Cochrane Review on Exercise Therapy for CFS
Edmonds et al. investigated the effectiveness of exercise therapy as a treatment for CFS in a 2004 Cochrane Review (1). The analysis included 5 randomised controlled studies. The primary measure being assessed was fatigue, according to the Chalder Fatigue Scale. The Cochrane Review concluded that some CFS patients may benefit from exercise therapy. The review stipulated that CFS patients who are similar to those included in the 5 studies should be offered exercise therapy. The review speculated that higher intensity exercise may lead to a poorer outcome and that 40% VO2 max exercise may be preferable to 75% VO2 max exercise.
The Cochrane Review emphasised that functional work capacity in the exercise group was not statistically significant from the control group after 12 weeks. The review also demanded higher quality studies of exercise therapy that incorporated different types of CFS patient groups. The review also encouraged a larger range of outcomes to be measured. It suggested adverse effects and quality of life be measured. The review also recommended that the outcome measures should be performed over a longer period of time to gauge the long term effectiveness of exercise therapy as a treatment for CFS patients.
The 5 Cochrane Review Examined Studies
The 5 studies that the Cochrane Review analysed were; Appleby 1995 (2), Fulcher 1997 (3), Moss-Morris 2003 (4), Powell 2001 (5) and Wallman 2004 (6). I will analyse each of these 5 studies independently. My analysis of the 5 studies will be distinct from the Cochrane Review, in that it will be more critique orientated as opposed to appraisal oriented.
Appleby et al. (2) used only an Oxford CFS criteria cohort for their exercise study. This cohort contains approximately 10% of CFS patients (When referenced to an International Classification of Disease ME/CFS cohort (7).) Consequentially, approximately 90% of the cohort studied were not true CFS patients.
Appleby et al. divided the study into four distinct groups. Group 1 were assigned GET and Fluoxetine, group 2 did GET and took a drug placebo, group 3 had an exercise placebo and took Fluoxetine and group 4 were given an exercise placebo and a drug placebo. Exercise therapy was performed 3 times a week for a period of 20 minutes. The intensity of the exercise was 75% of the individuals VO2 max.
The study concluded that there was no statistically significant difference between exercise therapy and the control group regarding functional work capacity. This was determined at 12 weeks and 26 weeks. The Appleby et al. study didn’t result in exercise therapy helping fatigue, when compared to the control group. This study also included the highest dropout rate, relative to the other four Cochrane Review studies.
The cohort used in the Appleby et al. study had a median CFS duration of just 2.3 years. This is a relatively short illness duration (for CFS) and hence any conclusions from this study should emphasise the relatively short median CFS duration of the Appleby cohort. Also the mandatory 20 minutes per session exercise therapy is a duration that exceeds the capacity of moderate and severely affected CFS patients. It can therefore be concluded that patients with only mild CFS were included in this Appleby et al. study.
Conclusions: Exercise therapy at 75% VO2 max is not useful for patients fulfilling the Oxford CFS criteria, with mild CFS and a short illness duration. The high dropout rate of the Appleby study may signify that 75% VO2 max exercise is contraindicated in the aforementioned CFS cohort.
Fulcher et al. (3) used only an Oxford CFS criteria cohort for their exercise study. This cohort contains approximately 10% of CFS patients (When referenced to an International Classification of Disease ME/CFS cohort (7).) Consequentially, approximately 90% of the cohort studied were not true CFS patients. 66 Oxford defined, CFS patients were divided into two groups. The first group were prescribed at least 5 sessions a week of GET. The initial sessions lasted between 5 and 15 minutes long and the exercise was graded up to a maximum length of 30 minutes. The first group performed at an intensity of 40% VO2 max, which is approximately 50% of the maximum heart rate. The first group were given the choice of walking, swimming or cycling. An exercise physiologist negotiated with each individual patient regarding adapting the exercise to the patient’s capacity and grading the exercise. The second group were prescribed flexibility and relaxation for 12 weeks. This ‘flexibility and relaxation’ mainly consisted of stretching. The second group performed this routine for 5 days a week starting at 10 minutes a day and increasing to a maximum of 30 minutes a day.
None of the ‘CFS’ patients included in this study had an appreciable sleep disturbance. Having sleep disturbance is a necessary condition to fulfil the Canadian Consensus Criteria (CCC) (8). It can therefore be concluded that none of the patients in this study fulfilled the CCC for CFS. In a De Becker et al. study (9), 94.8% of patients fulfilling the Holmes criteria (10) had a sleep disturbance. This study included 951 patients and those with a sleep disturbance graded it as 2.5 out of 3 in terms of severity. The De Becker et al. study also analysed 1578 patients fulfilling the Fukuda criteria (11) for CFS. 91.9% of the patients had a sleep disturbance, averaging 2.4 out of 3 in severity. It can therefore be concluded that the results of the Fulcher et al. study cannot be extrapolated to those suffering from CCC, Holmes criteria or Fukuda criteria CFS.
The Fulcher et al. study cohort comprised only patients who had been referred to a general hospital department of psychiatry. Patients fulfilling the Oxford criteria with a psychiatric disorder, comorbid disorder and/or a comorbid phobia were included in the Fulcher et al study. 41% of the study participants had in the past been treated for a comorbid disorder. The median illness duration of the study cohort was just 2.7 years which is a relatively short illness period for CFS patients. 5 outpatients that were too ill to attend the psychiatric department were excluded from the study.
It must be noted that patients with severe CFS would not have been eligible for this study due to the intensity and length of exercise required to be completed. Also the exercise group (group 1) started their exercise at 40% VO2 max which is roughly equivalent to 50% of the maximum heart rate. Many Postural Orthostatic Tachycardia Syndrome (POTS) patients and CFS patients would therefore be ineligible for this study due to their heart rate exceeding this percentage at rest. POTS (which often overlaps CFS) has the hallmark symptom of an increase in heart rate by over 30 beats per minute to 120 beats per minute on standing (12). The general formula to predict a healthy person’s maximum heart rate (in beats per minute) is 220-age. CFS patients often have a lower maximum heart rate than those healthy controls of a similar age. It can therefore be concluded that many CFS patients would not be typical of this studies cohort. This is due to many CFS patients exceeding 50% of their maximum heart rate at rest.
The patients contained within the exercise group (group 1) who rated themselves as better, didn’t show a significant improvement in muscle strength or peak oxygen consumption. Also over 40% of the patients in the study showed no or little improvement. Fulcher et al. were even cautious in their conclusion and emphasised that GET should only be prescribed appropriately and not in all circumstances.
Conclusions: Some of those patients fulfilling only the Oxford criteria and not the Holmes criteria, Fukuda criteria or the CCC for CFS may benefit from a certain type of exercise. The specific Oxford criteria patients that may benefit do not have an appreciable sleep disorder and may have a psychiatric disorder, comorbid disorder and/or a comorbid phobia and only have a short illness duration. Some patients within this select subset (which is very much an atypical CFS and unique subset) may benefit from starting GET at 40% of their VO2 max for several minutes a day.
This study isn’t freely available hence I have only read the abstract and not the details of the study. A small amount of information about this study is available in the Cochrane review (although the Moss-Morris study had just been submitted for publication when the Cochrane Review was published.) I will attempt to utilise this information and critique this study albeit in an inhibited manner.
Moss-Morris et al. (4) used the 1994 CDC CFS criteria cohort for their exercise study. This cohort contains approximately 40% of CFS patients (When referenced to an International Classification of Disease ME/CFS cohort (7).) Consequentially, approximately 60% of the cohort studied were not true CFS patients. The Cochrane Review rated the Moss-Morris study as the study that was of the least quality, relative to the other 4 studies reviewed. All 5 studies quality was evaluated by the CCDAN quality rating system which includes 23 different criteria.
The Moss-Morris study divided patients into 2 groups. Group 1 was prescribed GET for 12 weeks while group 2 was told to continue treatment as usual. Group 1 were told to exercise 4-5 times a week at 40% of their VO2 max, for a period of up to a maximum of 30 minutes. Exercising at 40% VO2 max equates to approximately 50% of the individuals maximum heart rate. The same criticisms of study exclusion apply to this Moss-Morris study as the previously discussed Fulcher et al. study. The criticisms that also apply to this study include using a CFS cohort that can reach 50% maximum heart rate while exercising compared to other CFS cohorts exceeding the 50% maximum heart rate while resting or standing. Consequently, this studies result can only be extrapolated onto a specific type of CFS patient.
The median length of illness in the Moss-Morris study was 3.1 years. This is a relatively short period when compared to other CFS cohorts. As a result of this short illness period being studied, the results of this study should not be imposed on those CFS patients with a long or medium illness duration. Only 11 out of 59 CFS patients in this study were unemployed. This is a relatively low number when compared to other CFS cohorts. As over 80% of this study cohort were working, it implied a level of CFS disability in this cohort that is mild as opposed to moderate or severe. The results of this study are therefore reflective mainly of mildly affected CFS patients.
The Moss-Morris study didn’t report on the CFS patients doing GET after 12 weeks. Consequently, the long term affects of GET have not been determined by this study. Also this study fails to measure many symptoms that are central to CFS such as pain, sleep, neurological symptoms and endocrine symptoms. The general effect of GET on CFS patients is therefore nebulous as the results of this study exclude the central components and potential biomarkers of CFS.
Moss-Morris et al. conclude that the reason GET was a partially effective treatment in this study may be due to “self-reported improvement by reducing the degree to which patients focus on their symptoms.” The authors continue to state that “there is no evidence that CFS patients disproportionately focus on symptoms, rather that common CFS definitions fail to describe patients’ symptomology adequately.”
Conclusions: I am not qualified to draw any definitive conclusions from this study as I have not read the study in its entirety, but rather secondary sources about the study. Having said that, I have cautiously determined that the results of this study mean that: CFS patients fulfilling only the CDC 1994 criteria, with mild CFS and CFS of a short duration may benefit from performing low (approximately 40% VO2 capacity) intensity GET. The benefit this specific subgroup will possibly gain may be due to “reducing the degree” they “focus on their symptoms” as the study authors state. This may only be a perceived and not a physiologically significant improvement.
Powell et al. (5) used only an Oxford CFS criteria cohort for their exercise study. This cohort contains approximately 10% of CFS patients (When referenced to an International Classification of Disease ME/CFS cohort (7).) Consequentially, approximately 90% of the cohort studied were not true CFS patients. Patients were divided into four groups. Group 1 were told to continue treatment as usual. Group 2 were the “minimum intervention group” and included two face to face sessions with a health professional totalling 3 hours, in combination with a prescription of GET. Group 3 were the “telephone intervention group” which involved 7 telephone calls of 30 minutes each with a health professional. Also GET was assigned to group 3. Group 4 were the “maximum intervention group” which entailed the same involvement as the “minimum intervention group” in conjunction with 7 one hour face to face sessions with a health professional. This Powell et al. study was not a pure GET study but rather a hybrid GET and cognitive behaviour therapy (CBT) study.
Patients were excluded from this study if they were confined to a wheelchair or bed which implies that the severely affected CFS cohort were barred from this study. Patients were also excluded from the study if they were taking any treatments. The paper doesn’t describe very articulately what treatments resulted in patient exclusion from the study, only to stipulate “other treatments.” Presumably this refers to any treatments outside of the realm of GET and CBT. A large portion of CFS patients are dependent on specific medications (and in most cases, many medications), hence a CFS cohort not taking any treatment is an atypical CFS cohort. The notion that only mildly or moderately affected patients were included in this study gains support when the intervention group criteria is analysed. This included up to 10.5 hours in consultations on top of the GET schedule. This temporally immense consultation duration would exceed the limits of severely affects CFS patients and some moderately affected patients. This is further evidence that the CFS patients in this study were only comprised of a unique subgroup.
The study originally selected 160 patients to be assigned to one of the 4 groups. Twelve of the eligible patients out of the 160 “refused to participate.” This makes the subgroup being studied even more atypical due to the possibility that these non-included patients had previously suffered from the effects of GET. Out of the 148 patients that continued to participate in the study, 21 dropped out which equates to 14%. This dropout number was not reflective in the published results of the study and such a high dropout rate could completely skew the results.
The Powell et al. study concludes that the improvement as a consequence of the treatment can be explained in terms of the changes of behaviour by the CFS group. This was due to the “physiological explanations they were given for their symptoms.” Prior to treatment, 81% of patients believed their condition was due to a virus however post treatment, only 23% maintained this belief. Prior to treatment, 67% believed their condition was due to a physical illness while 13% held this belief after treatment. 15% thought their condition was due to physical deconditioning prior to the study, while 81% held this view post treatment.
These results are not entirely surprising as the 3 intervention groups were told that their illness was not organic ad nauseum. The purpose of involving the intervention groups in this study was to convince them that their illness was psychosomatic and a result of deconditioning. This study did not reference any other organic illness studies in which the patients were told that their illness was psychosomatic. I suspect that regardless of the illness, organic or psychosomatic, that when health professionals continue to state that the illness is psychosomatic, a large portion of patients will believe this. This is especially true for studies such as this Powell et al. study in which success of treatment can be measured in terms of changed illness beliefs. If a study was to be conducted on cancer patients, with an identical methodology to the Powell et al. study, yet the cancer patients hadn’t been told of their diagnosis, a similar result to the Powell et al. study is likely. This is due to the suggestible nature of humans relating to appropriate authorities diagnosing illnesses. Also the study participants will be deemed to have not changed their illness beliefs and hence “failed” if they do not give an improved subjective outcome score. The Moss-Morris et al. (4) study authors support my argument by stating that alike studies may appear to be effective due to “self-reported improvement by reducing the degree to which patients focus on their symptoms.”
To clarify the above argument, let me demonstrate a hypothetical example from the Powell et al. study. As an individual in the study, I enter the study with the belief that my virus is organic. I am told by the medical professionals that this is not the case. I am told that my illness is psychosomatic and by not focusing on my symptoms, my illness will dissipate. Correspondingly, when I evaluate my symptoms as part of the study evaluation, I have two fundamental options. 1. I can state that my symptoms have not changed however this will indicate that I did not believe what the health professional told me and hence I did not complete the treatment correctly. 2. My success will be gauged upon stating that my symptoms have improved. I have been told that my symptoms are essentially imaginary hence if I write that my symptoms have improved, I am simply fulfilling what was asked of me by the health professional. This type of study is full of cognitive biases and to scientifically draw conclusions from it, a reference study is needed such as the hypothetical cancer study I have mentioned. Powell et al. have failed to provide a reference study to determine whether the same interventions with definitively diagnosed organic illness patients will yield different results. This is the vicious cycle that in the current paradigm renders CBT studies non-calibrated. Until this calibration has occurred, studies like this Powell et al. study are pseudoscience in nature.
A further criticism of this Powell et al. study involves the lack of information regarding what type of GET was performed. All of the other studies investigated by the Cochrane Review provide details pertaining to the type, duration and intensity of GET performed. As this information has been omitted, any conclusions drawn from this study are spurious. The study authors’ concede that one of the limitations of this study was the lack of placebo control group receiving therapist’s time. This lack of placebo, makes the results of this study even more nebulous as it cannot be determined if GET or CBT individually or in tandem were the effective treatment/s.
The study authors also concede that a further limitation of this study involved the subjective nature of the primary outcome measure- fatigue. Other studies analysed by the Cochrane Review determined outcome benefit by more objective, physiological means such as VO2 max. Self-reported measures are largely unreliable in the scientific and medical domain. It is therefore questionable as to why easily obtained objective measurements were not taken in this study.
Conclusions: I am reluctant to draw any conclusions from this study. Even the study participants- with mild to moderate Oxford defined CFS, not taking any treatments, that didn’t dropout of the study may not benefit from GET and patient education. The non-calibrated education fragment of the study and the entirely subjective primary outcome measure make this study largely unscientific. Also the type, duration and intensity of GET was not specified in this study. As the Appleby et al. (2) study resulted in less than favourable results at 70% VO2 max GET, the intensity of GET performed is highly relevant to the study outcome. To draw any conclusions from this study would be to neglect a highly relevant aspect of GET- exercise intensity.
Wallman et al. (4) used the 1994 CDC CFS criteria cohort for their exercise study. This cohort contains approximately 40% of CFS patients (When referenced to an International Classification of Disease ME/CFS cohort (7).) Consequentially, approximately 60% of the cohort studied were not true CFS patients. Patients were divided into 2 groups. Group 1 was prescribed GET which involved; walking, cycling or swimming. GET started at between 5 and 15 minutes duration and worked up to a maximum of 30 minute sessions. Intensity of exercise was determined by the mean heart rate that the patients achieved during the initial submaximal exercise test. Group 2 were prescribed relaxation and flexibility exercises such as stretching. Both groups performed their prescribed workout 3-4 times per week.
The cohort used in this study were not severely ill as they were initially able to do at least 5 minutes of exercise. Also they were able to attend weekly testing sessions at a University laboratory. Both of these tasks are generally unattainable tasks for those severely affected CFS patients. The study authors also concede that a skewed sample of “more robust and healthier subjects” may have been involved in this study. This was partly due to the maximal oxygen consumption testing that may “deter some people with CFS from participating.” The authors’ continue to question the suitability of using the maximal effort oxygen test for people with debilitating fatigue which is according to the authors, “exacerbated by physical activity.” This initial maximal effort oxygen test would not only deter those with severe CFS but a large portion of those with moderate CFS from participating in this study.
7 subjects out of the study group of 68 withdrew from the trials partway. Also 6 subjects out of the 68 were diagnosed with a major depressive order. A further 2 subjects had been diagnosed with dysthymia. The participants in this study were recruited from notices placed in medical surgeries and newspapers. A CDC 1994 CFS diagnosis was also required by each participant’s doctor in order to be included in the study. This step may have also skewed the study cohort as a large portion of doctors are unfamiliar with diagnosing CFS. A more accurate method to determine a study cohort could have been achieved through CFS specialists who are familiar with providing a CFS diagnosis. The hope of participating in the study may have also led some participants to “modify” their symptoms to fit into the CDC 1994 CFS definition. To overcome this method, the authors’ should have not alerted each individual to the required specific CFS diagnosis needed.
A further criticism of this Wallman et al. study involves the omission of the patient’s illness duration. This study was the only Cochrane Review study to not include this information. Also the number of study participants who were employed was also not supplied. This figure is useful if functional work capacity is to be determined as an outcome. The Wallman et al. study lasted for only 12 weeks hence the long term affects of GET aren’t clear from this study.
The study also fails to report or monitor many physical symptoms that are central to CFS including pain, sleep, neurological symptoms or endocrine symptoms. Wallman et al. speculate that the improvements noted in this study may be due do the abandonment of “avoidance behaviours.” It is therefore questionable if the GET itself had a beneficial effect by increasing fitness levels. The Wallman et al. study authors even concede that GET is not a cure for CFS.
Conclusion: Some patients fulfilling only the CDC 1994 CFS criteria and with mild CFS may benefit from a GET routine. This GET routine must be carefully structured and a heart rate monitor should be worn to determine exercise intensity levels. The target heart rate that should be aimed for has not been specified by Wallman et al.
Overall Conclusion Based on the Five Studies
The Cochrane Review concluded that there is “evidence that some patients may benefit from exercise therapy.” Also “Patients with CFS who are similar to those in these trials should be offered exercise therapy and their subsequent progress monitored.” Based on my analysis of the five Cochrane Review investigated studies, the patients with CFS “similar to those in the trials” that may benefit from GET are
- (My conclusion from the Fulcher et al. (3) study) Some of those patients fulfilling only the Oxford criteria that do not have an appreciable sleep disorder and may have a psychiatric disorder, comorbid disorder and/or a comorbid phobia and have only be ill for a short period.
- (My conclusion from the Moss-Morris et al. (4) study) Some CFS patients fulfilling only the CDC 1994 criteria, with mild CFS and CFS of a short duration.
- (My conclusion from the Wallman et al. (6) study) Some patients fulfilling only the CDC 1994 CFS criteria and with mild CFS.
Those patients that fulfil one of the subgroups above are the ones that may benefit from GET, although an increase in work capacity is unlikely as a result of the GET. The GET that these specific patients should have the option of performing involves low intensity exercise, 40% VO2 max and a heart rate monitor should be worn to monitor exercise intensity. Also an expert health practitioner should personally tailor this GET program for the specific individual and help monitor progressive measures to help avoid illness deterioration.
Based on the Cochrane Review’s conclusion, only these specific subgroups of CFS patients should be offered a certain type of GET. Therefore the Cochrane Review does not recommend GET for those CFS patients who don’t satisfy one of the aforementioned, atypical three CFS subgroups.
Part 1 of this meta-analysis can be found here: https://livingwithchronicfatiguesyndrome.wordpress.com/2010/09/10/a-meta-analysis-of-the-efficacy-of-graded-exercise-therapy-in-treating-cfs-part-1/
Part 3 is in the process of being written.
- Edmonds M, McGuire H, Price J. Exercise therapy for chronic fatigue syndrome (Cochrane Review). In: The Cochrane Library, Issue 3, 2004. Chichester, UK: John Wiley & Sons, Ltd.
- Appleby L. Aerobic exercise and fluoxetine in the treatment of chronic fatigue syndrome. National Research Register 1995.
- Fulcher KY, White PD. Randomised controlled trial of graded exercise in patients with chronic fatigue syndorme. BMJ 1997;314(7095):1647-52. and White PD, Fulcher KY. A randomised controlled trial of graded exercise in patients with a chronic fatigue. Royal College of Psychiatrists WinterMeeting, Cardiff. 1997.
- Moss-Morriss R, Wash C, Tobin R, Baldi JC. Mechanisms of change during a randomized controlled graded exercise trials for Chronic Fatigue Syndrome. Submitted for publication 2003.
- Powell P, Bentall RP, Nye FJ, Edwards RH. Randomised controlled trial of patient education to encourage graded exercise in chronic fatigue syndrome. BMJ 2001;322(7283):387-90.
- Wallman KE, Morton AR, Goodman C, Grove R, Guilfoyle AM. Randomised controlled trial of graded exercise in chronic fatigue syndrome. Medical Journal of Australia 2004;180(9):444-8.
- Neilson 2002, http://www.cfids-cab.org/cfs-inform/CFS.case.def/me.cfs.canadian.def03.txt
- Carruthers BM, Jain AK, De Meirleir KL, Peterson DL, Klimas NG, Lerner AM, Bested AC, Flor-Henry P, Joshi P, Powles ACP, Sherkey JA, van de Sande MI. Myalgic Encephalomyelitis/Chronic Fatigue Syndrome: Clinical Working Case Definition, Diagnostic and Treatment Protocols. J CFS 2002;11(1):7 – 116
- De Becker P, McGregor N, De Meirleir K. A definition-based analysis of symptoms in a large cohort of patients with chronic fatigue syndrome. J Intern Med 2001;250:234-240.
- Holmes GP, Kaplan JE, Gantz NM, Komaroff AL, Schonberger LB, Straus SE, et al. Chronic fatigue syndrome: a working case definition. Ann of Intern Med 1988;108:387-389.
- Fukuda K, Straus SE, Hickie I, Sharpe MC, Dobbins JG, Komaroff A, and the International Chronic Fatigue Syndrome Study Group. Chronic Fatigue Syndrome: a comprehensive approach to its definition and study. Ann Intern Med 1994;121:953-959