Research brief: CBT & GET have little effect on pain in PACE Trial

View the Post on the Blog

The first in a new series of (mercifully) short pieces on recent research. By Simon McGrath

A new study used data from the large PACE Trial to see if CBT and Graded exercise (GET) improved ME/CFS pain. (The main trial itself found a moderate effect of CBT and GET on self-reported fatigue and function.) The study looked at muscle and joint pain frequency, and fibromyalgia status. And it compared CBT and GET to both specialised medical care (SMC) and APT (pseudo-pacing).

The questions are, did CBT or GET have any effect on these - and if so, how big was the effect? The answers were unimpressive:

​

So CBT helped a little with muscle pain compared with SMC (but not APT), a little with joint pain compared with SMC (but not APT), and not at all with fibromyalgia. Graded exercise fared a little better, but the effect was small at best.


What does a 'small' effect mean?

How small is small? For muscle and joint pain, the study used 'Cohen's d' to measure the effect of the treatments, and the values ranged from 0.24 to 0.31. Generally, 0.2 is considered a small effect, 0.5 is moderate and 0.8 is large (over 1 is very large), but the best way to gauge size is to look at a graph of the data for control vs treatment. The graph below shows an effect size of 0.27, typical in this study. The control group is in red, the treatment group - showing a small improvement effect (lower score) - is in black. The shaded pink area shows the overlap between the two groups. As you can see, there is a definite treatment effect overall, but it's rather a small one.

The effect on Fibromyalgia status needs to be measured slightly differently, so the authors used the "Number Needed to Treat" measure. On average, 12 patients need to be treated with GET for one patient to no longer meet the criteria for fibromyalgia. This too is a small effect (note it may be a case of moving from 'just meeting' the criteria, to 'just not meeting criteria', rather than recovery).


No evidence on effect on pain severity

Disappointingly, the study only measured pain frequency, not pain severity, with the authors acknowledging this approach "may not be as clinically meaningful" as using both pain frequency and severity. Many patients with severe pain would agree. Note that as this was an outpatient study, the severely affected - who often have the worst pain - were excluded.


Treatment effects on pain were independent of ‘change in fatigue’.

You might expect that if CBT and graded exercise helped patients recover then symptoms would improve across the board. But surprisingly, the study found that gains in pain were not linked to improvements in fatigue.


Summary

CBT and Graded Exercise had a small or insignificant effect on muscle and joint pain frequency, and Fibromyalgia in the PACE Trial. There was no evidence on the ability of CBT and Graded exercise to reduce pain severity.


Read more on the Phoenix Rising thread about this study

This article first appeared on the ME Association's website

Simon McGrath tweets on ME/CFS research

Support Phoenix Rising​

Phoenix Rising is a registered 501 c.(3) non profit. We support ME/CFS and NEID patients through rigorous reporting, reliable information, effective advocacy and the provision of online services which empower patients and help them to cope with their isolation.

There are many ways you can help Phoenix Rising to continue its work. If you feel able to offer your time and talent, we could really use some more authors, proof-readers, fundraisers, technicians etc. and we'd love to expand our Board of Directors. So, if you think you can help then please contact Mark through the Forum.

And don't forget: you can always support our efforts at no cost to yourself as you shop online! To find out more, visit Phoenix Rising’s Donate page by clicking the button below.

​

​

View the Post on the Blog

 

Thanks Simon.

This a comment from the thread on the paper: I thought I'd give it more exposure. If anyone can make the table easier to read, feel free to do so in a post underneath (and I'll try to link to it):
-----

Anyone else think it's a little suspicious that they only concentrated on muscle pain and joint pain given the table below. This is all the data we got for the other three pain symptoms.

Looking at tender lymph nodes, for example, the raw changes in percentages were:
-8% (APT),
-4% (CBT),
0% (GET)
and
-7% (SMC).

Or in percentage terms of those with the symptom at baseline:
26.3% reduction (APT),
10.7% reduction (CBT),
1% reduction (GET)
and
22.1% reduction (SMC only).

The results for sore throat, although not as clear cut, also don't suggest much difference or evidence for CBT and GET (headache is intermediary)

All they say is:

"We chose muscle pain and joint pain as individual symptom outcomes, before analysis of outcome data, since these were the most commonly reported [symptoms] at baseline".

(note: protocol doesn't mention they are going to do this, so there is a chance they may have had a feel for how things were going, or know from other situations)

Table 2. Presence of individual symptoms, valid % (n)
APT APT CBT CBT GET GET SMC only SMC only
Baseline 52 weeks Baseline 52 weeks Baseline 52 weeks Baseline 52 weeks
Symptom n=159 n=151 n=161 n=145 n=160 n=144 n=160 n=149
Muscle pain 71 (113) 60 (91) 64 (102) 50 (73) 71 (114) 50 (72) 72 (115) 60 (89)
Joint pain 52 (84) 49 (74) 50 (81) 34 (49) 51 (81) 39 (56) 52 (83) 45 (67)
Headache 37 (59) 31 (47) 39 (63) 28 (40) 37 (59) 29 (42) 36 (58) 34 (50)
Sore throat 32 (51) 31 (50) 33 (53) 28 (41) 29 (46) 31 (44) 30 (48) 27 (40)
Tender lymph nodes 31 (50) 23 (35) 32 (51) 28 (41) 29 (46) 29 (41) 32 (51) 25 (37)

---
Another point for what it's worth: the numbers whose data we don't have are:
APT: 8,
CBT: 16
GET: 16
SMC only: 11
i.e. the biggest numbers are for CBT and GEt. One wonders what their data might be like.

 

These 'Research briefs' are a great idea and this was a good one to start with - though I still don't get the 'cohen' chart concept - despite SImon's help!

 

Thanks Simon.

Anyone else think it's a little suspicious that they only concentrated on muscle pain and joint pain given the table below. This is all the data we got for the other three pain symptoms.

Looking at tender lymph nodes, for example, the raw changes in percentages were:
-8% (APT),
-4% (CBT),
0% (GET)
and
-7% (SMC).

Or in percentage terms of those with the symptom at baseline:
26.3% reduction (APT),
10.7% reduction (CBT),
1% reduction (GET)
and
22.1% reduction (SMC only).

The results for sore throat, although not as clear cut, also don't suggest much difference or evidence for CBT and GET (headache is intermediary)

All they say is:

(note: protocol doesn't mention they are going to do this, so there is a chance they may have had a feel for how things were going, or know from other situations)

Table 2. Presence of individual symptoms, valid % (n)
APT APT CBT CBT GET GET SMC only SMC only
Baseline 52 weeks Baseline 52 weeks Baseline 52 weeks Baseline 52 weeks
Symptom n=159 n=151 n=161 n=145 n=160 n=144 n=160 n=149
Muscle pain 71 (113) 60 (91) 64 (102) 50 (73) 71 (114) 50 (72) 72 (115) 60 (89)
Joint pain 52 (84) 49 (74) 50 (81) 34 (49) 51 (81) 39 (56) 52 (83) 45 (67)
Headache 37 (59) 31 (47) 39 (63) 28 (40) 37 (59) 29 (42) 36 (58) 34 (50)
Sore throat 32 (51) 31 (50) 33 (53) 28 (41) 29 (46) 31 (44) 30 (48) 27 (40)
Tender lymph nodes 31 (50) 23 (35) 32 (51) 28 (41) 29 (46) 29 (41) 32 (51) 25 (37)
---
Another point for what it's worth: the numbers whose data we don't have are:
APT: 8,
CBT: 16
GET: 16
SMC only: 11
i.e. the biggest numbers are for CBT and GEt. One wonders what their data might be like.

I think that choosing the two most frequent symptoms, plus fibromyalgia as a comorbid disease is a justifiable selection - though it's a good point that the results for other pain symptoms aren't as good. Just looking at the results for other symptoms, my take on it is:

• Headache: CBT best (CBT 39% -> 28% = -11%; GET -8%; APT -6%; SMC -2%)
• Sore throat: looks like nothing made a difference
• Tender lymph nodes: APT best (-8%, all others probably not significant)

I don't know how much difference that makes to the overall study findings which found CBT/GET made small or insignificant difference to the two most common pain symptoms. Including all symptoms would have made the results a bit weaker, but they are already pretty unimpressive!

 

This is such a great idea to do research briefs on the PACE trial!

Thanks Simon

 

Or in percentage terms of those with the symptom at baseline:
26.3% reduction (APT),
10.7% reduction (CBT),
1% reduction (GET)
and
22.1% reduction (SMC only).

That's gotta hurt.

 

Nice article - particularly like the graph. Makes it easy to understand the small effect these ''treatments'' have.

 

I think that choosing the two most frequent symptoms, plus fibromyalgia as a comorbid disease is a justifiable selection - though it's a good point that the results for other pain symptoms aren't as good. Just looking at the results for other symptoms, my take on it is:

• Headache: CBT best (CBT 39% -> 28% = -11%; GET -8%; APT -6%; SMC -2%)
• Sore throat: looks like nothing made a difference
• Tender lymph nodes: APT best (-8%, all others probably not significant)

I don't know how much difference that makes to the overall study findings which found CBT/GET made small or insignificant difference to the two most common pain symptoms. Including all symptoms would have made the results a bit weaker, but they are already pretty unimpressive!

It wasn't that long a paper so I think they should just have given the same data for all the pain symptoms esp. given it is known there are different sorts of pain (and drugs that target different types of pain) and the title, "Pain in chronic fatigue syndrome: response to rehabilitative treatments in the PACE trial."

It would, of course, made the discussion section (conclusions, etc.), and how people will interpret the findings, less clear cut, although as you have pointed out, even for those three, the results aren't clear cut themselves.

 

Its worth pointing out that they didn't follow the proper fibromyalgia diagnostic criteria as they left out the tender points. I had always thought that was a major part of the diagnosis.

We used the American College of Rheumatology criteria (Wolfe et al. 2010) to diagnose fibromyalgia, with the exception of tender points.

 

Its worth pointing out that they didn't follow the proper fibromyalgia diagnostic criteria as they left out the tender points. I had always thought that was a major part of the diagnosis.

We used the American College of Rheumatology criteria (Wolfe et al. 2010) to diagnose fibromyalgia, with the exception of tender points.

I thought that too, but it seems that the 2010 criteria don't require a tender point examination:
The American College of Rheumat... Wolfe 2010] - PubMed - NCBI

CONCLUSION: This simple clinical case definition of fibromyalgia correctly classifies 88.1% of cases classified by the ACR classification criteria, and does not require a physical or tender point examination.

So I don't really understand the authors point about 'with the exception of tender points when the paper they cite does not require tender points'. I'm afraid at that point, I thought 'whatever' and left it at that.

 

user9876 said: ↑

Its worth pointing out that they didn't follow the proper fibromyalgia diagnostic criteria as they left out the tender points. I had always thought that was a major part of the diagnosis.

We used the American College of Rheumatology criteria (Wolfe et al. 2010) to diagnose fibromyalgia, with the exception of tender points.​

I thought that too, but it seems that the 2010 criteria don't require a tender point examination:
The American College of Rheumat... Wolfe 2010] - PubMed - NCBI

CONCLUSION: This simple clinical case definition of fibromyalgia correctly classifies 88.1% of cases classified by the ACR classification criteria, and does not require a physical or tender point examination.​

So I don't really understand the authors point about 'with the exception of tender points when the paper they cite does not require tender points'. I'm afraid at that point, I thought 'whatever' and left it at that.

From my post to the thread http://forums.phoenixrising.me/inde...eatments-in-the-pace-trial.24886/#post-382774 :

An aside really - this isn't important.

They say they are using the Wolfe et al. (2010) Fibromyalgia criteria at least three times. This doesn't make much sense given this trial started in 2005!

Also, the Wolfe et al. (2010) criteria requires counting somatic symptoms http://www.ncbi.nlm.nih.gov/pubmed/20461783 but there is no mention of them doing this.

The protocol (http://www.biomedcentral.com/1471-2377/7/6) tells us they planned to use:

4. Presence or absence of fibromyalgia (using chronic widespread pain criteria only and not tender points) [41]

Wolfe F, Smythe HA, Yunus MB, Bennett RM, Bombardier C, Goldenberg DL, al. : The American College of Rheumatology 1990 Criteria for the Classification of Fibromyalgia. Report of the Multicenter Criteria Committee.
Arthritis Rheum 1990, 33:160-172.​

This matches the description in the text so is presumably what they used.

 

I see Julius Bourke and Peter White, who led the (spun) PACE Trial pain results study, are the people leading an expensive trial in the UK that was formerly funded by the CFS Research Foundation and now by Action for ME:
http://www.actionforme.org.uk/get-informed/news/our-news/biomedical-research-move-agreed

 

I see Julius Bourke and Peter White, who led the (spun) PACE Trial pain results study, are the people leading an expensive trial in the UK that was formerly funded by the CFS Research Foundation and now by Action for ME:
http://www.actionforme.org.uk/get-informed/news/our-news/biomedical-research-move-agreed

Thread here on that development: http://forums.phoenixrising.me/inde...-research-foundation.30309/page-3#post-479775

 

While I have elsewhere cast doubt on the distributions involved, I want to comment that, even if you assume PACE really did deal with Gaussian distributions, using Cohen's d instead of Pearson's measures would reduce the primary reported effects in the original study to something scarcely worth mentioning.

This is in line with the extremely modest results on the only objective scale, which would not be considered clinically significant if we were talking about a serious condition like heart failure where improvements are notably difficult. (I notice that even the alleged "control" group gained 20 meters in the six-minute walk, which I therefore subtract from the gains claimed for therapy. This still leaves a bias in the form of the hidden assumption that none of those who declined a second test (around 30% in each group) would have scored lower. This is where the strategy of making objective measures secondary paid off for the PACE authors. They could use subjective measures without objective backing while claiming their study included objective measures. Had those objective measures been primary, failure to participate in both before and after tests would have removed many subjects from the trial data.)

If my criticisms about the distribution in use are valid, the standard deviation, which plays a critical role in all such measures, could be mainly the result of arbitrary bounds and sample size.