NICE Support for GET (Fulcher & White): RCT of GET in patients with the CFS

[Valentijn]

With the PACE trial coming under heavy fire from several directions, there's been a bit of a discussion about what happens if it officially sinks to the bottom of Wessely's metaphorical ocean. Most specifically, what happens to NICE?

Although the current NICE guidelines predate PACE, PACE has been reportedly used to delay a review of the NICE guidelines. So one possibility is that such a review will become more feasible or even urgent. But in that case - what support remains for CBT and GET in NICE?

In the case of GET, there are 6 studies. This thread is for discussing one of them, authored by Kathy Fulcher and Peter White. It's an older study, from 1997, thus it predates this forum and the close scrutiny warranted for biopsychosocial papers.

 

[Valentijn]

Full Text: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2126868/

Fulcher KY, White PD. Randomised controlled trial of graded exercise in patients with the chronic fatigue syndrome. BMJ 1997;314:1647-52.

Design: Randomised controlled trial with control treatment crossover after the first follow up examination.

Setting: Chronic fatigue clinic in a general hospital department of psychiatry.

Subjects: 66 patients with the chronic fatigue syndrome who had neither a psychiatric disorder nor appreciable sleep disturbance.

Interventions: Random allocation to 12 weeks of either graded aerobic exercise or flexibility exercises and relaxation therapy. Patients who completed the flexibility programme were invited to cross over to the exercise programme afterwards.

Main outcome measure: The self rated clinical global impression change score, “very much better” or “much better” being considered as clinically important.

Results: Four patients receiving exercise and three receiving flexibility treatment dropped out before completion. 16 of 29 patients rated themselves as better after completing exercise treatment compared with eight of 30 patients who completed flexibility treatment. Analysis by intention to treat gave similar results (17/33 v 9/33 patients better). Fatigue, functional capacity, and fitness were significantly better after exercise than after flexibility treatment. 12 of 22 patients who crossed over to exercise after flexibility treatment rated themselves as better after completing exercise treatment. 32 of 47 patients rated themselves as better three months after completing supervised exercise treatment. 35 of 47 patients rated themselves as better one year after completing supervised exercise treatment.

Conclusion: These findings support the use of appropriately prescribed graded aerobic exercise in the management of patients with the chronic fatigue syndrome.

 

[Valentijn]

Fortunately there were already some advocates asking some good questions and raising some excellent points, 18 years ago. Those are published, along with a reply and some extra data from the authors, at www.ncbi.nlm.nih.gov/pmc/articles/PMC2127632/pdf/9361550.pdf

Summaries of the published comments:

MEA ( @charles shepherd & Anne Macintyre):

• Group who rated themselves as better (CGI) showed no comparative VO2max or muscle strength increase compared to other patients
• Percentage of patients receiving disability benefits not published
• Illness duration average of 2.7 years may make results inapplicable to newly ill patients
• Severe patients excluded
• Authors dismissed potential biological defects
• Subset may have disordered muscle metabolism

Alan Franklin:

• If improvement CGI scores included "a little better"(3), number of improvers in control group would have been nearly the same as in GET
• The type of patients included (able to attend clinic, good sleep, no psych problem) would be a small proportion of CFS patients
• Careful GET prescription is needed, not just "aerobic exercise"

Mike Sadler:

• 40% of screened patients were excluded
• Oxford criteria shouldn't exclude anxiety and depression
• BMI, age, sex comparisons between groups not in paper (added in author response)
• Not stated if CGI score of 1-2 was post-hoc
• CGI score of 1-3 would be more valid indicator of improvement (better versus no change or worse)
• No data regarding consultation rates, use of drug treatment, time off work

Ellen Goudsmit:

• 40% reported little or no improvement
• Variables not compared between those "very much better" and the rest, such as onset, various CFS symptoms
• Authors may have selected a CFS subgroup prone to fast recovery
• Authors fail to explain why 20 patients with no psych disorder are taking full-dose antidepressants
• Many CFS patients cannot tolerate full dose antidepressants, so some fatigue and malaise maybe have been caused by those drugs
• Necessary to characterize the responders versus the non-responders, to determine who should have this treatment

Authors' Reply (Peter White, Kathy Fulcher):

• Significant proportion returned to work (at long-term followup, non-controlled outcome)
• Patients should return to exercise after fever from triggering illness abates
• They use the same treatment principle with severely disabled patients in hospital
• Reduced activity may cause reduced oxidative metabolism
• Sleep disturbance may cause low cortisol
• CGI "a little better" classified as negative result before they collected any data (but not prior to randomization or treatment?)
• 20 patients took fulldose antidepressants to prevent relapse of previous comorbid depressive illness
• Baseline data of age, sex, BMI, illness duration excluded due to journal requirements

 

[Valentijn]

The NICE appendix for the review of the evidence is at https://www.nice.org.uk/guidance/cg53/evidence/full-guideline-appendix-1-196524110

The brief list of the evidence is at page 57 of the pdf. The extended comments are from pages 287-289. The validity assessment summary table is at page 437.

The NICE reviewers gave this study a very high rating for "level of evidence" at 1++. Evidence is rated from 1-4, with 1 being the best and 4 the worst. They also gave it a "validity score" of 17, which is quite high.

Some of that rating is due to it being a randomized and controlled trial. Another "strength" is that it used some objective outcome measures, including VO2max. Positive outcomes where claimed regarding CGI (Clinical Global Impression), VO2max, fatigue scores, and some SF36 scores.

According to the validity assessment, the also rated "good" for baseline comparibility of groups, follow-up, handling of drop-outs, statistical analysis, sample size calculations, and "adequate" for the treatment comparability and outcome objectivity.

 

[A.B.]

The NICE reviewers gave this study a very high rating for "level of evidence" at 1++. Evidence is rated from 1-4, with 1 being the best and 4 the worst. They also gave it a "validity score" of 17, which is quite high.

If the evidence is so good for GET, how come the already biased PACE trial failed to yield similarly positive results?

It seems that as time passes, the results of these studies become more and more modest. We've gone from impressive results to the point of questioning whether these interventions do anything at all.

 

[Sean]

Main outcome measure: The self rated clinical global impression change score, “very much better” or “much better” being considered as clinically important.

The NICE reviewers gave this study a very high rating for "level of evidence" at 1++.

How can any trial of any form of physical exercise therapy with no objective physical measure for primary outcome, be considered reliable or even relevant?

It speaks disturbing volumes about the peer-review process and culture that this sort of study is even allowed to be conducted, let alone published, let alone subsequently given a very high rating as evidence.

 

[Valentijn]

If the evidence is so good for GET, how come the already biased PACE trial failed to yield similarly positive results?

Despite their claims to exclude psych diagnoses and "appreciable sleep disturbance", 20 patients (out of the 66 total) were taking full dose anti-depressants. Another 10 were on low-dose tricyclic antidepressants for sleep problems.

The claim in the author's response to Goudsmit seems to be that the patients were cured of their "depressive illness" at the clinic, and were taking the meds (indefinitely?) just in case of possible relapse.

So I think they had a stereotypical group of Oxford fatigue patients with a lot of psych issues, and not many physically disabled ME patients. That makes it a lot easier to get the participants to improve their exercise habits.

 

[Valentijn]

How can any trial of any form of physical exercise therapy with no objective physical measure for primary outcome, be considered reliable or even relevant?

They used VO2max as a secondary outcome measurement, which showed some mild improvement in the GET patients versus the stretching/relaxation controls.

But ... the GET group started out with an average VO2max score of 31.8. For a group which is aged almost 38 years on average, and is 70% composed of women, that's a very normal score. The control group started a bit lower with a VO2max of 28.2, but was 79% women.

So the participants weren't deconditioned at all to start with, and accordingly it isn't meaningful to use a measurement which reflects conditioning.

Those 3 extra males in the GET group could also make a huge difference in scores, since men score much higher at VO2max. Not sure if the type of GET used would affect men and women differently though, regarding VO2max improvements.

 

[user9876]

If PACE as a large trial shows no improvements on objective measures or no correlations on objective measures this should bring doubt into all the non-blinded GET and CBT trials that use only subjective measures or that have failed to publish the objective ones,

 

[A.B.]

Do you know what criteria they used to diagnose CFS? I'm guessing Oxford. This would allow the patient cohort to contain of a substantial number of patients that didn't actually have CFS but something else, such as an episode of depression. In which case the study would demonstrate that exercise is useful for some symptoms of depression.

 

[WillowJ]

It speaks disturbing volumes about the peer-review process and culture that this sort of study is even allowed to be conducted, let alone published, let alone subsequently given a very high rating as evidence.

Yes. It should have been corrected in the preliminary reviews.

Failing that, at least the analysis should have been corrected in the pre-publication reviews.

Failing that, evidence reviews should have downgraded it severely, to: oh, yeah, there was this study and they did a few things right but some of the errors they made were so severe that the results can be safely dismissed from evidence (except to point out the failure of what they were testing).

Also in the third step, post-publication review, they should listen to anyone with a sensible criticism and not this endless deferring to authority. Like:

Because it was:
The Lancet
and
Queen Mary University of London ,

then clearly everything is ok (why would we check?, because, ).

Because it's time for and . For chournalists, Horton, the SMC, QMUL, and KCL.

Not for the patients, of course. They don't eat cake. What do the patients eat, again? Banana leaves, right? Or is it eucalyptus leaves; can never remember.

^end of this is satire, in case not evident. The first part doesn't need to be satire, because unfortunately true.

Edit: specifics are for PACE. I forgot what thread I was on... sorry.

 

[Valentijn]

Do you know what criteria they used to diagnose CFS? I'm guessing Oxford.

Definitely Oxford. And "27 (41%) had successfully been treated for a comorbid disorder beforehand but still met criteria for the chronic fatigue syndrome."

Based on the 20 on antidepressants during the trial for depressive illnesses, I'm guessing all of those "cured" comorbid diagnoses were psychiatric.

 

[WillowJ]

Sorry! I forgot what thread I was on. All this relates to PACE. Feel free to ignore.

Spoiler: criteria problems and lack of efficacy for anyone in PACE

Do you know what criteria they used to diagnose CFS? I'm guessing Oxford. This would allow the patient cohort to contain of a substantial number of patients that didn't actually have CFS but something else, such as an episode of depression.

They used (a modification of?) Oxford and then and only then sub-grouped by Reeves 2005 (Empirical inclusion--a modification which is based on, but destroys the requirements of, Fukuda) and by a modification of the London criteria (which failed to match London criteria, either, I think by failing to require post-exertional malaise, though someone else may be able to explain better). Of course, subgrouping after requiring everyone to have "a main symptom of fatigue" is not the same as independently selecting various groups.

In which case the study woukd demonstrate that exercise is useful for some symptoms of depression.

In which case the study demonstrates that exercise and CBT are fairly useless for most people who have "a main symptom of fatigue" and no blindingly obvious non-CFS diagnosis, who can get to clinic, including those who may be depressed as a primary or secondary diagnosis.

And should not be routinely recommended.

Remember, the bit about people getting better was pure spin. They might have put down on a questionnaire which had no correlation to their actual abilities, that they felt "a little better" or possibly "much better" (but the trial moved "a little better" into some sort of a positive response threshold from being in a neutral threshold, so it's unlikely they had many takers for "much better") or that they could do 1-2 more things slightly better on an questionnaire about how far they could walk or carrying things up stairs, after being pressured to think that they could/should say this. Or their provider may have marked this on their behalf. Or some could walk a few steps further. But even after this kind of maybe-there-maybe-not-there-or-not-important improvement, the mean rating was as poor as people with other very serious diseases.

Questionnaires and everything are explained here (even with a song!):
https://www.youtube.com/user/MEAnalysis/videos

 

[Sean]

DELETED, coz WillowJ.

 

[Sidereal]

Their reply makes no sense I'm afraid. If 20 patients were on antidepressants for relapse prevention, then they would still be considered to have a depressive illness. You don't magically lose a diagnosis of major depression during a period of remission while you're being maintained on antidepressants. Lol.

 

[Valentijn]

Their reply makes no sense I'm afraid. If 20 patients were on antidepressants for relapse prevention, then they would still be considered to have a depressive illness.

Goudsmit made a very good point related to this as well - the antidepressants themselves could be the cause of ongoing fatigue in those patients.

 

[Sidereal]

Goudsmit made a very good point related to this as well - the antidepressants themselves could be the cause of ongoing fatigue in those patients.

Yeah but even putting aside the issue of antidepressants, the fact that their mood is currently ok doesn't mean that they don't have residual fatigue and sleep issues from their underlying depressive illness.