Medical practice

How therapy and exercise 'may help some with CFS'

"Chronic fatigue syndrome patients' fear of exercise can hinder treatment," The Guardian reports.

Chronic fatigue syndrome (CFS) is a long-term condition that causes persistent and debilitating fatigue. We do not know what causes the condition and there is no cure, though many people improve over time.

Treatments for CFS aim to reduce symptoms, but some people find certain treatments help, while others don't.

The news coverage is further analysis of a trial from 2011, which investigated four different treatments for CFS.

This study suggested adding either cognitive behavioural therapy (CBT) or graded exercise therapy (GET) to a person's medical care saw some improvements in their symptoms of fatigue and physical function.

CBT is a type of "talking therapy" designed to change patterns of thinking and behaviour, while GET is a structured exercise programme that aims to gradually increase how long a person can carry out a physical activity.

The current analysis assessed a range of possible factors to see whether these might explain how CBT and GET improved symptoms.

The findings suggested the treatments could be having an effect at least in part by helping to reduce fear avoidance beliefs, such as worrying exercise would make symptoms worse.

However, this study does have limitations, including the fact the researchers have looked at a lot of different possible factors, and some of the statistical associations may arise by chance.

The researchers aim to use these findings to help them improve these treatments or develop new ones.

As the authors make clear, it is important to note this study did not look at what causes CFS.

Where did the story come from?

The study was carried out by researchers from King's College London and other UK universities.

It was funded by the UK Medical Research Council, the Department of Health for England, the Scottish Chief Scientist Office, the Department for Work and Pensions, the National Institute for Health Research (NIHR), the NIHR Biomedical Research Centre for Mental Health at the South London and Maudsley NHS Foundation Trust, and the Institute of Psychiatry, Psychology and Neuroscience at King's College London.

The study was published in the peer-reviewed medical journal, Lancet Psychiatry.

The UK news headlines covering this complex study have all tended to miss the point slightly. The headlines either focus on the already published results (The Independent), or talk about "fear of exercise" exacerbating CFS (The Daily Telegraph and the Daily Mail) or hindering treatment (The Guardian).

This study did not look at what causes or "exacerbates" CFS, or hinders treatment. It assessed how CBT and GET might have improved fatigue and physical function.

It found at least part of the treatments' effects seemed to be down to reducing people's "fear avoidance beliefs", such as worrying exercise would make their symptoms worse.

The Daily Telegraph's suggestion that the study says "people suffering from ME [myalgic encephalopathy] should get out of bed and exercise if they want to alleviate their condition" is particularly unhelpful, and feeds the idea that people with CFS are "lazy": this is not the case.

CFS is a serious condition that can cause long-term illness and disability, and it is not reasonable to suggest people with CFS should simply get up and do some exercise.

People living with CFS need to talk to their doctors about what is appropriate for them and, if an exercise programme is recommended as part of their treatment, that this is done in a structured way. If anything, attempting to exercise before the body is ready to can reverse the rehabilitation process.

What kind of research was this?

This was an analysis of data from a randomised controlled trial of different treatments for CFS, which attempted to investigate how these treatments might work.

The trial was called PACE (adaptive Pacing, graded Activity and Cognitive behaviour therapy; a randomised Evaluation trial). It compared four different treatments in 641 people with CFS:

  • specialist medical care alone
  • specialist medical care with adaptive pacing therapy, which involves balancing periods of activity with periods of rest
  • specialist medical care with cognitive behaviour therapy (CBT)
  • specialist medical care with graded exercise therapy (GET)

These treatments are described in more detail in our analysis of this study from 2011.

It found adding CBT or GET to medical care gave moderate improvements in physical function and fatigue compared with medical care alone.

In this study, researchers wanted to see if they could identify what factors (mediators) CBT and GET might be influencing to give rise to these improvements.

The researchers had planned these "secondary" analyses of the PACE trial in advance, so they were able to collect all the relevant data they needed during the trial.

This is a more robust approach than carrying out ad hoc analyses after the study is completed. These secondary analyses tend to be used to generate hypotheses that can be further investigated in future studies.

What did the research involve?

The researchers carried out analyses of the PACE trial data to identify possible mediators (factors than can influence the effectiveness of treatments).

This essentially involved looking at whether the effects of CBT or GET were still statistically significant if the researchers adjusted for the potential mediators in their analyses.

The idea is that if CBT or GET work by changing one or more of the mediators, adjusting the analyses to essentially "remove" changes in these mediators will also reduce or remove the effects of CBT or GET on the outcomes.

They also looked at the effect of CBT and GET on these mediators, and the relationship between the mediators and the outcomes.

At the start and various other points during the PACE trial, the researchers measured certain factors they thought could be potential mediators.

Most of these mediators were measured using the Cognitive Behavioural Responses Questionnaire (CBRQ), while a few were measured using specific tests.

These factors included the level of participants':

  • fear avoidance beliefs – such as being afraid exercising would make symptoms worse
  • symptom focusing – thinking a lot about symptoms
  • catastrophising – such as believing they would never feel right again
  • embarrassment avoidance beliefs – such as being embarrassed by symptoms
  • damage beliefs – such as the belief that symptoms show they are damaging themselves
  • avoidance or resting behaviour – such as staying in bed to control symptoms
  • all-or-nothing behaviour – behaviour characterised by periods of high activity and subsequent long periods of resting
  • self-efficacy – feelings of control over symptoms and the disease
  • sleep problems – measured using the Jenkins Sleep Scale
  • anxiety and depression – measured using the Hospital Anxiety and Depression Scale (HADS)
  • fitness and perceived exertion – measured using a step test
  • walking ability – measured as the maximum distance a person could walk in six minutes

For their analyses, the researchers took into account the participants' level of these mediators 12 weeks into the trial. The exception was the walk test, which was assessed at 24 weeks.

The researchers also looked for mediators of the effect of CBT and GET at 52 weeks. These outcomes were measured using the physical function subscale of the Short Form (SF)-36 and the Chalder Fatigue Scale respectively.

Individuals with missing data were excluded from the analyses. The researchers also adjusted for a range of potential confounders in their analyses.

What were the basic results?

The researchers found fear avoidance beliefs appeared to be the strongest mediator of the effects of both CBT and GET on physical function and fatigue compared with specialist medical care. It seemed to account for up to 60% of their effect on these outcomes.

For GET, adjusting for participants' increase in exercise tolerance (how far they could walk in six minutes) substantially reduced the effects of GET, but not CBT.

A number of other factors also seemed to be mediators of CBT or GET (compared with specialist medical care alone or adaptive pacing therapy), but the effects tended to be smaller. Fitness and perceived exertion did not appear to be mediating the effects of treatment.

How did the researchers interpret the results?

The researchers concluded fear avoidance beliefs were the most important mediators of the effects of CBT and GET.

They say that: "Changes in both beliefs and behaviour mediated the effects of both CBT and GET, but more so for GET."


This study has tried to pick apart how cognitive behavioural therapy (CBT) and graduated exercise therapy (GET) affected fatigue and physical function in the PACE randomised controlled trial (RCT).

Its findings suggest this could partly be a result of CBT and GET reducing fear avoidance beliefs, such as the fear that exercise would make symptoms worse. But these treatments were less effective in cases where fear avoidance beliefs remained.

The researchers also identified other factors (mediators) that seemed to be playing a role, such as GET increasing the maximum distance an individual could walk in the six-minute walk test.

The advantages of the study include that this is a pre-planned analysis of an RCT, as well as the fact that after the treatments were started, mediators and outcome were measured in temporal order (i.e. “one after the other”). The latter means that it is possible that the treatments are influencing the mediators, which are then influencing outcomes.

The authors acknowledge that the outcomes were showing changes by 12 weeks when the mediators were measured, so it is possible that they were both affecting each other. However, without measurements of the mediators before 12 weeks they were not able to look at this more closely to see if they could be certain which change came first.

The study only measured some potential mediators, and the authors note they could not rule out the possibility unmeasured factors are influencing the results. They did adjust for a range of confounders to try to reduce this chance, however.

Another potential limitation was the main analysis excluded participants with missing data. This is appropriate if those with missing data are missing at random, but if particular types of people – such as those for whom the treatments are not working as well – are more likely to be missing data, this can bias the results.

The researchers did a separate analysis that included incomplete data to look at whether this might be a problem, and this did not differ very much from the original analysis. This suggested missing data was not having a large effect.

The analyses also only included mediators and outcomes assessed at one point, although they were measured multiple times. The authors say they are analysing this additional data, as well as looking at the mediators together, rather than singly. They say the multiple analyses may have made it more likely some of their significant findings were down to chance.

NHS Attribution