• Welcome to Phoenix Rising!

    Created in 2008, Phoenix Rising is the largest and oldest forum dedicated to furthering the understanding of and finding treatments for complex chronic illnesses such as chronic fatigue syndrome (ME/CFS), fibromyalgia (FM), long COVID, postural orthostatic tachycardia syndrome (POTS), mast cell activation syndrome (MCAS), and allied diseases.

    To become a member, simply click the Register button at the top right.

Longitudinal mediation in PACE RCT of rehabilitative treatments for CFS: modelling & design conside

Dolphin

Senior Member
Messages
17,567
Free full text: http://www.trialsjournal.com/content/16/S2/O43

Longitudinal mediation in the PACE randomised clinical trial of rehabilitative treatments for chronic fatigue syndrome: modelling and design considerations


Kimberley Goldsmith1*, Trudie Chalder1, Peter White2, Michael Sharpe3 and Andrew Pickles1

  • *Corresponding author: Kimberley Goldsmith
Author Affiliations

1Institute of Psychiatry, Psychology & Neuroscience, King's College London, London, UK

2Centre for Psychiatry, Wolfson Institute of Preventive Medicine, Barts and the London School of Medicine, Queen Mary University, London, UK

3Psychological Medicine Research, Department of Psychiatry, University of Oxford, Oxford, UK

For all author emails, please log on.

Trials 2015, 16(Suppl 2):O43 doi:10.1186/1745-6215-16-S2-O43

The electronic version of this article is the complete one and can be found online at:http://www.trialsjournal.com/content/16/S2/O43


Published: 16 November 2015
© 2015 Goldsmith et al.

This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Background
Clinical trials require large monetary and time commitments and should provide information on both whether and how treatments work. Treatment mechanisms can be studied using mediation analysis, allowing refinement of treatments. Mediation studies often use only single contemporaneous measures of mediator and outcome limiting the conclusions that can be drawn. Longitudinally measured mediators and outcomes, such as those in the Pacing, Graded Activity, and Cognitive Behaviour Therapy: A Randomised Evaluation trial (PACE, ISRCTN 54285094) allow for more realistic estimates of mediated effects.

Methods
Autoregressive models accounting for measurement error were used to study treatment effect mediation of cognitive behaviour therapy (CBT) and graded exercise therapy (GET) in PACE. Fear avoidance and physical function were used as example mediator and outcome; these were measured at baseline and three times post-randomisation as part of the trial design. Model fit criteria, Wald tests and comparisons of parameter estimates were used.

Results
Longitudinal SEM were more flexible and gave what were likely more plausible estimates of mediated effects. Constancy of mediator - outcome effects over time and across treatment groups increased precision. For CBT and GET, 46% and 53% of the overall effect were mediated through fear avoidance.

Conclusions
Trials should be designed to include multiple measurements of mediators and outcomes so that more realistic mediation models can be used. Longitudinal models may have more power to detect mediated effects. Approximately half of the effect of each of CBT and GET were on physical function was mediated through reducing avoidance of fearful situations.

 

Dolphin

Senior Member
Messages
17,567
Physical function refers to the SF-36 physical function scores.

It's not that surprising that changing fear avoidance might change how people report their impairments.

The SF-36 physical function questionnaires asks whether you are limited regarding certain activities. The possible responses are: "Yes, limited a lot"; "Yes, limited a little"; "No, not limited at all". Change people's perceptions of what they feel they need to avoid and you'll change how they respond to questionnaires asking about their limitations.
 

A.B.

Senior Member
Messages
3,780
For CBT and GET, 46% and 53% of the overall effect were mediated through fear avoidance.

46% and 53% of 0% is still 0%.

On a more serious note, I'm interested in the statistical analysis which supposedly provides insight into causality. That one can deduce causality alone in a PACE trial like setting sounds hard to believe. I'm guessing they're playing word games as usual.
 
Last edited:
Messages
13,774
First report on PACE mediation:

Results
The IVs were weak, with a maximum R2 change of 0.03. The five strongest IVs were therefore used in the 2SLS in each case. There was modest mediation of CBT and GET effects (approximately 20% of the total effect). The IV-derived estimators were somewhat different in magnitude than the BJK estimators and were less precise. There is scope for modelling a common effect of mediators on outcomes across trial arms.


http://www.trialsjournal.com/content/12/S1/A144

Second:

Fear avoidance beliefs had the largest mediated effect on both fatigue and physical function for both CBT (fatigue CBT vs APT −1·22, 95% CI −0·52 to −1·97; physical function CBT vs APT 1·54, 0·86 to 2·31) and GET (fatigue GET vs APT −1·86, −0·80 to −2·89; physical function GET vs APT 2·35, 1·35–3·39). This accounted for 51% of the overall effect on physical function for GET and 37% for CBT, as compared with APT. The proportions were 61% and 34% for the same comparisons for the fatigue outcome.

http://www.thelancet.com/journals/lanpsy/article/PIIS2215-0366(14)00069-8/fulltext

Now the third:

Results

Longitudinal SEM were more flexible and gave what were likely more plausible estimates of mediated effects. Constancy of mediator - outcome effects over time and across treatment groups increased precision. For CBT and GET, 46% and 53% of the overall effect were mediated through fear avoidance.

I won't be happy til they get above 50% for CBT too. Surely they can find a way?!
 

user9876

Senior Member
Messages
4,556
Given it didn't make it into their main paper I assume the results didn't support the hypothesis. The idea of throwing ever more complex stats at poor data is somewhat strange.

This could be interesting "Autoregressive models accounting for measurement error" I've not seem anything they have done exploring measurement error and I don't see how an auto-regressive model would help since the biggest measurement error is because the measurement system is influenced differently by different treatment arms. I've never seen any discussion of error distributions from them or errors due to non-linearities in questionnaire based scales.

The previous mediator paper made some fundamental in ignoring possible temporal relationships between mediator and endpoints or initial variables. Also I would wonder about independence of variables.

Of course with no detail hard to say anything,

This statement feels very worrying
Longitudinal SEM were more flexible and gave what were likely more plausible estimates of mediated effects.
More flexible in that could be manipulated towards the hypothesis that they believe in "plausible'
 

Bob

Senior Member
Messages
16,455
Location
England (south coast)
The previous mediation analysis (which failed to demonstrate any mediation effects despite them concluding that they had demonstrated mediation effects) said that there would be another (longitudinal) mediation analysis.

It looks like they have managed to increase the (purported) fear-avoidance mediation effect for both CBT and GET for physical function in the new abstract. I can't understand how they could have an increased the mediation effects using the same data, in this analysis, so look out for creative statistical methods, and smoke, and mirrors.
 
Last edited:

Snow Leopard

Hibernating
Messages
5,902
Location
South Australia
Looking at the data 3 different times and getting 3 different results. *shrugs*

Physical function refers to the SF-36 physical function scores.

It's not that surprising that changing fear avoidance might change how people report their impairments.

The SF-36 physical function questionnaires asks whether you are limited regarding certain activities. The possible responses are: "Yes, limited a lot"; "Yes, limited a little"; "No, not limited at all". Change people's perceptions of what they feel they need to avoid and you'll change how they respond to questionnaires asking about their limitations.

I agree, that is the most likely explanation. To truly claim there is less fear avoidance, there needs to be objective evidence that the patients actually did more stuff.
 

user9876

Senior Member
Messages
4,556
Looking at the data 3 different times and getting 3 different results. *shrugs*



I agree, that is the most likely explanation. To truly claim there is less fear avoidance, there needs to be objective evidence that the patients actually did more stuff.
Really we need to look at the exact questions and how their semantics link and how they may correlate. To refer to a set of questions as a 'scale showing x' can be very misleading.
 

Daisymay

Senior Member
Messages
754
To truly claim there is less fear avoidance, there needs to be objective evidence that the patients actually did more stuff.

I'm useless as statistics but if I may just reiterate the obvious, as well as exposing the statistical and methodological problems with PACE, they need to be challenged on the fact that patients are not saying they are fearful of exercise due to irrational, illogical wrong illness beliefs as is assumed n PACE, people are justifiably unhappy at/not prepared to/fearful of exercising beyond their own personal capabilities as they know from past experience it will lead to an exacerbation of their symptoms which may last for days, weeks, months. Patients reluctance to exercise beyond their capabilities is indeed a rational and logical response to a serious physical disease where post exertional malaise is a defining, and objectively verifiable symptom.
 

Bob

Senior Member
Messages
16,455
Location
England (south coast)
For Reference - Details from Previous Mediation Analysis

It's all academic because there were no proven mediation effects in the previous paper and, as someone said 50% of zero is still zero. But, for reference, the (purported) fear-avoidance mediation effects from the previous paper were as follows...

Fatigue

CBT vs SMC 17%
CBT vs APT 34%

GET vs SMC 33%
GET vs APT 61%

Physical Function

CBT vs SMC 27%
CBT vs APT 37%

GET vs SMC 43%
GET vs APT 51%
 
Last edited:

Dolphin

Senior Member
Messages
17,567
I previously tried to find out which questions made up the fear avoidance questionnaire.
http://forums.phoenixrising.me/inde...alysis-of-pace-trial.34927/page-2#post-545617

----

Here are the questions for the Cognitive Behavioural Responses Questionnaire (CBRQ).

The text says:
We measured several of the putative mediators using the Cognitive Behavioural Responses Questionnaire (CBRQ); these were five cognitive measures: catastrophising, fear avoidance beliefs, damage beliefs, symptom focusing, and embarrassment avoidance beliefs, and two behavioural measures: all-or-nothing behaviour and avoidance or resting behaviour.

Appendix says:
Putative mediating variables at 12 weeks (24 weeks for walk test) Cognitive Behavioural Response Questionnaire This is a self-rated questionnaire designed to measure patients’ cognitive and behavioural responses to illness (5, 6). Items are rated on a 5-point Likertscale ranging from ‘strongly disagree’ to ‘strongly agree’. Five cognitive and two behavioural subscales can be derived, with higher scores on each scale indicating greater impairment of the response in question: catastrophising (range 0 to 16), fear avoidance beliefs (0 to 24), damage beliefs (0 to 20), symptom focusing (0 to 24), embarrassment avoidance beliefs (0 to 24), all-or-nothing behaviour (0 to 20) and avoidance/resting behaviour (0 to 32). To clarify, an example of an item from the cognitive subscale fear avoidance is “I am afraid thatI will make my symptoms worse if I exercise”. An example of an item from the behavioural subscale avoidance behaviour is “I tend to avoid activities that make my symptoms worse”.

I'll try to mark the ones that are definitely or possibly part of a scale - the definite ones are underlined. I have a PhD thesis that looked at them but it used four cognitive factors rather than five (it didn't have "damage beliefs). It had the first 11 as fear avoidance when there are only 6 of them (I think many of them are damage beliefs). It also had the next 6 questions [i.e. "I worry that I may become permanently bedridden because of my symptoms" to "I will never feel right again"] as catastrophising while there is only supposed to be 4 (although figures in the Appendix only add to 40 questions while there are 42 questions so the 6 questions may all be catastrophising].

Please indicate how much you agree or disagree with the following statements about your current symptoms by ticking the appropriate box.

Views about your symptoms
STRONGLY DISAGREE

DISAGREE NEITHER

AGREE NOR DISAGREE

AGREE

STRONGLY AGREE

I am afraid that I will make my symptoms worse if I exercise [said to be fear avoidance in Chalder et al. 2015]

My symptoms would be relieved if I were to exercise [I imagine this is fear avoidance]

Avoiding unnecessary activities is the safest thing I can do to prevent my symptoms from worsening [I imagine this is fear avoidance]

The severity of my symptoms must mean there is something serious going on in my body (Damage - from http://bjp.rcpsych.org/content/suppl/2011/08/30/bjp.bp.110.082974.DC1/ds82974.pdf)

Even though I experience symptoms, I don't think they are actually harming me

When l experience symptoms, my body is telling me that there is something seriously wrong

Physical activity makes my symptoms worse

Doing less helps symptoms

Symptoms are a signal that I am damaging myself [said to be damage beliefs in Chalder et al. 2015]

I am afraid l will have more symptoms if I am not careful

I should avoid exercise when l have symptoms [fear avoidance from "Mediators of the Effects on Fatigue of Pragmatic Rehabilitation for CFS" (Wearden & Emsley))]

I worry that I may become permanently bedridden because of my symptoms

lf I push myself too hard I will collapse

My illness is awful and I feel that it overwhelms me

lf I overdo things it will cause a major relapse (catastrophising - from "Mediators of the Effects on Fatigue of Pragmatic Rehabilitation for CFS" (Wearden & Emsley))

I will never feel right again (catastrophising - from http://bjp.rcpsych.org/content/suppl/2011/08/30/bjp.bp.110.082974.DC1/ds82974.pdf

When l experience symptoms, l think about them constantly (this was classed as symptom focusing in a PhD once)

I worry when I am experiencing symptoms (this was classed as symptom focusing in a PhD once)

When I am experiencing symptoms it is difficult for me to think of anything else (symptom focusing - http://bjp.rcpsych.org/content/suppl/2011/08/30/bjp.bp.110.082974.DC1/ds82974.pdf)

I think a great deal about my symptoms [symptom focusing from "Mediators of the Effects on Fatigue of Pragmatic Rehabilitation for CFS" (Wearden & Emsley))]

My symptoms are always at the back of my mind (this was classed as symptom focusing in a PhD once)

I spend a lot of time thinking about my illness (this was classed as symptom focusing in a PhD once)

I am embarrassed about my symptoms (this was classed as embarrassment avoidance beliefs in a PhD once)

I worry that people will think badly of me because of my symptoms [embarrassment from "Mediators of the Effects on Fatigue of Pragmatic Rehabilitation for CFS" (Wearden & Emsley))]

The embarrassing nature of my symptoms prevents me from doing things (embarrassment from http://bjp.rcpsych.org/content/suppl/2011/08/30/bjp.bp.110.082974.DC1/ds82974.pdf)

I avoid social situations because I am scared my symptoms will get out of control (this was classed as embarrassment avoidance beliefs in a PhD once)

I am ashamed of my symptoms (this was classed as embarrassment avoidance beliefs in a PhD once)

My symptoms have the potential to make me look foolish in front of other people (this was classed as embarrassment avoidance beliefs in a PhD once)

--------------
Two behavioural measures: all-or-nothing behaviour and avoidance or resting behaviour. I believe these last 13 questions are them

I stay in bed to control my symptoms (avoidance http://bjp.rcpsych.org/content/suppl/2011/08/30/bjp.bp.110.082974.DC1/ds82974.pdf)

When I experience symptoms, I rest [avoidance or resting behaviour from "Mediators of the Effects on Fatigue of Pragmatic Rehabilitation for CFS" (Wearden & Emsley))]

I tend to avoid activities that make my symptoms worse [avoidance or resting behaviour (Chalder et al. 2015 appendix)]

I tend to nap during the day to control my symptoms (this was classed as avoidance or resting behaviour in a PhD once)

I tend to overdo things when I feel energetic (this was classed as all-or-nothing behaviour in a PhD once)

| find myself rushing to get things done before I crash (this was classed as all-or-nothing behaviour in a PhD once)

I tend to overdo things and then rest up for a while (this was classed as all-or-nothing behaviour in a PhD once)

I tend to do a lot on a good day and rest on a bad day [all-or-nothing behaviour from "Mediators of the Effects on Fatigue of Pragmatic Rehabilitation for CFS" (Wearden & Emsley))]

I sleep when I'm tired in order to control my symptoms (this was classed as avoidance or resting behaviour in a PhD once)

l avoid making social arrangements in case I'm not up to it. (this was classed as avoidance or resting behaviour in a PhD once)

I avoid exerting myself in order to control my symptoms (this was classed as avoidance or resting behaviour in a PhD once)

I'm a bit all or nothing when it comes to doing things (this was classed as all-or-nothing behaviour in a PhD once)

I avoid stress situations (this was classed as avoidance or resting behaviour in a PhD once)
 

Dolphin

Senior Member
Messages
17,567
Here's the SF-36 physical functioning subscale. You score 0 for each "Yes, limited a lot"; 5 for each "Yes, limited a little"; and 10 for each "No, not limited at all".

The following questions are about activities you might do during a typical day. Does your health now limit you in these activities? If so, how much?

Yes, limited a lotYes, limited a littleNo, not limited at all

Vigorous Activities (such as running, lifting heavy objects, participating in strenuous sports)
Vigorous Activities (such as running, lifting heavy objects, participating in strenuous sports) Yes, limited a lot
Vigorous Activities (such as running, lifting heavy objects, participating in strenuous sports) Yes, limited a little
Vigorous Activities (such as running, lifting heavy objects, participating in strenuous sports) No, not limited at all

Moderate activities (such as moving a table, pushing a vacuum cleaner, bowling or playing golf)
Moderate activities (such as moving a table, pushing a vacuum cleaner, bowling or playing golf) Yes, limited a lot
Moderate activities (such as moving a table, pushing a vacuum cleaner, bowling or playing golf) Yes, limited a little
Moderate activities (such as moving a table, pushing a vacuum cleaner, bowling or playing golf) No, not limited at all

Lifting or carrying groceries
Lifting or carrying groceries Yes, limited a lot
Lifting or carrying groceries Yes, limited a little
Lifting or carrying groceries No, not limited at all

Climbing several flights of stairs
Climbing several flights of stairs Yes, limited a lot
Climbing several flights of stairs Yes, limited a little
Climbing several flights of stairs No, not limited at all

Climbing one flight of stairs
Climbing one flight of stairs Yes, limited a lot
Climbing one flight of stairs Yes, limited a little
Climbing one flight of stairs No, not limited at all

Bending, kneeling or stooping
Bending, kneeling or stooping Yes, limited a lot
Bending, kneeling or stooping Yes, limited a little
Bending, kneeling or stooping No, not limited at all

Walking more than one mile
Walking more than one mile Yes, limited a lot
Walking more than one mile Yes, limited a little
Walking more than one mile No, not limited at all

Walking half a mile
Walking half a mile Yes, limited a lot
Walking half a mile Yes, limited a little
Walking half a mile No, not limited at all

Walking one hundred yards
Walking one hundred yards Yes, limited a lot
Walking one hundred yards Yes, limited a little
Walking one hundred yards No, not limited at all

Bathing or dressing yourself
Bathing or dressing yourself Yes, limited a lot
Bathing or dressing yourself Yes, limited a little
Bathing or dressing yourself No, not limited at all
 

Snow Leopard

Hibernating
Messages
5,902
Location
South Australia
It is important to note that a change on the SF-36 PF score of 10, is either one question changed from limited a lot to not limited at all, or two questions moving half a step from limited a lot to limited a little, or from limited a little to not at all.

How meaningful is this level of change really?
 

Sean

Senior Member
Messages
7,378
It is important to note that a change on the SF-36 PF score of 10, is either one question changed from limited a lot to not limited at all, or two questions moving half a step from limited a lot to limited a little, or from limited a little to not at all.

How meaningful is this level of change really?
This is a core issue that PACE has to be confronted on, the serious disconnect between their abstract number waving and extravagant sales pitch, and the actual practical real world benefits to patients of those numbers, which is basically none.
 

Kati

Patient in training
Messages
5,497
You know, the PACE trial will be retracted and this group of so-called scientists will still write and publish papers.

And the band played on.
 

MeSci

ME/CFS since 1995; activity level 6?
Messages
8,231
Location
Cornwall, UK
Is it just me, or does the last sentence of the 'conclusions' make no (grammatical) sense at all?
Approximately half of the effect of each of CBT and GET were on physical function was mediated through reducing avoidance of fearful situations.

I don't know if it's brain fog, but I can't get my head round what they are trying to say here.
 

A.B.

Senior Member
Messages
3,780
They probably rushed this out of the door just to publish something and distract from the slowly sinking PACE trial (therefore typos).
 

Bob

Senior Member
Messages
16,455
Location
England (south coast)
Is it just me, or does the last sentence of the 'conclusions' make no (grammatical) sense at all?
My guess is that its supposed to mean: "Approximately half of the effect of each of CBT and GET were on physical function was mediated through reducing avoidance of fearful situations."

I hear that fiction publishers are cutting back on proof-readers these days.