The GETSET Trial: Letters to The Lancet and a response from the authors | 26 March 2018

March 26, 2018


GETSET was a clinical trial in secondary care of a new approach for people with ME/CFS and it was published in the Lancet on 22 June 2017.

The trial used the principles of graded exercise therapy, delivered in a self-help booklet, with up to four sessions of assistance offered remotely by trained physiotherapists via phone or Skype.

A total of five letters have now appeared in the Lancet critiquing elements of the trial, and the authors have now issued a response to all of them (below).

ME Association position

In view of the fact that the results from this clinical trial are going to form part of the research evidence that NICE will be examining – in relation to the new clinical guideline – ME Association trustees decided that we should make a very careful and thorough response to the study. We published our review on 28th June 2017.

ME Association Review of GETSET

For the benefit of anyone who just requires a summary of the key points and concerns that we made, our main criticisms were:

  1. GES (Graded Exercise Self-help) had little or no effect on physical function and did not result in participants claiming that GES had any significant overall effect on their ME/CFS
  2. The study design – i.e. addition of a new primary outcome measure – was altered after the trial had started
  3. Patients were recruited using the very broad NICE diagnostic criteria for ME/CFS – which meant a significant proportion did not meet Fukada and are unlikely to have met the Canadian diagnostic criteria for ME/CFS
  4. The trial recruited an unrepresentative number of people who were likely to have been in the mild category of illness severity, could exercise, and were enthusiastic about taking part.  This is not representative of the overall ME/CFS population
  5. The trial relied on self-reported and subjective outcome measures rather than employing any form of objective outcome measure (e.g. use of an actometer) for measuring changes in physical activity levels
  6. Examining and reporting on the results after only 12 weeks of treatment is far too short to draw any meaningful conclusions about the efficacy and safety of GES
  7. The study was not blinded – for obvious reasons. However, we believe that this sort of approach to activity management, where there has to be considerable individual variation in approach, is not really suitable for assessment in randomised controlled trials that rely on subjective outcome measures

The Lancet – Graded exercise self-help for chronic fatigue syndrome in GETSET

Authors' reply

Lucy V Clark, Francesca Pesola, Janice M Thomas, Mario Vergara-Williamson, Michelle Beynon, Peter D White.

Published: 24 March 2018

In the graded exercise therapy guided self-help treatment (GETSET) trial,1 we found that addition of guided graded exercise self-help (GES) to specialist medical care (SMC) safely improved fatigue and physical functioning more than did the comparison treatment of SMC alone.

Frank Twisk suggests that GES is not a rehabilitative intervention as the GES group did not return patients to normal levels of fatigue and physical functioning, and Joan Crawford is unimpressed with the effect size of GES for improving functioning. We suggest that an improvement in the main symptom of fatigue with an effect size of 0·53 is clinically useful.1 We have already acknowledged the small size of the effect on physical functioning (0·20), but our finding that the effect size was greater in those with the worst baseline physical functioning suggests this might represent a ceiling effect.

Therapists reported that 29% of patients did not adhere more than “slightly” to the exercise programme, despite 88% attending at least 75% of their guided support. This area requires more research to understand non-adherence and how it might be improved.2

We used the National Institute for Health and Care Excellence (NICE) definition of chronic fatigue syndrome as these are the commonest criteria used by UK clinicians in secondary care.3 A subgroup analysis using two other definitions of chronic fatigue syndrome made no significant difference to the findings. There were no significant differences in safety outcomes between treatment arms once missing data were accounted for.

Anna Wood is concerned about adherence to GES and missing data; both were reported in the paper and neither significantly altered our findings or conclusions.1 The use of outcome assessments by mail and available resources prevented the use of objective measures, such as actigraphy, as suggested by Crawford. We suggest that using patient-reported outcome measures is sensible in symptom-defined illnesses such as chronic fatigue syndrome. The largest trial of pacing therapy suggested that it had a worse, not better, safety profile when compared to more active rehabilitation interventions similar to GES.4

In response to Robert Saunders, we can confirm that all patients reported post-exertional malaise (PEM) and/or fatigue at baseline (as is required to meet NICE criteria for chronic fatigue syndrome); no patients were excluded based on having PEM. A previous trial showed that post-exertional fatigue improved more after graded exercise therapy compared to after both SMC alone and pacing therapy.5The lack of a significant difference in safety outcomes between treatment arms reinforces the safety of GES for chronic fatigue syndrome.

Karen Kirke is concerned about severely affected patients with chronic fatigue syndrome. We acknowledge that all participants were required to attend a clinical assessment, which would exclude those that were housebound.1 That being said, use of teletherapy for patients too unwell to attend clinic is an innovative approach that merits further research.6, 7

PDW reports grants from the UK National Institute of Health Research and the Sue Estermann Fund during the conduct of the study, and personal fees from Swiss Re-insurance company outside the submitted work; he is an appointed member of the Independent Medical Experts' Group, a non-governmental body, which advises the UK Ministry of Defence about its Armed Forces Compensation Scheme. PDW provided unpaid advice to the UK Department for Work and Pensions until 2015. All other authors declare no competing interests.

The Lancet – Letters of concern

Frank Twisk

Published: 24 March 2018

The results of the guided graded exercise self-help trial (GETSET) by Lucy Clark and colleaguesreaffirm that graded exercise therapy (GET) is not a rehabilitative treatment for chronic fatigue syndrome.2

Although fatigue (measured by Chalder Fatigue Questionnaire [CFQ]) and physical functioning (assessed using the Short Form-36 physical function [SF-36 PF] subscale) improved in the guided graded exercise self-help (GES) group (and the non-intervention group), these effects were by far insufficient to come close to the normal levels defined previously3 by one of the GETSET authors (CFQ ≤6 and SF-36 PF ≥85).

The observation that ”physiotherapists reported that […] 42% adhered to GES completely or very well, […] 30% moderately well, and […] 29% slightly or not at all”1 makes far-reaching conclusions impossible.

Also, the trial1 selected patients using National Institute for Health and Care Excellence criteria, rather than the commonly used Fukuda criteria for chronic fatigue syndrome.4 Although limiting the analysis to participants who met these Fukuda criteria did not change the results, the authors also state that the ”relative absence of intervention by diagnostic subgroup interactions might be related to the study not being powered to detect all interactions“.1

Finally, the conclusion that GET is a safe intervention is at odds with another GETSET publication,which reported that a chronic fatigue syndrome patient subgroup deteriorated due to GET and observed that the “deteriorated group reported experiencing more barriers to GET, including a worse exacerbation of symptoms in response to GET”.5

Overall, the conclusion that GET is an effective and safe treatment for chronic fatigue syndrome is not substantiated by the outcomes.

I declare no competing interests.

Joan Crawford

Published: 24 March 2018

The mean score for physical functioning (measured by the Short Form-36 [SF-36] subscale)1 for people aged 35–44 years is 93·3 (SD 13·4).2

Participants in the GETSET trial by Lucy Clark and colleagues3 had a mean age of 38·1 years in the active treatment arm (guided graded exercise self-help [GES] group) and 38·7 years in the control group.

After the short 12-week intervention, participants' mean self-reported physical functioning score using the SF-36 scale was 55·7 (23·3) in the GES group, an increase from 47·3 (22·2) before the intervention, compared with a slight increase in physical functioning in the control group from 50·1 (22·6) to 50·8 (25·3).3

Clark and colleagues state that “GES significantly improved fatigue and physical functioning compared with SMC [specialist medical care] alone”. However, this modest change in a self-reported outcome measure in a non-blinded trial, in which bias is highly likely without use of objective measures of activity, is extremely unlikely to translate to a clinically meaningful improvement for patients. Nor does this change return them anywhere close to healthy and robust good physical functioning compared with controls, which patients heartily desire.

Overall, these observations suggest such modest improvements are unlikely to be due to countering deconditioning and exercise avoidance as proposed by the authors, but are much more likely to be accounted for by nothing more than the placebo effect, in other words a desire by patients to please their therapists.

Moreover, there was no attempt during this trial to obtain measurable objective activity levels to ascertain if patients increased exercise and activity or complied with the intervention at all.

I declare no competing interests.

Anna Wood

Published: 24 March 2018

The CONSORT statement on harms notes that “it is important to report participants who are non-adherent or lost to follow-up because their actions may reflect their inability to tolerate the intervention.”1

It is thus welcome that Lucy Clark and colleagues2 report physiotherapist-rated data on adherence; they considered that only 42% of participants adhered to guided graded exercise self-help (GES) completely or very well.

The protocol notes that to “measure departure from intended treatment, participants will be asked at follow-up whether they adhered to the booklet and guidance, and how much PA [physical activity] they undertook in the past week.”3 It would be useful if Clark and colleagues could now also publish such data.

A review of eight surveys (n=4338) found 51% of patients with chronic fatigue syndrome or myalgic encephalomyelitis reported being harmed by graded exercise therapy (GET).4 Unsurprsingly, such findings have led to concern about GET's safety. It would therefore be preferable if researchers also used and reported on other objective tools to measure adherence, such as actometers and heart rate monitors, which are much more readily available now than in the past.

According to the GES Therapist Manual “a central concept of GET and GES is to maintain exercise as much as possible during a CFS/ME [chronic fatigue syndrome/myalgic encephalomyelitis] setback.”5More adherence data could help investigate whether the lower rates of harm found in research studies are due to patients not complying with this central concept and instead using pacing, which has been associated with a lower rate of harm in surveys.4

I declare no competing interests.

Robert Saunders

Published: 24 March 2018

I was surprised to read that the GETSET trial by Lucy Clark and colleagues1 excluded participants who had physical contraindications to exercise, as such a criterion would appear to exclude anyone suffering from post-exertional malaise (PEM).

In its 2015 report,2 the US Institute of Medicine concluded: “There is sufficient evidence that PEM is a primary feature that helps distinguish ME/CFS [myalgic encephalomyelitis/chronic fatigue syndrome] from other conditions”. Indeed, post-exertional neuroimmune exhaustion is a compulsory requirement for a diagnosis of myalgic encephalomyelitis under the International Consensus Criteria.3

The disputed National Institute for Health and Care Excellence criteria that were used in the GETSET trial1 are less precise, stating that “post-exertional fatigue and/or malaise” must be present for a diagnosis of chronic fatigue syndrome or myalgic encephalomyelitis to be considered, whereas “physical or mental exertion makes symptoms worse” is only listed as a possible symptom, not a requirement.4

Whether Clark and colleagues interpreted post-exertional fatigue to mean the abnormal physical responses to exertion that have been reported in other myalgic encephalomyelitis and chronic fatigue syndrome studies is unclear.5 If they did, it would appear that the GETSET participation requirements were self-contradictory. If they did not (as the inclusion of patients with high baseline Short Form-36 physical function scores suggests), this should be made clear.

Nevertheless, given the inherent bias of unblinded trials that rely on subjective outcome measures,6with less than one in five participants having reported any benefit from guided graded exercise self-help, the evidence GETSET provides that this strategy is unlikely to be beneficial to most people diagnosed with chronic fatigue syndrome (however it is defined) should not be overlooked.

I declare no competing interests.

Karen D Kirke

Published: 24 March 2018

Lucy Clark and colleagues1 (July 22, 2017, p 363) reported that guided graded exercise self-help (GES) for chronic fatigue syndrome was “more useful in those with worse physical functioning”, defined as a Short Form-36 physical function (SF-36 PF) score of 40 or less.

As follow-up GES sessions were done by Skype and telephone, this result might suggest to some clinicians the possibility of modifying and extending this intervention to patients with severe and very severe chronic fatigue syndrome, all of whom are housebound and some of whom are bedbound. Indeed, the accompanying commentary acknowledges that patients with severe chronic fatigue syndrome were ineligible for this secondary-care trial, but suggests that GETSET shows that “graded exercise is effective even in individuals with severe symptoms.”2

GETSET's sample was higher-functioning than in previous trials, with a baseline mean SF-36 PF score of 49, compared to PACE's 38 and FINE's 30.1, 3, 4 Housebound patients have been reported to have a mean SF-36 PF of 17.5

Extrapolating GETSET's reported modest short-term gains for a small number of higher-functioning patients to the severe and very severe group is likely to be inappropriate. When trials included some severe patients, such as FINE's 11% non-ambulatory patients, no short-term or long-term benefit to physical functioning for rehabilitation over usual care was found.4 When patients undertook graded exercise therapy with chronic fatigue syndrome specialised therapists, 39% of severe to very severe patients reported being very much worse afterwards, compared to 22% of mild to moderate patients.6

Available evidence suggests that with interventions aimed at increasing activity, severe to very severe patients with chronic fatigue syndrome do worse, not better, than their milder counterparts.

I declare no competing interests.


7 thoughts on “The GETSET Trial: Letters to The Lancet and a response from the authors | 26 March 2018”


    Peter White and his colleagues essentially define ME/CFS as an *objective* reduction in physical activity brought about by erroneous cognitions and consequent behaviour. It is therefore self-contradictory for White et al to say “We suggest that using patient-reported outcome measures is sensible in symptom-defined illnesses such as chronic fatigue syndrome.”

    Apparently White et al want to define ME/CFS objectively but measure it subjectively. It is noteworthy that they desire their behavioural-deconditioning concept of ME/CFS to have an objective status but wish to avoid the improvement they allege derives from their GETSET study being measured objectively (and therefore more stringently). To my mind that’s trying to have your cake and eat it.

    However, in the previous sentence White et all say “The use of outcome assessments by mail and available resources prevented the use of objective measures, such as actigraphy, as suggested by Crawford.” So are they saying they would have *liked* to use objective measures but could not because of resource constraints and reporting methodology? If so then what sense does it make to then immediately argue that subjective measures are the better choice?

    One can only conclude that the shifting around and inconsistencies in White et al’s position on the crucial issue of how outcomes are measured reflects, in the final analysis, a lack of integrity in their approach to research.

  2. peter white has multiple and very serious conflicts of interest.
    these on their own should call into question the validity of the study
    jeremy bearman

    1. I agree Jeremy. His conflicts of interest and determination to suppress the data of the PACE study make it impossible to believe he has the necessary objectivity and neutrality to conduct credible research into ME/CFS.

  3. Don’t forget that a part of the NICE/CFS patient group got worse after GETSET:

    Cheshire, A, Ridge, D, Clark, L, and White, P.
    Why patients with chronic fatigue syndrome/myalgic encephalomyelitis improve or deteriorate with graded exercise therapy.
    J Psychosom Res. 2016; 85: 59

    This study confirms GET is not only effective, but even potentially harmful.

    1. Frank, I think you meant to say “GET is not only INEFFECTIVE, but even potentially harmful.”

    2. Indeed Frank. Also on White’s premises CBT-GET is unfalsifiable: “There were substantially different treatment recommendations between patient support organisations and medical sources. Since expectations can determine response to treatment, these different views may reduce the engagement in and effectiveness of rehabilitative therapies recommended by national guidelines and supported by systematic reviews.”

      White’s position appears to be that if GET works then it’s effective, and if it doesn’t work then the patient’s expectations are to blame. This means that for White there is no experiment whatsoever that could falsify GET. It’s hard to credit that this could really be the position of a professional scientist, but that’s how it appears.

  4. “We have already acknowledged the small size of the effect on physical functioning (0·20), but our finding that the effect size was greater in those with the worst baseline physical functioning suggests this might represent a ceiling effect.”[/QUOTE]
    It is disappointing that there are again making this point without sharing information about the other subgroup i.e. with the higher baseline scores.

    I tried to deal with this in my (unpublished) letter.
    (Copied from an earlier message)
    I decided to focus on a specific point that I did not see others raising in discussions. Most of my other letters use more references:

    [QUOTE]Response to: Guided graded exercise self-help plus specialist medical care versus specialist medical care alone for chronic fatigue syndrome (GETSET): a pragmatic randomised controlled trial

    The SF-36 physical functioning (PF) outcomes for those in the guided graded exercise selfhelp (GES) group who had a baseline SF-36 PF score ≥45 (call them group B) must have been particularly poor in the GETSET trial.1 We are told that those with a baseline score ≤40 (group A) made up approximately 40% of the sample and ended with average score of 56.9. This means the average outcome for group B, the higher functioning group at baseline, was actually lower, at around 54.9. Also, by definition, group A increased by an average of at least 16.9 (56.9-40). However the whole sample only increased by an average of 8.4. This means that an upper bound on the average increase for group B would be only approximately 2.7, in comparison to the increase of 16.9 for group A. This is an extreme scenario and the difference in improvements was most likely higher than 14.2. It would be interesting if Clark and colleagues could give the exact figure so everyone would be aware of the magnitude of the difference in the response.

    Clark and colleagues say the poor results may be due to a ceiling effect. More than 90% of healthy working-age people score 90 or more.2 Therefore, the mean score of 54.9 for group B and 55.7 overall suggests that if there is a ceiling in the effectiveness of GES, it is a long way below normal functioning. I do not believe this was made clear to readers.

    Tom Kindlon

    Competing interests: I work in a voluntary capacity for the Irish ME/CFS Association.


    1. Clark LV, Pesola F, Thomas J, Vergara-Williamson M; Beynon M, White PD. Guided graded exercise self-help plus specialist medical care versus specialist medical care alone for chronic fatigue syndrome (GETSET): a pragmatic randomised controlled trial. The Lancet. June 22, 2017 doi:10.1016/S0140-6736(16)32589-2

    2. Wilshire CE, Kindlon T, Matthees A, McGrath S. Can patients with chronic fatigue syndrome really recover after graded exercise or cognitive behavioural therapy? A critical commentary and preliminary re-analysis of the PACE trial. Fatigue. 2017;5:1–4.[/QUOTE]

    The journal did take a long time to reply (over 6 months): I am left wondering whether they asked the investigators for their opinion. I could imagine the researchers might try to put the journal off the letter.

    It is now on PubPeer

    There’s a reasonable possibility those in the group with an initial score of 45+ on the SF-36 physical functioning subscale actually decreased on average.

Comments are closed.

Shopping Basket