
Does anyone have an copy of the original Cochrane review (pre-2017)? It looks like Larun changed some of the data in the 2017 version (or it might be she was losing track in the comments more than I am!), but I can't tell what.

(Bob refers to some non-significant data that I can't now find anywhere)
Click to expand...

Do you mean the FINE data: -2.12 [ -4.49, 0.25 ] ?

If so that's still there on page 65 of the pdf, in Analysis 1.2

I was just about to try to work out what was going on with Larun's response about that.

I'm sure I was told that old Cochrane reviews (post 2014?) are still available, but I couldn't see how.

PS: I've found downloading the pdf is much more convenient than using their webpage for searching through, and trying to follow all the points being made.
 
Lucibee said:
Does anyone have an copy of the original Cochrane review (pre-2017)? It looks like Larun changed some of the data in the 2017 version (or it might be she was losing track in the comments more than I am!), but I can't tell what.

(Bob refers to some non-significant data that I can't now find anywhere)

ETA: Found it here: https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD003200.pub7/information#whatsNew

ETA2: No differences.
Click to expand...

I thought that Bob had referred to them using the alternative CFQ marking scheme which is possibly where the non-significant data comes in. In Fine they were using the binomial scheme and didn't get a significant result but then they wrote a letter saying it would be significant if they used the 'likert' (which isn't likert) type marking. Then PACE switched and I have a feeling that Cochrane also used this.

Personally I think any research using the CFQ should be rated as poor quality. But the switching of marking schemes raises a real issue as some improve on one scheme and get worse on the other. I'm pretty sure the movement is towards a switch to improve with a move from binomial to 'likert' on the pace data. But I think the validation work was done on the binomial scoring system if its done on both then one piece of validation work is plainly wrong.
 
Adrian said:
I thought that Bob had referred to them using the alternative CFQ marking scheme which is possibly where the non-significant data comes in. In Fine they were using the binomial scheme and didn't get a significant result but then they wrote a letter saying it would be significant if they used the 'likert' (which isn't likert) type marking. Then PACE switched and I have a feeling that Cochrane also used this.

Personally I think any research using the CFQ should be rated as poor quality. But the switching of marking schemes raises a real issue as some improve on one scheme and get worse on the other. I'm pretty sure the movement is towards a switch to improve with a move from binomial to 'likert' on the pace data. But I think the validation work was done on the binomial scoring system if its done on both then one piece of validation work is plainly wrong.
Click to expand...

It was more to do with the inclusion of the unplanned, unpublished 33-pt analysis data in the SMD instead of binomial data, and then relegating the SMD data on the basis that it was more difficult to understand. Frankly, any pooling of inconsistent data is going to be dangerous (and potentially uninterpretable), so I can sort of see why they did it, but they should have been much more up-front about it.

CFQ simply isn't useful as a repeated measures tool. It's just too crude. It measures "worsening of fatigue compared with when you were last well". How anyone thought they could realistically use it to measure improvement in fatigue, when the whole scale is biased towards "getting worse", is just daft. Use of the binomial scale sort of addresses some of this (0=better/no change; 1=bit worse/much worse), but without sorting out the baseline issues, it's useless.

Larun et al. following PACE's lead on the switch just shows how they fail to understand what the measures they are analysing actually do.
 
Last edited:
Another point that Tom and Bob picked up on was the downgrading of evidence obtained at follow-up (rather than at end of treatment).

The main excuse she gave Bob was therapy contamination, but we know from PACE that this wasn't the case until at least 2.5 year follow-up. I would hope that the other studies had taken steps to minimise this anyway, so to mention it as an issue without discussing it in the review seems odd.

When Tom brought this up, it was all about baseline differences between groups (possibly due to drop-outs being more severely affected?) - which she conceded could be addressed by the individual patient data analyses (by adjusting for baseline). So I guess we'll have to wait and see - unless it's the IPD review that's been pulled...
 
You must log in or register to reply here.
Back
Top Bottom