The Heterogeneity of ‘CFS’
The Prevalence of CFS
Gunn et al. (1) on behalf of the CDC estimated that the 1993 CFS prevalence was between 0.002% and 0.0073%. A 1991 study by Ho-Yen et al. (2) estimated that 0.13% percent of the UK population were inflicted with CFS. Reyes et al. (3) in 2003 figured that 0.24% of Wichita residents had CFS. This same study also determined that only 0.18% of the Wichita cohort had CFS for a duration of at least 1 year. A 1999 study by Jason et al. (4) determined that 0.42% of the United States population suffers from CFS. A Royal College Report (5) from 1996 estimated that between 1% and 2% of Britons had CFS. On behalf of the CDC, Reeves et al. (6) discerned that 2.54% of the US population had CFS as of 2007. In 1997, Wessely et al. (7) predicted that 2.6% of the Great Britain population was inflicted with CFS.
The CFS prevalence estimates are not increasing uniformly with time hence this alone does not provide a mechanism for the inconsistent illness rate estimates. The differing CFS criteria used and the variable operationalized nature of these criteria are sufficient aspects to explain the varied CFS incidence figures. These aspects will be discussed in detail in subsequent sections.
The CDC itself as an autonomous institute, has over a 14 year period produced a CFS prevalence estimate that has increased over 1,500 fold (1) (6). The median CDC CFS incidence estimate of 0.24% by Reyes et al. (3) in 2003 determined that only 16% of their CFS cohort had been privy to their CFS diagnosis and received treatment. Since this 2003 estimate, the CDC (Reeves et al. (6)) has revised the CFS prevalence rate by increasing it a further 10 fold to 2.54%. Jason et al. (8) have hypothesised that this most recent CDC CFS prevalence estimate outlier can be attributed to a broadening of the Fukuda et al. CFS definition. It has been suggested that this criteria broadening is due to the misconstrued theoretical definition when applied in practise.
The CDC’s discrepancy in estimating CDC criteria prevalence has eventuated from the same criteria- the Fukuda et al. criteria. Jason et al. (8) have emphasised that these criteria are polythetic in that only four criteria out of eight are necessary for a CFS diagnosis. Jason et al. continue to state that due to this polythetic criteria, different groups of patients are erroneously being compared. This is due to symptoms (that are often considered hallmark CFS symptoms) such as post-exertional malaise and cognitive impairments not being necessary for a CFS diagnosis. This can potentially skew the GET study results as the nature of post exertional malaise makes it inextricably linked to a worsening of symptoms after GET.
The Fukuda et al. criteria are also problematic when applied in an operationalized manner due to the lack of clarity and absence of a well-defined cohort. Jason et al. (9) have determined that 38% of the population with a major depressive disorder diagnosis were misdiagnosed with CFS under the Fukuda et al. CFS criteria. This is despite a concession by the CDC that “It is incorrect that…. CFS is a form of depression.” (10) In 2003, the CDC estimated that 0.24% of the population had CFS according to the 1994 CDC CFS criteria (3). In 2007, the CDC estimated that 2.54% of the population had CFS in accordance with the same 1994 CDC CFS criteria (6). This 10 fold increase in CFS prevalence as determined by the CDC highlights how studies using the same criteria are not necessarily using the same patients. Let us now hypothetically assume that the 2003 CDC CFS estimate can be considered as an accurate representation of the CFS cohort. This would invalidate the 2007 CDC CFS estimate. If a GET study was performed on this 2003 cohort, this group would only be representative of 10% of the 2007 CDC CFS estimate cohort. Let us now hypothetically assume that the 2007 CDC CFS estimate is representative of a true CFS cohort. Consequentially, the 2003 CDC CFS estimate would be erroneous. If GET was performed in this 2007 CDC cohort, then the results could not be extrapolated onto the 2003 CDC cohort due to it comprising a small subsection of the 2007 CDC cohort. Due to the aforementioned reasons in tandem with the broadness of the Fukuda et al. criteria, GET studies that are found to be effective for a non-ubiquitous percentage of a Fukuda cohort cannot be superimposed onto another Fukuda cohort. This is due to the absence of well-characterised CFS patients, nebulous criteria and the polythetic nature of the Fukuda criteria. In 1975, Spitzer et al. wrote that the usefulness of a diagnostic category is limited by its level of reliability. The extent that a diagnostic category is unreliable is consequentially limited by its validity for clinical research. These words by Spitzer et al. support the notion that the definitive flaws of the Fukuda criteria (and indeed wide discrepancy among CFS prevalence estimates) invalidate the bulk of GET related literature. The only conclusion that can be drawn from this GET literature using nebulous criteria is that GET is useful for those that it helped in the study. No further conclusions can be extrapolated, even for those CFS patients in an identical cohort.
Physiological Abnormalities distinguishing CFS cohorts
Jason et al. (11) have presented research suggesting that key differences exist between CFS patients with high levels of infectious and inflammatory processes in comparison to those CFS patients with low levels of infectious and inflammatory processes. The authors also highlight a possible CFS subgroup that has cognitive impairments as a primary symptom and other symptoms of a lower intensity. Jason et al. conclude their paper by strongly advocating the use of subgrouping CFS cases in order to aid researchers in determining a higher level of understanding of this illness including possible preventions, treatments and cures.
The physiological effects of exercise on CFS patients will be discussed in a subsequent section within this meta-analysis. These physiological exercise effects do not present consistently across the Fukuda et al. CFS cohort. De Becker et al. (12) have written that; histological, metabolic, muscle oxidative capacity, atrophy of fast twitch fibres, lactic acid produced during exercise and myopathic features on electromyograms are all exercise related physiological effects that have found to be abnormal among CFS patients. De Becker et al. emphasise that these abnormalities are not consistent among CFS patients and the authors conclude that CFS is a heterogeneous illness. De Becker et al. also state that GET studies may be finding varied results among their CFS cohorts due to the variability in patient populations and subsequent muscle abilities. These inconsistent exercise related physiological abnormalities among CFS patients support the theory that even among a specific and autonomous CFS definition, CFS patients have varying abilities and will respond differently to exercise. Any GET conclusions from one CFS criteria subgroup studied will not be relevant to others fulfilling the same CFS criteria. This is due to the aforementioned exercise related physiological effects differing across CFS criteria. Rather than GET studies attempting to define CFS patients through criteria such as Fukuda et al. more relevant subgrouping should occur through patient fragmentation according to exercise related physiological abilities such as the previously mentioned De Becker et al. list of exercise effects on CFS patients.
Lane et al. (13) performed a study to detect enterovirus sequences within the quadriceps of CFS patients. 20.8% of the CFS patients were positive for this test. Lane et al. also tested the entire CFS cohort in this study for abnormal lactate response to exercise. The authors stated that an abnormal lactate response to exercise indicates an impaired muscle energy metabolism. Lane et al. found an association between an abnormal lactate response to exercise and the presence of enterovirus sequences in the quadriceps. This studies correlation result provides evidence for the heterogeneity of CFS, especially in an exercise related domain. Lane et al. have demonstrated that regardless of CFS criteria, subgroups of CFS patients will respond differently to exercise.
DeLuca et al. (14) performed a cognitive based study on CFS patients fulfilling the 1994 CDC CFS criteria. Patients were excluded if; they had an illness onset exceeding four years, had symptoms of less than a moderate severity (at the time of testing) or had a psychiatric disorder up to five years prior to their CFS onset. In this study, patients with moderate to severe CDC 1994 defined CFS were divided into two groups. The first group of CFS patients had a current or previous psychiatric disorder (with an onset during the patients CFS) while the second group of CFS patients did not have a current or previous psychiatric disorder. These two groups were compared to a healthy although relatively sedentary control group. The CFS group without psychiatric disorders performed worse at the cognitive tests than the CFS patient cohort with a psychiatric disorder. This study result emphasises that the 1994 CDC CFS definition encompasses distinct groups of patients that either a) have separate illnesses under the CFS umbrella or b) have the same illness that manifests in different symptoms and consequential differing abilities. If a) is correct then a blanket GET treatment should not be recommended for completely distinct illnesses without studies into GET for these distinct illnesses that fall within the all-encompassing CDC’s CFS definition. If b) is correct then the heterogeneous nature of CFS including different patient’s abilities and symptoms requires CFS subgroups to be studied in regards to GET as a treatment as opposed to the current paradigms over simplistic panacea.
The DeLuca et al. cognitive impairment study result also supports the notion that those CFS patients with a psychiatric disorder component have a distinctly different illness than those CFS patients void of a psychiatric disorder. This is due to the neat psychiatric disorder dichotomy that correlates with the cognitive impairment symptom (or lack thereof.) The study result also highlights how different subgroups of CFS patients respond differently to stimuli. DeLuca et al. have speculated that the lack of consistent medical findings in the general CFS domain is a direct result of the heterogeneity of the CDC’s 1994 CFS criteria. The authors continue on to state that finding diagnostic variables may lead to identification of more homogeneous CFS subgroups which may consequentially aid in identification of causative agents.
Lerner et al. (15) performed a study on a specific subgroup of CFS patients who had severe CFS and collectively unique electrocardiogram abnormalities (when compared to control electrocardiogram abnormalities.) These CFS patients also had high antibodies to the human cytomegalovirus. These patients were treated with intravenous ganciclovir. 24 weeks after the IV treatment, 72% of the CFS patients had reached remission. The patients who didn’t respond to the ganciclovir intravenous treatment had lower human cytomegalovirus antibodies originally in comparison to the group that did respond to treatment.
See et al. (16) also performed a study that supports the CFS comprising distinct illness state hypothesis. This study provided CFS patients with alpha interferon in a placebo cross-over designed double blinded study. The only CFS patients who benefited from taking alpha interferon were the CFS patients who had a reduced Natural Killer cell function. These study result suggests that CFS patients don’t ubiquitously respond to treatments but rather a specific and well-defined subgroup of CFS patients may respond to a recommended CFS treatment. These studies therefore support the hypothesis that the term ‘CFS’ can currently be defined as an accumulation of multiple distinct illnesses that have unique etiologies and require individualised treatments. The blanket recommendation of any treatment to an apparent heterogeneous group of illnesses that are actually distinct illnesses is erroneous. GET should therefore not be recommended ubiquitously to CFS patients.
A 1993 CDC study by Mawle et al. (17) compared CFS patients fulfilling the 1988 CDC CFS criteria against demographically similar controls. The immune testing only revealed minor differences between the CFS group and the controls. The CFS patients were then grouped into how they were feeling on the testing day and the nature of their CFS onset. After this grouping, significant differences between the CFS groups and the controls became evident. The sudden onset CFS group had much higher levels of CD8 cells that were expressing CD11b. This group also had significantly higher levels of Il-2 than the corresponding controls. The gradual onset CFS group had lower levels of IL-1a and IL-1b than the controls. The gradual onset group also had a lower percent of CD56 cells. The more severely ill patients had higher IL-2 levels and lower CD8 cells that were expressing CD25. They also had lower levels of IL-1a when compared to the controls. This study result highlights that a holistic approach to “CFS” may reveal very few abnormalities when compared to controls however significant abnormalities are present in the patients when a reductionist approach is adopted. This relates to GET in that CFS patients need to be subgrouped if meaningful data is to be collected. Also this study result emphasises that different types of CFS, based on onset, immunological markers and severity, seem to indicate distinct illness groups. Rather than CFS data being collected for one group and extrapolated onto another CFS group, specific studies are required to determine if treatments such as GET may be beneficial for CFS subgroups. Until this data is obtained, a recommendation of GET for CFS patients is a fallaciously reasoned approach. This is due to the heterogeneous nature of CFS that consists of quite distinct subgroups with differing features that may indeed be distinct illnesses.
Jason et al. (18) divided CFS patients into two groups- those with normal cortisol levels and those with abnormal cortisol levels (based on 5 testing occasions.) The CFS patients deemed to have abnormal cortisol levels did not improve over the study period in immunological or self reported areas. The CFS patients with normal cortisol levels experienced improvements in both immunological and self reported areas. This result indicates that a subset of CFS patients (those with normal cortisol levels based on a graph of 5 testings) may skew CFS study results due to spontaneous improvements. GET studies on CFS patients should therefore take this factor into account and determine cortisol levels over a period of time for the patients included in their study. It is conceivable that the CFS patients in GET studies are improving due to the nature of their CFS subgroup as opposed to their GET treatment. Having said this, some GET studies also involve a CFS control group that partakes in such activities as stretching as opposed to aerobic exercise. This Jason et al. result also supports the notion that determining accurate CFS subgroups is a prerequisite to determining relevant and applicable CFS treatments. This study supports the scientific literature previously mentioned that CFS is heterogeneous in nature and extrapolating GET study results onto other subgroups, different CFS criteria or indeed the same CFS criteria is erroneous due to these false, non-homogenous and poorly characterised divisions of CFS patients.
Illnesses Beneath the CFS Umbrella
Chronic Fatigue Syndrome is in theory a distinct disorder that is well-characterised and homogeneous. In practise, many other conditions fulfil the CFS criteria. CFS criteria generally stipulate that an alternative diagnosis excludes a CFS diagnosis. This is problematic in practise due to the large number of CFS-like conditions that produce symptoms that satisfy the CFS criteria. These illnesses/conditions include; autoimmune disorders, cardiac and respiratory disease, connective tissue disorders, side effects of medications, thyroid disease, Addison’s Disease, Lyme disease, celiac disease, irritable bowel syndrome, immunodeficiency, multiple sclerosis, sleep disorders, rheumatic diseases, psychiatric disorders, psychological disorders, anxiety, CMV, EBV, HHV-6, parvovirus, fibromyalgia, malnutrition, mitochondrial dysfunction, neuropathy, Cushing syndrome, B12 deficiency, depression, enteroviruses, adrenal insufficiency, obesity, parasites, food allergies, hormonal abnormalities, coxsackie viruses, post viral fatigue, post traumatic stress disorder, seasonal affective disorder, multiple chemical sensitivity and systemic lupus.(19) (20) It is true that a number of these illnesses/conditions may not be included in GET studies on CFS due to exclusionary criteria however a sufficiently large portion of these aforementioned illnesses/conditions aren’t considered as exclusionary criteria in GET studies. GET studies do not require patients to have excluded the above list of conditions through testing in order to be included in a GET study. The GET studies do not perform these extensively listed tests on the CFS patients included. A large portion of patients included in GET studies will fulfil at least one of the above illnesses/conditions. Consequentially, the GET studies are including a scatter pot of illnesses/conditions and the results of these studies are therefore as equally as non-definitive. For GET studies on CFS patients to produce meaningful results, the study authors would be required to distinguish the above list of illnesses from one another in CFS patients. It is only then that GET studies would be comparing like with like. Once this fragmentation is made, results from the studies can be translated to the corresponding “CFS” subgroups.
The CFS cohort is extremely nebulous and more difficult to subgroup due to the high number of CFS patients that have one or multiple illnesses/conditions from the above list. As mentioned previously, Lane et al. (13) demonstrated that the CFS subgroup with an enterovirus infection had an abnormal lactate response to exercise and an impaired muscle energy metabolism. This was in contrast to an enterovirus-negative CFS patient cohort. The above list of CFS-like conditions with symptoms that can potentially mimic CFS provides evidence that the over simplistic viewpoint that GET studies are comparing similar patients is flawed. A haphazard array of illnesses is currently being studied in unison which is consequentially producing meaningless data.
Despite the similar illness exclusion stipulation that is part of the CDC’s 1994 CFS criteria, the CDC website (21) alludes to a study result which underlies the theory that CFS is heterogeneous. This study result dictates that EBV, Ross River virus and Coxiella burnetti virus will lead to a post-infective state that fulfils the CDC’s 1994 CFS criteria in 12% of instances. This consists of a 12% post virus state cohort that can be labelled and diagnosed as a post-viral fatigue cohort. The CDC criteria include this post-viral fatigue cohort amongst their broad CFS criteria. Consequentially, this cohort is studied as a CFS cohort however the uniqueness of this group’s onset requires a fragmentation from the CFS cohort to produce a subgroup. The inclusion of this 12% post-viral fatigue cohort into a CFS cohort is sufficient to skew study results. Rather than including the post-viral fatigue cohort into the omni-encompassing CFS cohort, an effort should be made to subgroup these symptomatically similar albeit clinically distinct illnesses.
The post viral fatigue study that is referred to on the CDC website was performed by Hickie et al. (22) on behalf of the CDC. This study was one part of a subset study comprising the ‘Dubbo Studies.’ Hickie et al determined that following EBV, RRV and Coxiella burnetti onset, different patients took varying lengths of time to recover. The data indicates that 35% of this viral cohort had post viral fatigue at 6 weeks from baseline. 27% had post viral fatigue at three months, 12% at six months and 9% at 12 months. A parallel Dubbo study (23) determined that after 24 months, only 3% of the cohort still had post viral fatigue. The Fukuda et al. CFS criteria requires an illness duration of at least six months as a necessary prerequisite before the diagnosis of CFS can be given. The above data from Hickie et al. (22) shows a continuum that is atypical of a general CFS cohort, as opposed to this subgroup.
Professor Andrew Lloyd held a hegemonic role in the Dubbo studies. A graph titled ‘Natural History of Symptoms’ (24) presented by Professor Lloyd containing data from the Dubbo studies illustrates the post viral fatigue patients’ symptomatic percentage versus time. This graph illustrates that time, not treatment was the instigating factor leading to this cohort’s remission. After 24 months, virtually all of the former post viral fatigue (and interchangeable) ‘CFS’ patients had recovered. The mean illness duration of the CFS patients included in GET studies is typically 24 months. It is therefore conceivable that the GET patients studied have naturally reached remission based on time alone, independent of the GET treatment. The Hickie et al. data consisting of the high remission rates in the post viral fatigue cohort contrasts a Cairns et al. (25) review of CFS remission rates. Cairns et al. reviewed 14 studies that were primarily focused on determining CFS illness duration. The review concluded that the median recovery rate from CFS without systematic intervention was 5%. This result is the antithesis of the Hickie et al. result which saw the post viral fatigue cohort almost ubiquitously reach a remissions status within 24 months, with remission status being non-dependant on treatment. The Fukuda et al. criteria includes this post viral fatigue cohort within the heterogeneous CFS label. It is therefore evident that the CFS remission rates are heterogeneous depending on CFS subgroup. It is also apparent that post viral fatigue diagnosis should be an exclusionary criteria among GET CFS studies due to the almost ubiquitous remissions reached by this cohort within a short temporal frame.
Vollmer-Conner et al. (26) performed a study titled ‘An empirical delineation of the heterogeneity of chronic unexplained fatigue in women.’ The authors concluded that chronic unexplained fatigue can be differentiated by; obesity, sleep hypnoea, depression, a physiological stress response, sleep disturbance, interoception and menopausal status. This list of conditions, previously in this meta-analysis, was deemed to be in some instances misclassified as CFS. The current Fukuda et al. CFS criteria fails to adequately distinguish CFS from this list of heterogeneous CFS-like conditions. A portion of GET studies attempt to exclude some of these conditions however no GET study has adequately excluded all of these conditions. This is partly due to the overlapping nature of fatigue as such symptoms as depression, obesity and sleep disturbance can be the primary cause of fatigue or indeed a secondary effect of the primary fatigue source. Determining this fatigue hierarchy is often difficult and problematic. This Vollmer-Conner et al. result is further evidence that the patients included in GET studies are not homogenous and the cohort comprises patients of different illnesses rendering the conclusions inadequate and nebulous.
A study by Jason (27) provides further evidence that the overlapping and secondary symptoms in CFS are the primary symptoms of unrelated disorders. Jason explains that major depressive disorder is a prime example of a psychiatric disorder that shares symptoms such as fatigue, sleep disturbance, joint pain, muscle pain and impaired concentration with CFS. These symptoms along with the fatigue accompanying major depressive disorder, provides a fulfilment of the Fukuda et al. CFS criteria. Jason writes that these patients could be misclassified as CFS patients in studies. Friedberg et al. (28) and Komaroff et al. (29) have shown that the fatigue accompanying CFS does not share equal prominence as the fatigue that often tandems depression. This provides further evidence that it is erroneous to include these depression causing fatigue patients in GET studies. Depressed patients may fulfil a Fukuda et al. diagnosis based on symptomatology alone however the nature, prevalence and severity of these symptoms clearly distinguishes depressed patients from CFS patients.
Comparing CFS Criteria
Institutes are recommending GET ubiquitously to all CFS patients regardless of subgroup or CFS criteria filled. The CDC website (30) even recommends GET to “Very ill people who are housebound or bedbound.” As discussed in Part 2 of this meta-analysis, GET studies have almost exclusively been performed on patients fulfilling either the Oxford CFS criteria or the Fukuda et al. CFS criteria. The extrapolation of this data onto other CFS criteria cohorts is erroneous.
Jason et al. (31) performed a study titled ‘A Comparison of the 1988 and 1994 Diagnostic Criteria for Chronic Fatigue.’ The Jason et al. study revealed that the patients fulfilling the 1988 criteria were a more severely functionally impaired cohort that were more symptomatic compared to those fulfilling only the 1994 CDC CFS criteria. Jason et al. concluded that those fulfilling the 1988 CDC criteria may be a clinically separate group of patients. The authors also wrote that the 1994 criteria may encompass a heterogeneous cohort of patients with a range of illness severities and disabilities. This study by Jason et al. highlights how patients fulfilling different criteria not only have different symptoms but also varying degrees of disability. It is therefore fallacious to impose the result of a study focused on one criteria onto a distinctly different criteria. If one disregards the heterogeneity of CFS in terms of distinct illness subsets, the scientific method requires specific studies targeted at the varying levels of CFS patients severity continuum. Studies performed on a mildly affected cohort such as the Fukuda et al. cohort should not be extrapolated onto severely affected patients. This is especially true of GET studies in which the very aspect being measured- functionality is what distinguishes these cohorts.
Sampson (32) performed an extensive analysis of cohorts containing CFS, fatigued and mood disordered patients. Sampson concluded that broad CFS definitions such as the Oxford and Fukuda et al. definitions include patients with a psychological/psychiatric fatigue etiology. By including these primary psychological/psychiatric patients, Sampson suggests that the study results are clearly going to show that psychosocial factors are prevalent in CFS. It is a tautology and these studies such as GET studies provide us with no more information than we started with. These studies should focus their attention onto the underlying fatigue that is the product of CFS itself. The process of including fatigue caused by a psychological/psychiatric etiology results in two quite distinct cohorts being falsely linked. The consequential data and results are therefore not CFS typical but rather just fatigue typical. Due to fatigue’s heterogeneous nature, these studies are haphazard and reveal nothing about the differing components of various fatigue etiologies.
Jason et al. (33) performed a study comparing patients who fulfilled the Fukuda et al. CFS criteria and patients who filled a specific ME criteria. The authors also compared these two groups to a group that had chronic fatigue due to psychiatric causes. This Jason et al. study concluded that a significant difference existed between these three groups. These differences were mainly consistent of neurological, fatigue, cardiopulmonary abnormalities neuropsychiatric and rheumatological symtoms. Jason et al. determined that those meeting the ME criteria were significantly different from those patients in the psychiatric fatigue group. Also the Fukuda et al. group resembled the psychiatric group much more than the ME group resembled the psychiatric cohort. Jason et al. speculated that it may be erroneous to synthesize CFS study results that use different definitions. This conclusion by Jason et al. is diametrically opposed to the actions of many GET studies and institutes which extrapolate specific CFS criteria GET results onto other CFS criteria groups.
The Canadian Consensus Criteria for CFS
Carruthers et al. (34) compiled the actual Canadian Consensus Criteria (CCC) for CFS in a paper titled ‘Myalgic Encephalomyelitis/Chronic Fatigue Syndrome: Clinical Working Case Definition, Diagnostic and Treatment Protocols. These CCC attempt to distinguish a more homogonous type of patient- an area in which the previous criteria such as the Oxford criteria and Fukuda et al. criteria have failed. The CCC specifies that the fatigue resulting from post-exertional malaise usually lasts for a 24 hour plus duration. This fatigue is also associated with an impairment of cognitive abilities. The CCC also states that this reactivity following exertion is often delayed to the point of occurring the next day or in some instances later. The CCC continues on to emphasise that exercise will generally improve patients with fatigue caused by depression, in contrast, exercise will cause adverse effects on CCC patients. This CCC composed by Carruthers et al. distinguishes these CCC CFS patients from those heterogeneous CFS patients that fulfil the Oxford and/or Fukuda et al. criteria. In order for a patient to fulfil the CCC, exercise must be contraindicated in that patient. Therefore by definition, GET is not suitable for CCC patients. If a patient improves as a consequence of exercise, they are automatically excluded from the CCC.
The CCC also questions a key aspect of the Fukuda et al. CFS criteria. This aspect involves the Fukuda et al. focus on one arbitrary symptom- long-lasting fatigue. The CCC emphasise that this is just one symptom across the spectrum of CFS symptoms that are of great importance such as post-exertional malaise, pain, sleep disturbances and cognitive impairment. The CCC concludes that the Fukuda et al. criteria are flawed in that they make it very difficult for physicians to separate the fatigue consequential of CFS from the fatigue caused by other illnesses.
The criteria aspect of the CCC stipulates that a patient fulfilling the CCC must meet criteria for fatigue, post-exertional malaise and/or fatigue, sleep dysfunction and pain. Combined with these, the patient must also have at least two neurological or cognitive manifestations and at least a symptom from two of the following categories; autonomic, neuroendocrine or immune manifestations. There are further criteria that need fulfilling for a patient to be diagnosed with CCC CFS however the above is an outline of some of the necessary conditions. It is evident that these CCC exclude the majority of patients fulfilling the Fukuda et al. criteria. Also these CCC would exclude most of the patients involved in GET studies.
Jason et al. (35) have studied the CCC cohort and concluded that the CCC CFS patients share many key features. These include; similar symptoms, severe impairment, comparable levels of fatigue and weakness and similar neurological symptoms. The symptoms that are common to CCC patients can interfere with everyday living ability. Jason et al. concludes that CCC patients are a distinct group from chronic fatigue and psychiatrically fatigued patients. This study supports the notion that CCC patients comprise a distinct and homogonous subgroup. Due to the different symptoms and impairment degree in this CCC cohort, compared to a Fukuda et al. cohort, GET results from a Fukuda et al. cohort should not be transposed onto this CCC cohort.
In contrast to the Fukuda et al. and Oxford CFS criteria, the CCC isn’t polythetic. By stipulating that specific symptoms are necessary for a CCC CFS diagnosis, the CCC is an improvement on the Fukuda et al. criteria due to the lack of broadness and unambiguousness in interpretation by practising clinicians. The patient that fulfils the CCC will be approximately homogonous and study results on this cohort can be extrapolated ubiquitously to CCC patients worldwide.
Further Evidence that CFS is Heterogeneous
The CDC has recently stated (36) that “Because CFS is clinically heterogeneous and comprised of various subtypes, it is essential that treatable causes of the illness are identified to improve clinical management.” This indicates a definitive admission from the CDC that “CFS is clinically heterogeneous.” Despite this concession, the CDC website does not share this definitive viewpoint and isn’t committing to ruling out a common CFS etiology that explains all CFS cases. The CDC website (37) states that “While a single cause for CFS may yet be identified, another possibility is that CFS represents a common endpoint of disease resulting from multiple causes.” The varying nature of the diseases causing CFS already provide many specific and distinct mechanisms to explain the end disease state that fulfils the CDC’s CFS criteria. The range of responses to treatments also underlie the conclusion that CFS doesn’t represent a “common endpoint of disease” but rather a cluster of distinct diseases which have distinct endpoints however these diseases also incidentally fulfil the broad and omni-encompassing CDC CFS criteria.
The recently conducted UK PACE trial (38) that attempts to treat CFS patients with GET alongside CBT produced a study protocol that was published in 2007. This study protocol explained in detail the precise methods and cohorts that would be studied amongst this trial. I will not provide a thorough critique of the PACE trial in this section however it is worth noting that the broad Oxford CFS criteria were the diagnostic criteria used in this study. When referenced to an International Classification of Disease ME/CFS cohort (39), only 10% of Oxford criteria patients are true CFS patients. This PACE trial study protocol states that “The trial will recruit new patients from secondary care clinics run by three different disciplines (immunology, infectious disease and psychiatry) in six different centres in both England and Scotland. This recruitment plan will ensure sufficient heterogeneity to allow generalisation of the findings.” As discussed in detail previously, the findings pertaining to one CFS broad criteria cohort such as Oxford of Fukuda et al. cannot be extrapolated or “generalised” onto other cohorts. This PACE trial also aimed to recruit a heterogeneous cohort rather than a homogonous cohort. This is problematic in that the immunological, infectious disease and psychiatric patients recruited have distinct illnesses with a common arbitrary symptom in fatigue. By blurring the boundary between these already subgrouped and distinct illness cohorts and by creating a less well defined CFS patient, the findings of this PACE trial cannot even be applied to any cohort, heterogeneous or homogeneous.
A study by Pierce et al. provides further evidence against the generalisation of GET findings onto CFS patients. This study states that in the most generous scenario, GET has only relived and not cured a specific and small minority of CFS patients. Also that the empirical evidence supports the notion that exercise is the primary cause of both the negative effects concerning physiology and quality of life in a significant portion of CFS patients. Pierce et al. continue on to state that due to the heterogeneity of CFS, a universal therapy such as GET is unsuitable unless the prescriber has a high degree of confidence that the patient’s condition will not deteriorate. The authors state that they cannot think of a more inappropriate therapy for CFS patients. Pierce et al. provide the further argument that blanket recommendations of GET are not appropriate for heterogeneous groups of patients such as CFS patients. A large portion of this cohort already respond in a negative manner to exercise hence an erroneous oversimplification of the CFS cohort as homogenous may be the key instigator of detrimental CFS treatments such as GET. Pierce et al. expand on their heterogeneous CFS argument by stating that no studies to date have managed to investigate all of the nuanced aspects relating to CFS. The large body of empirical evidence suggests that CFS is caused by a complex series of variable events that can potentially affect multiple bodily systems. The authors also mention that one of the defining factors of ‘CFS’ involves the large range of varying symptoms between patients which suggest that different bodily functions are affected when comparing different patients, all to varying degrees. This suggests that the bodily systems regulating the exercise response may not be affected in some CFS patients although in other CFS patients, these systems are likely to be severely affected. Consequentially, this differentiation between ‘CFS’ patients, requires therapies to be individually tailored to patients rather than broad and blanket recommendations such as GET imposed onto the entirety of CFS patients.
The very nature of recruiting CFS patients to complete GET studies is fraught with difficulties. Jason (27) mentions that the current recruitment methods are inadequate due to the type of questions that the study authors ask the potential trial participants. He cites an example of the patient indicating that they have post-exertional malaise. Rather than asking subsequent questions pertaining to this symptom, GET study authors fail to gather details on this symptom. Jason suggests standardized questions that include the duration, frequency and severity of individual symptoms as well as onset, pattern, intensity and associated factor details. This detailed questioning of CFS patients is also endorsed in a study performed by Hawk et al. (41) Jason also uses the analogy of a lupus diagnosis which involves a combination of factors including clinical judgement, patient reports and objective measures to determine a diagnosis. It is impossible to complete this checklist when classifying CFS patients according to the Fukuda et al. criteria, due to the very nature of these criteria. Consequentially, GET studies using the Fukuda et al. criteria are limited by this criteria’s inadequacy in diagnosing a homogenous group of patients or indeed a true CFS group of patients.
The application aspect of the Fukuda et al. criteria is also flawed due to the non-standardised nature of the criteria. Jason et al. (33) have discerned that the determining factor concerning whether a patient fulfils CFS criteria involves the very way the questions are phrased to the patients. This study discovered that the number of patients indicating post-exertional malaise or fatigue varied from between 40.6% to 93.8% depending on the manner in which the question was asked. This significant result requires CFS criteria such as the Fukuda et al. criteria to standardize their questions in order to fulfil the fundamental study reliability component. The very diagnosis of CFS is often contingent upon fulfilling a quotient of symptomatic criteria. This Jason et al. result indicates that a large portion of patients studied may be falsely diagnosed due to something as variable as the phrasing of questions.
The very nature of many CFS criteria involves patients being asked if they have certain symptoms. This method of diagnosis is doubly problematic due to the potential for suggestive bias. If an individual presents undiagnosed to a clinician, it is imperative for the patient to expound upon their symptoms to avoid this suggestive bias. If the clinician or researcher asks the patient if they have certain symptoms (as is the case in all GET studies), the possibility of suggestive bias arises. For instance, if the patient is asked if they are exhausted, a large majority of patients and indeed a large portion of the population will indicate “yes.” When a healthy cohort (42) were questioned as to if they were exhausted, 100% answered “yes.” When this same healthy cohort was asked if they were chronically exhausted, 30% still gave the self-evaluated answer- “yes.” If this exhaustion is a significant symptom for the patient then they will likely mention it to the clinician without the need for a symptomatic question response. By planting the symptom suggestion in the patients mind, a larger portion of subjects will state that they have the symptom as opposed to the patients presenting their symptoms independently of a clinical suggestion bias. This factor has the potential to produce patient cohorts that are more prone to suggestion bias than patient cohorts that legitimately have CFS that is affecting their functionality.
The cohorts used in GET studies such as the Fukuda et al. criteria cohorts are also nebulous due to the lack of diagnostic tests used by the diagnosing clinicians. The symptomatic questionaries that require fulfilling are vague and lack the definitiveness of objective tests. The Barnum affect is defined as “the tendency to accept certain vague or worthless information (43).” The vague and descriptive writings that amount to CFS questionnaires can potentially influence GET studied cohorts in this manner. This cognitive bias parallels the suggestive bias discussed in the previous paragraph. There are several tests that should be used by CFS study authors to support a CFS criteria diagnosis. One such test is the Natural Killer cell function test. Fletcher et al. (44) found that “NKCC (natural killer cell cytotoxicity) was a good predictor of CFS status.” GET study authors have failed to make use of the range of objective tests in order to comprise their CFS study cohorts. This further compromises the accurate CFS diagnosis and homogenous nature of CFS patients involved in GET studies.
A study by Joyce et al. (45) on CFS criteria and remission or improvement status concluded that the less restrictive the CFS criteria used, the better the prognosis. This study result highlights how the broad CFS criteria such as Oxford and Fukuda et al. contain a higher portion of patients that will improve regardless of treatment. The GET studies that have found an improvement in CFS patient’s symptoms have all used a broad criteria such as Oxford of Fukuda et al. Consequentially, these cohorts have the best prognosis, indicating that either these patients have a distinctly separate illness to more restrictive CFS criteria cohorts, these patients aren’t as severely ill as other CFS criteria cohort patients or possibly these patients had symptoms that would have spontaneously improved, regardless of the GET treatment. This Joyce et al. study provides further evidence that broad CFS criteria studies are including very different patients to more restrictive CFS criteria cohorts.
It has been determined that the studies finding that GET is a beneficial CFS treatment have used an Oxford or Fukuda et al. criteria cohort. These criteria have been demonstrated to be; heterogeneous, containing distinct and incomparable illness entities, non-operational, non-standardised, too broad, polythetic, containing non-CFS fatigue causing conditions and lacking fundamental practical requirements. Consequentially, these aforementioned GET studies have produced non-applicable conclusions that have been fallaciously applied to all CFS patients. It is erroneous for these GET study results to be imposed on different CFS cohorts and it has been demonstrated that the flawed nature of the broad CFS criteria used by these GET studies makes superimposing the study results onto an identical CFS criteria cohort fallacious.
1. Gunn, W.J., Connell, D.B., & Randall, B. (1993). Epidemiology of chronic fatigue syndrome: The Centers-for-Disease-Control study. In B.R. Bock & J. Whelan (Eds.), Chronic Fatigue Syndrome. (pp. 83-101). New York: John Wiley & Sons
2. Ho-Yen D O, McNamara I. 1991. General practitioners’ experience of the chronic fatigue syndrome. British Journal of General Practice. 41(349):324-326.
3. Reyes M, Nisenbaum R, Hoaglin DC, Unger ER, Emmons C, Randall B, Stewart JA, Abbey S, Jones JF, Gantz N, Minden S, Reeves WC. Prevalence and incidence of chronic fatigue syndrome in Wichita, Kansas. Arch Intern Med 2003;163:1530-6
4. Jason LA, Richman JA, Rademaker AW, et al. A community-based study of chronic fatigue syndrome. Arch Intern Med1999; 159: 2129-37.
5. Royal Colleges of Physicians, Psychiatrists and General Practitioners (1996). Chronic fatigue syndrome; Report of a joint working group of the Royal Colleges of Physicians, Psychiatrists and General Practitioners. London, UK: Royal College of Physicians of London.
6. Reeves, W.C., Jones, J.J., Maloney, E., Heim, C., Hoaglin, D.C., Boneva, R., Morrissey, M., & Devlin, R. (2007). New study on the prevalence of CFS in metro, urban and rural Georgia populations. Population Health Metrics 2007, 5:5 doi:10.1186/1478-7954-5-5
7. Wessely, S., Chalder, T., Hirsch, S., Wallace, P., & Wright, D. (1997). The prevalence and morbidity of chronic fatigue and chronic fatigue syndrome: a prospective primary care study. American Journal of Public Health, 87, 1449-1455.
8. Jason LA et al. (2010) The Development of a Revised Canadian Myalgic Encephalomyelitis Chronic Fatigue Syndrome Case Definition, American Journal of Biochemistry and Biotechnology 6 (2): 120-135, 2010
9. Jason, L.A., N. Najar, N. Porter and C. Reh, 2009a. Evaluating the centers for disease control’s empirical chronic fatigue syndrome case definition. J. Disabil. Policy Stud., 20: 93-100. DOI: 10.1177/1044207308325995American Journal of Biochemistry and Biotechnology 6 (2): 120-135, 2010
11. Jason, L. A., Corradi, K., & Torres-Harding, S. (2007). Toward an Empirical Case Definition of CFS. Journal of Social Service Research, 34(2), 43-54.
12. De Becker P, Roeykens J, Reynders M, McGregor N, De Meirleir K. (2000), Exercise capacity in chronic fatigue syndrome. Arch Intern Med. 2000 Nov 27;160(21):3270-7.
13. Lane RJ, Soteriou BA, Zhang H, Archard LC (2003). Enterovirus related metabolic myopathy: a postviral fatigue syndrome. J Neurol Neurosurg Psychiatry. 74 (10): 1382–1386.
14. DeLuca, J., Johnson, SK., Ellis, SP and Natelson, BH. Cognitive functioning in patients with chronic fatigue syndrome devoid of psychiatric disease. Journal of Neurology, Neurosurgery, and Psychiatry, 1997, 62, 151-155.
15. Lerner AM, Zervos M, Dworkin HJ, Chang, CH and O’Neill W. A unified theory of the cause of chronic fatigue syndrome. Infectious Diseases in Clinical Practice 1997;6:230-243.
16. See DM, Tilles JG. Alpha interferon treatment of patients with chronic fatigue syndrome. Immunol Invest 1996; 25: 153-164.
17. Mawle AC, Reyes M, Schmid DS. Is chronic fatigue syndrome an infectious disease? Infect Agents Dis. 1993 Oct;2(5):333-41.
18. Jason, L.A., Torres-Harding, S., Maher, K., Reynolds, N., Brown, M., Sorenson, M., Donalek, J., Corradi, K., Fletcher, M.A., & Lu, T. Baseline cortisol levels predict treatment outcomes in chronic fatigue syndrome non-pharmacologic clinical trial. Journal of Chronic Fatigue Syndrome, 14, 39-59 (2007).
22. Hickie et al. Post-infective and chronic fatigue syndromes precipitated by viral and non-viral pathogens: prospective cohort study, BMJ. 2006 September 16; 333(7568): 575.
25. R. Cairns and M. Hotopf, A systematic review describing the prognosis of chronic fatigue syndrome, Occupational Medicine 2005;55:20–31
26. Vollmer-Conna U, Aslakson E, White PD. An empirical delineation of the heterogeneity of chronic unexplained fatigue in women.
27. Jason. L, Problems with the New CDC CFS Prevalence Estimates, http://www.iacfsme.org/IssueswithCDCEmpiricalCaseDefinitionandPrev/tabid/105/Default.aspx
28. Friedberg, J.R., & Jason, L.A. (1998). Assessment and treatment of Chronic Fatigue Syndrome. Washington, D.C.: American Psychological Association.
29. Komaroff, A.L., Fagioli, L.R., Geiger, A.M., Doolittle, T.H., Lee, J., Kornish, R.J., Gleit, M.A., Guerriero, R.T. (1996). An examination of the working case definition of Chronic Fatigue Syndrome. The American Journal of Medicine, 100, 56-64.
31. Jason et al. A Comparison of the 1988 and 1994 Diagnostic Criteria for Chronic Fatigue Syndrome,
32. Sampson, CLOSE ANALYSIS OF A LARGE PUBLISHED COHORT TRIAL INTO FATIGUE SYNDROMES AND MOOD DISORDERS THAT OCCUR AFTER DOCUMENTED VIRAL INFECTION, http://www.iacfsme.org/BULLETINSUMMER2010/Summer2010SampsonAnalysisCohortTrial4481/tabid/435/Default.aspx
33. Jason LA, Helgerson J, Torres-Harding SR, Carrico AW, Taylor RR. Variability in diagnostic criteria for chronic fatigue syndrome may result in substantial differences in patterns of symptoms and disability. Eval Health Prof 2003 Mar;26(1):3-22
34. Carruthers et al. Myalgic Encephalomyelitis/Chronic Fatigue Syndrome: Clinical Working Case Definition, Diagnostic and Treatment Protocols
35. Jason et al, Comparing the Fukuda et al. Criteria and the Canadian Case Definition for Chronic Fatigue Syndrome, Jason et al.
38. White et al. Protocol for the PACE trial: A randomised controlled trial of adaptive pacing, cognitive behaviour therapy, and graded exercise as supplements to standardised specialist medical care versus standardised specialist medical care alone for patients with the chronic fatigue syndrome/myalgic encephalomyelitis or encephalopathy BMC Neurology 2007, 7:6doi:10.1186/1471-2377-7-6
40. S. Pierce & P.W. Pierce, The physiology of exercise intolerance in patients with myalgic encephalomyelitis (ME) and the utility of graded exercise therapy
41. Hawk, C., Jason, L.A., & Torres-Harding, S. (2007). Reliability of a chronic fatigue syndrome questionnaire. Journal of Chronic Fatigue Syndrome, 13, 41-66.
44. Fletcher et al. Biomarkers in Chronic Fatigue Syndrome: Evaluation of Natural Killer Cell Function and Dipeptidyl Peptidase IV/CD26. PloS ONE, May 2010
45. Joyce J, Hotopf M, Wessely S. The prognosis of chronic fatigue and chronic fatigue syndromes: a systematic review. QJ Med 1997;90:223-233.
This was part 3 of my Meta-Analysis of the Efficacy of Graded Exercise Therapy in Treating CFS
Part 4 is currently being written