1. Patients launch a $1.27 million crowdfunding campaign for ME/CFS gut microbiome study.
    Check out the website, Facebook and Twitter. Join in donate and spread the word!
ME/CFS and Beating the Clock
For Jody Smith, the ticking of a clock was enough at one time to chase her back to her bed. But with the passage of time, she has been able to reclaim her living room ...
Discuss the article on the Forums.

Measuring disability in patients with CFS: reliability and validity of the WSAS (PACE

Discussion in 'Latest ME/CFS Research' started by Dolphin, May 7, 2011.

  1. Dolphin

    Dolphin Senior Member

    Messages:
    6,472
    Likes:
    4,754
    Note, this study reports some PACE results - but it is only data from before the interventions so not so interesting.
  2. Dolphin

    Dolphin Senior Member

    Messages:
    6,472
    Likes:
    4,754
    List of WSAS questions

    I always prefer to have a look at the questionnaires when I can to see what they are measuring:

    (From full (unpublished) PACE Trial protocol)

  3. Dolphin

    Dolphin Senior Member

    Messages:
    6,472
    Likes:
    4,754
    Correlations before treatment may not be the safe as correlations after treatment

    Firstly, I think a problem is that no truly objective outcome measures are used.

    I think better data would have been got by comparing the WSAS scores with:
    which is an outcome measure in the PACE Trial (but wasn't reported in the Lancet paper), so they could have done the comparison either before and/or after treatment.

    Similarly they do have actometer data before treatment in the PACE Trial but this was not reported.

    The 6 minute walking test and 2 minute step test have some value. But they could be influenced by how well you think you are rather than how well you actually are.

    I think a problem with this study is that no correlations are reported for after treatment.

    We are supposed to assume that the same correlations will apply; however, they may not - there may be response biases (including specific to a therapy like CBT) that cause people to answer certain questionnaires differently e.g. to please the therapist. Maybe they would all still correlate together as they are all somewhat fairly subjective, or maybe they would not.

    The accuracy of a questionnaire after an intervention is at least equally, if not more, important.

    ETA: They explicitly say:
    By the way, I think there is a good chance that this paper was written with the PACE Trial results in mind, just as I think
    was published with the PACE Trial results in mind (it actually used data from the mid-1990s); people will recall that it was used to give the mean and standard deviation for the Chalder Fatigue scale for the general population (based on people who had attended GPs in the previous year i.e. it excluded a lot of the healthier people) when the protocol paper already quoted previous research by Trudie Chalder et al which gave a threshold for "fatigue casesness" - a definition which most likely didn't give the results they wanted (it is quite strict while the new one is much looser).
  4. Esther12

    Esther12 Senior Member

    Messages:
    5,101
    Likes:
    4,903
    Thanks for looking through this Dolphin. From my quick look at the paper, I couldn't see any interesting data that I wanted to dig into. I wish we could access the raw data.

    edit: I totally agree about the possibility of different treatments, particularly CBT, having an impact upon the way these different outcome measure correlate with one another. But there's nothing in this paper that would let us assess whether that's the case.
  5. Dolphin

    Dolphin Senior Member

    Messages:
    6,472
    Likes:
    4,754
    The results in Figure 1 might look less impressive if we had all the data

    I thought I'd highlight the following as people might not be aware of it if they just flick through the paper.

    There are two cohorts used, the PACE Trial participants and another group:
    They say:
    However, those aren't the only variables they could have checked for - Table 1 shows they had various other variables - I think if they should have simply listed them all if it were the case they did check them. Also, having similar baseline scores doesn't mean they all responded to therapy in a similar way.
  6. Dolphin

    Dolphin Senior Member

    Messages:
    6,472
    Likes:
    4,754
    Hyping CBT and GET:
    I can't remember that paper 2005 off the top of my head but most studies haven't published this data. Indeed, the famous Sharpe et al. (1996) CBT study in the BMJ was supposed to publish the data but they didn't - funny that.
  7. Dolphin

    Dolphin Senior Member

    Messages:
    6,472
    Likes:
    4,754
    They sneak in points that don't seem to be proven:
    Simply having a partner (or not as the case may be) tells one very little about "oversolicitous behaviour".
  8. Dolphin

    Dolphin Senior Member

    Messages:
    6,472
    Likes:
    4,754
    The WSAS correlated most strong with depression in the (untreated) CFS patients

    I think the authors may be correct that the WSAS may partly measure, in some people anyway, depression and anxiety. However, as predicted, I don't particularly like the way they make the point (in particular the underlined part).

    I think a "cautious attitude to physical activity" would have made the basic point.
    -----
    On the correlation with depression, etc, they say in the introduction:
    The corollary of this is that improvements may represent improvements in depression.
  9. Esther12

    Esther12 Senior Member

    Messages:
    5,101
    Likes:
    4,903
    I noticed the over solicitous thing as well, and seemed to remember the previous evidence in support of this being amusingly weak.

    You might know this better than I, but from what I remember, the emergence of the idea that 'over solicitous' relationships were bad for outcomes came from a study in which they predicted that those who were less happy in their relationships would be more at risk of CFS (stress, depression, etc), and then actually found that it was very marginally the opposite. So they then came up with a new psychosocial story (as one always can).

    They didn't seem terribly concerned about the psychosocial harm that may be done to carers who then had to worry that their attempts to support their loved one were actually perpetuating their illness. That's a psychosocial strain!
  10. Dolphin

    Dolphin Senior Member

    Messages:
    6,472
    Likes:
    4,754
    What data were we not shown?

    As I pointed out above, we are not given correlations with some measures we know were used in the PACE Trial. One reason for this is that some may not have been measured in the other cohort. However, as with a lot of these studies, they may decide not to show us some data that doesn't suit them.
  11. Hope123

    Hope123 Senior Member

    Messages:
    1,141
    Likes:
    531
    Dolphin, I looked at this scale briefly before and believed I e-mailed you on this prior but the improvement on the WSAS scores in PACE post-intervention was on the level of something like 3 points (don't remember exactly) and some past studies, done in other conditions, seem to indicate than a 7-10 point difference (or some number around there) is needed to mean something significant.

    This is somewhere in the deep recesses of my mind but I also recall that WSAS was primarily developed/ validated/ used in psychiatric disorders rather than other disorders so the question might be how valid is it for ME/CFS.

    Haven't read this paper yet.
  12. Dolphin

    Dolphin Senior Member

    Messages:
    6,472
    Likes:
    4,754
    Yes, you did send me something before - thanks. Here it is:
    Also, the SMC group went up by 3 points so, when adjustments were made to baseline scores, CBT was 3.6 points better than SMC and GET was 3.2 points better. I think it's easier to criticise those sorts of figures as perhaps not being of much significance (especially if one factors in any response biases, and perhaps improvements in depression, etc.).

    I wouldn't say this is the most exciting paper in the world to read. I just can read these sorts of papers relatively easily now as I know the questionnaires, etc quite well so like to read them.
  13. Snow Leopard

    Snow Leopard Senior Member

    Messages:
    2,203
    Likes:
    1,536
    Australia
    The problem is that if there is such a 'fear', the resulting behaviour is not at all changed with CBT, which brings the whole model of CBT into question

    http://www.ncbi.nlm.nih.gov/pubmed/21414449 (and the studies cited in that paper).
  14. oceanblue

    oceanblue Senior Member

    Messages:
    1,174
    Likes:
    342
    UK
    True, but worth noting that they are now claiming the WSAS data refutes this bunkum, so we can quote this if the 'loved ones being too supportive' theory is used again.
  15. oceanblue

    oceanblue Senior Member

    Messages:
    1,174
    Likes:
    342
    UK
    Now some doubt over relationship between reported disability and physical capacity:

    shame they didn't discuss this in the PACE paper.
  16. Esther12

    Esther12 Senior Member

    Messages:
    5,101
    Likes:
    4,903
    Or go in to detail here! Have I missed the data for this point? On my read through, I couldn't see any figures that let us know exactly what was being referred too (I am feeling particularly dim at the moment though).
  17. Dolphin

    Dolphin Senior Member

    Messages:
    6,472
    Likes:
    4,754
    Yes, interesting point about strain on partner.
    I don't recall off-hand what research they might be referring to; they reference 38 & 39:
    I imagine the data could be interpreted in other ways.
  18. oceanblue

    oceanblue Senior Member

    Messages:
    1,174
    Likes:
    342
    UK
    Table 1 shows only a weak correlation between the subjective WSAS score and the objective 6MWT (r=-0.261) and Step test(r=0.128). As both should be related to true underlying disability/physical capability, you would expect a bigger correlation than this, hence their suggestion that reported disability might not be such an accurate measure of actual disability.
  19. Esther12

    Esther12 Senior Member

    Messages:
    5,101
    Likes:
    4,903
    Hmmm... they look later than the paper I was thinking of. It could have been as late as 98, but I didn't think the initial paper involved Wessely. I read it long before I was keeping notes on this stuff though (and I've recently lost most of my notes!). Maybe I've just started a CFS myth!
  20. Esther12

    Esther12 Senior Member

    Messages:
    5,101
    Likes:
    4,903

    I'm a bit shoddy on statistics stuff, so don't really understand how these things are calculated, or having much understanding of their significance. Dolphin suggested 'The Cartoon Guide to Statistics' as an appropriate text for my level of ignorance, but I've yet to get around to it.

    I should really take the time to re-read the paper properly, and do some googling. I didn't take much in from it first time around. I'm off to read the wikipedia page on the 'Pearson product-moment correlation coefficient'.

See more popular forum discussions.

Share This Page