Graded Exercise Therapy for ME... or You?

A critique of the Powell et al methodology

By Farrah Tate

This article was written to accompany the March 2001 Talking Point article by Paul Leverenz and Farrah Tate. Whilst reading these comments, please refer to a copy of the following reference article:

Why have we focused so much attention on this one article? It is important that our members are aware of the current trend in psychiatric research, focussing on Cognitive Behaviour Therapy (frequently abbreviated simply as CBT). Persons with ME/CFS must demand that peer-reviewed and respected journals only publish studies based on sound scientific protocol. In the future, it should be our group objective to submit a response to any further papers like Powell et al paper as an incorporated body (the ME/CFS Society of SA).

Subject Recruitment and Criteria

Study participants were recruited by the Oxford criteria. Unfortunately, this criteria does not diagnose ME/CFS patients exclusively, but rather may allow inclusion of patients who have the symptom of “chronic fatigue” as a result of various other illnesses and disorders. Particularly, patients with mental and primary depressive orders may be mistakenly “diagnosed” as ME/CFS sufferers by these criteria. As Bagnall (2001) suggests, there is evidence for this occurring in the present study, whereby a number of patients were excluded (after their primary selection according to the Oxford criteria) for various reasons, including psychiatric exclusions. Moreover one consultant physician confirmed the diagnosis and we have not been advised whether this physician has a background knowledge of, and experience with, the illness of ME/CFS. I personally find it difficult to accept that one individual alone (per patient) could constitute such a major diagnosis, particularly when the criteria are so subjective. Perhaps a panel including ME/CFS experienced physicians (and possibly other health professionals) would have provided a more valid and accurate diagnosis.

David (2001) provides an interesting discussion of this in her Rapid Response, and reminds us that an expert CDC diagnostic criteria working group (1994) rejected the Oxford criteria. Of course, it must not be forgotten that there has also been much discussion regarding the validity of the CDC criteria. David’s literature observations indicate that the Oxford criteria are conveniently used and “widely accepted” by adherents to the CBT model. David goes on to state “This definition is used for one reason and for one reason only. In practice, they tend to exclude people with any kind of disease process (ongoing infections, neurological problems etc.), leaving psychiatrists and psychologists with a sample dominated by individuals whose health is likely to improve with CBT.”

Yet another potential flaw in subject recruitment is the source of patient recruitment: a “chronic fatigue” clinic. Again, the cross-section of patients utilising this clinic may be far from representative of a true ME/CFS patient population. If this were the case, accurate subject selection would depend even more on valid diagnostic criteria, which as stated, were not applied. This misrepresentation is further evidenced by the fact that at base-line a quarter to a half of “CFS” patients were “working” (refer to Table 2). Clark (2001) suggests that this is “an indication that the study was biased toward those with milder forms of the illness.” We know – through associating or working with patients with ME/CFS – that typical “work” is out of the question for most individuals who have not entered a recovery phase of the illness. It is critical that more research is carried out (and by non-psychiatric researchers) to determine a true indication of the levels of disability (as quantified by objective outcomes such as specific measures of working capacity).

Another frequent problem – and one with significant ethical basis – also noted in the current publication methodology is that, as Clark states, “Those with the severest symptoms have been consistently ignored in research and are least likely to access the few specialist services that do exist.” He continues with “We call for clinicians to take responsibility for those most profoundly affected.” Certainly the sickest individuals (long term and acute in the initial stages of the illness) are significantly marginalised. Surely this contributes to, and continues to perpetuate, the myth that there is no organic basis to illness of ME/CFS. Moreover, the exclusion from the study of patients confined to wheel chair or bed shows complete ignorance of the illness; ME/CFS certainly may fluctuate to the degree where people attain different degrees of limited mobility on a monthly or even weekly basis.

According to this deficiency in recruitment criteria and protocol, perhaps the title of the paper should be aptly renamed to “Randomised controlled trial of patient education to encourage graded exercise in patients with chronic fatigue as a symptom” or “.... patients with symptomology meeting the Oxford diagnostic criteria.”


Even though patients were randomised into the three levels of intervention and control groups, as Chaudhuri (2001) comments, there is no data presented in the paper to provide evidence that patients in each intervention group were adequately matched for levels of physical fitness before entering the study. One would usually consider randomisation sufficient to eliminate the risk of directed bias BUT base-line data would have been useful to validate this assumption. Further, comparison of base-line measurements with post-intervention measurements could have provided an important indicator of improved exercise capacity or fitness. Of course, it must be noted that such measurements are expensive, require substantial technician skill, and hence may not have been practically possible, even if desired by the authors.

Treatment Conditions

Chaudhuri (2001) states: “Frequent early contacts with patients in the three intervention groups (and not in the control group) might have confounded the outcome measures by positively influencing the results. This view is certainly supported by the difference between the control and the three intervention groups (and lack of difference between the individual intervention groups) emerging at the end of three months, with little changes thereafter. By speaking to their patients, Powell et al might have provided them with a coping strategy that the control group could not access.” Therefore, improvement cannot be stated as unequivocally being a result of the exercise therapy or indeed educational intervention alone. The latter is unlikely as Clark (2001) adds: “Had education itself been the significant intervention at least some differences between the groups could have been expected.”

One also wonders why an “activity diary” was not mentioned in the methodology. If not completed, there is less evidence to support the assumption that activities were carried out as instructed – i.e. with reference to type, intensity and duration. This would seem particularly important since measurements of physical fitness were not alluded to, and therefore were presumably not carried out.

Outcome Measures

Much discussion has ensued regarding the problems associated with the “fatigue scale” developed (conveniently) by Hickie and colleagues, and the limitations of the MOS subscale, both used to assess outcome (David 2001; Goudsmit 2001). In general, both these measures are particularly subjective and cannot differentiate between patients with CFS and depression, or between degrees of fatigue. David (2001) mentions: “According to Bentall et al, the physical functioning subscale scores ‘directly address this issue.’ They don’t. They do not cover avoidance as much as perceived limitations in terms of running, walking at least one block, climbing at least one flight of stairs, moving a table, lifting or carrying groceries, go bowling, and bathe/dress oneself. The scores do not indicate whether these or any other activities were reduced by 5%, 15%, or by 50%.” [Note: ‘this issue’ (as stated above) refers to the difference in avoidance of activity at base-line and after intervention for treatment groups as compared with the control group.]

Accordingly, the only symptoms assessed as intervention outcome measures were fatigue, emotional distress and sleeping problems. Many ME/CFS patients could think of numerous other symptoms (including cognitive dysfunction, sore throat/glands and orthostatic intolerance) that might warrant measurement through significant levels of disablement imposed. A more appropriate outcome measure would have been to assess how many intervention treated subjects were capable of performing specific (and realistic to employment potential) working tasks at the end of the trial as compared with the baseline at the beginning of the intervention. A comparison with the controls (not receiving any level of the intervention treatment) should give some indication of usefulness of this intervention in establishing workability. Given author emphasis on commenting that the therapy is “shorter,” “requiring less therapist skill” than CBT and “cost effective” it would seem logical that they make some comment regarding out-come in terms of the level of labour/working capacity restored. After all, this is a very important component of government assessment of the extent of the cost of an illness to the country. Whether this workability is sustainable (and is therefore cost effective in the long-term) is another question altogether – certainly patients should have been followed up several times further, many months beyond the completion of the intervention.




Copyright ME/CFS Australia (SA) Inc

Reprinted from