Chronic fatigue syndrome: bias in the BMJ.
Ellen Goudsmit. PhD. C.Psychol.
Health psychologist and archivist.
Chronic fatigue syndrome (CFS) is a common, potentially disabling illness, which carries a substantial socio-economic burden1. The true prevalence of this disorder is unknown but may have been underestimated in the past2. Without doubt, the condition deserves serious consideration both from the research scientists and physicians in clinical practice.
The British Medical Journal (BMJ) is the official organ of the British Medical Association, the largest professional body of physicians in the UK. Readers expect the journal to publish original research on CFS and to keep them up-to-date with developments documented elsewhere. The publication of a broad range of views allows practitioners to make informed decisions and is an essential part of the scientific process. For clinical and epidemiological purposes, patients with CFS are currently defined by the CDC criteria, which were developed by the International study group and modified in 19943.
In September 2000, I reviewed the content of the publications on CFS, which have appeared in the BMJ since 1995 to see if the nature of the papers reflected the global research, clinical opinion and changes in the diagnostic approach in relation to this condition.
Methods and Results
A search of Medline for publications in the BMJ between January 1995 and August 2000 on chronic fatigue syndrome and myalgic encephalomyelitis (ME) identified 41 articles and letters. They included 6 original papers, three editorials, one review, a case history, a book review and a number of letters. A similar search using the BMJ’s own search engine plus an independent database identified three additional items, all written by the BMJ editorial staff. To analyse the content of the published papers and the diagnostic criteria used by the authors in case selection, I targeted original research papers (including short papers), editorials and review articles identified by this search. There were six papers featuring original research4-9.
Not a single paper used the latest CDC-criteria for case selection. Where reported, the definitions used for case selection in these papers were the Oxford criteria10 though one paper did not make it clear except to note that the patients fulfilled criteria for "neurasthenia"6. Only one report of original research included findings inconsistent with a psychiatric explanation4. One challenged the adequacy of a measure to assess abnormal illness behaviour5 while two supported the management of CFS using cognitive behavioural therapy (CBT) and/or graded exercise7,8. In a paper, which confirmed previous observations of altered neuroendocrine control in CFS, the authors concluded, without any direct experimental evidence, that the abnormal responses might be the result of prolonged inactivity or a disturbance of the sleep-wake cycle6. Moreover, a short report misrepresented the illness ME9, thought to be a subgroup of CFS.
The only review during this period was an extract taken from Clinical Evidence. All the authors were mental health professionals11. There were also three editorials on the illness12-14. One was written by two psychiatrists12. Another dealt with childhood CFS and was written by a paediatrician13. The third, a commentary on the Royal Colleges Report, was written by a virologist14. All three editorials expressed views consistent with a psychiatric explanation of CFS. The book review covered a text co-authored by two of the aforementioned psychiatrists and was highly complimentary15. However, given that it was written by a former colleague of one of the authors (not declared), this is perhaps not surprising.
Between 1995-2000, none of the published papers on CFS used the currently accepted international diagnostic criteria (CDC) to define their patient population. Since the Oxford criteria are less specific than all the other published guidelines, the conclusions of these papers cannot be applied to all those suffering from CFS. The paper on abnormal illness behaviour was an appropriately designed study since it compared CFS patients with people suffering from multiple sclerosis (MS)5. However, the paper by Lane et al was the only report focusing on the possible role of non-psychiatric influences on fatigue in CFS4. It was the only research of its kind published in the BMJ during the past five years.
Most editors of medical journals respect the need for balance in the content of papers, particularly on controversial topics such as CFS. However, in relation to CFS, the editors and reviewers have clearly leaned towards the psychological and psychiatric aspects during the period in question. The fact is that most of the papers in the journal have emphasized the role of inactivity, mood disorders and/or maladaptive beliefs. There were no papers on the immunological or virological aspects nor any item referring to new research on these topics published elsewhere. I believe that the result has been to give readers the impression that the majority of patients with CFS are suffering from phobic avoidance and following unhelpful advice.
Responding to earlier suggestions of bias, the editor had claimed that: “We don’t consider ourselves to be pushing any theory: we are simply sorting among the 5000 papers submitted to us to find the best”16. However, this is hard to believe, given the inadequacies in the various papers supporting the psychiatric view. For example, Sharpe et al7 indicated in their report that the patients who were going to receive CBT spent more than twice the amount of time in bed compared to the comparison group, despite having the same level of disability and fatigue. This and one other finding suggest that the researchers had inadvertently included more people with psychological problems in the treatment arm, and this alone could explain their favourable response to CBT. The paper on graded exercise was similarly flawed in terms of patient selection and did not provide any long-term follow up8.
The claim to publish ‘best evidence’ is also difficult to reconcile with the uncritical review of treatments11. Firstly, this overlooked a number of relevant trials in order to support their contention that CBT was “effective” and that graded exercise produced “substantial improvements”. Secondly, it disregarded notable flaws in the ‘successful’ trials, such as the fact that the only symptoms assessed were fatigue and emotional distress and that all had included patients with psychiatric disorders; a subset likely to respond to CBT and exercise7 8. Thirdly, the review failed to note that the follow-up of Sharpe et al had not confirmed the initially reported differences between the groups. In my view, this paper was partial, biased and clearly misleading.
Another example is the editorial, which suggested that many doctors still advocate the "rest cure" and implied that most patients take that advice12. The fact is that the strategy described has not been recommended in relation to the clinical management of CFS in any medical journal in the past two decades, nor has it been advocated by the two national patient groups during this time. The most commonly used coping strategy is in fact pacing, but this was not mentioned here, or in the review. Sadly, errors are not always corrected, nor flaws discussed. Although multiple criticisms and shortcomings of the review of treatments were reported in the electronic form of the journal (eBMJ), none were published in the paper version despite the promise made by the BMJ to a colleague. Moreover, in a personal communication to another colleague, the editor of Clinical Evidence accepted that there were shortcomings in the review but this was not made known to the readers of the BMJ.
There is no question, given the evidence, of interpretative flexibility. Anyone replicating the analysis will find the same studies, with the same conclusions. A trial on cognitive-behaviour therapy is not a study on the role of enteroviruses, and conclusions that chronic fatigue may be perpetuated by inactivity cannot be interpreted as a suggestion that the illness may be complicated by an abnormal immune response. In this respect it is also worth noting that I am not the first to have documented a far from balanced editorial policy in the BMJ. An independent report published more than seven years ago, noted that the mainstream British medical journals tended to limit their coverage of CFS to papers favouring a psychological explanation17.
Another sign of editorial bias can be found in articles written ‘in-house’. In the past, one of the BMJ editors revealed his support for the psychiatric explanation and reinforced the negative stereotype of the CFS patients by writing a factually incorrect commentary18. For instance, he claimed that the patients with CFS had manipulated the World Health Organization (WHO) and persuaded them to "include myalgic encephalomyelitis under the diseases of the nervous system in ICD-10." This is simply untrue, as a call to the WHO would have confirmed. He was also incorrect when he claimed that "supporters of myalgic encephalomyelitis.... landed a Myalgic Encephalomyelitis Act on the British statute books, requiring an annual report to be made to parliament on its causes, effects and treatment". There is no ME Act in the existing British statute books.
Taken in conjunction with this background, my knowledge of papers which have been rejected and the fact that between 1995-2000, only one paper was published in the BMJ linking CFS with a non-psychological aetiology, there remains little doubt that the editorial policy of the BMJ is uncritically supportive of the psychiatric view of CFS. This has seriously compromised the quality of information provided on CFS to the readers of the BMJ. My analysis supports the view that the journal’s has consistently ignored the non-psychiatric professional views on CFS. Based on my knowledge of this illness, I am unable to find sufficient scientific reason to justify this stance.
Flawed scientific research is harmful since the principle of self-correction by science cannot be applied. In addition, the editor's decision on the selection of scientific communications and choice of authors for review articles and editorials on CFS should not only be fair, but must be seen to be fair. I was unable to see this transparency or balance of opinion on CFS in this search of papers published during the past five years in the BMJ.
A request for a more evidence-based, as opposed to theory-led editorial policy was rejected by the editor. Arguments included that the ‘preference’ was no different to that in similar journals, and that it reflected editorial freedom.
We accept the principle of editorial freedom, but should a mainstream medical journal have a policy, which promotes one theory at the expense of others? Should a scientific publication censor all the research supporting alternative theories? Should the editor overlook major flaws and disinformation as part of the policy? Does all this not undermine the scientific process? If the editor is to have this ‘freedom’, should the journal not declare its policy so readers are aware of it and realise that they need to seek information on the immunological and virological aspects elsewhere? And should that journal claim to support evidence-based medicine when it selects articles on one illness largely on the basis of the editor’s personal preferences?
Details of analysis of other journals (same time period)
Since this analysis was completed, there has been no apparent change in editorial policy. Thus there have been no original reports of neurological, immunological or virological findings, nor items alluding to such research published elsewhere. The BMA Ethics Committee was unwilling to consider a complaint regarding the journal’s editorial policy.
This paper was written with help from two medical colleagues, who wish to remain anonymous.
Chronic fatigue syndrome: bias in the Lancet
Evidence of bias
I conducted a Medline search of papers in the Lancet from January 1995 to August 2000. This identified 38 articles and letters, including four papers written by mental health professions from King’s College Medical School and their colleagues (1-4). Except for one small news item, all details of the research on the immunological and neurological aspects of the illness and alternative explanations were relegated to the letters page. Two editorials challenged specific issues: domination of psychiatrists in the Royal Colleges Working Party and the flaws in a study on hydrocortisone (5-6).
To assess if the original papers represented ‘best evidence’, two medical colleagues assessed the quality of the publications. One was a trial of hydrocortisone on patients with normal cortisol concentrations, who arguably did not require and would not benefit from supplementation (1), while the second was a longitudinal study of “post-infectious fatigue” where only six cases actually became ill following infection (2). Both these studies emanated from the team at King’s. The third – from Dutch proponents of the CBT model - reported the response of a heterogeneous sample to fluoxetine (3). Lastly, a review hypothesized that CFS was indistinguishable from conditions such as tension headaches and globus syndrome, which not only misrepresented the illness but trivialised it (4). Arguments favouring the recategorisation of CFS as a functional somatic syndrome included the preponderance amongst women and the prevalence of child abuse, although the latter has not been documented in controlled research on CFS and certainly not outside America.
The apparent preference for papers consistent with a psychological explanation is a relatively new phenomenon as an analysis of papers from 1990-1995 shows.
In this time period, I found 34 items consisting of 4 studies, one viewpoint article (discussing the psychiatric aspects of the illness) and 20 letters and short reports. The four studies included two epidemiological surveys (from King’s College CFS Clinic), one report on magnesium and one article discussing the evidence of immune activation found in strictly-defined CFS. The letters addressed additional issues such as the possible role of HHV-6 reactivation, Gilbert’s syndrome as well as psychological issues and neurally mediated hypotension.
Altogether, the publications covered a wide range of opinions.
An analysis of papers from 1.1.85 to 31.8.90, before the increase in interest in CFS, identified no original papers. There were, however, 14 letters, discussing a variety of findings, including changes in interferon and red blood cell morphology and a follow-up report on those involved in the Iceland epidemic.
Like the BMJ, the Lancet has denied its readers the opportunity to make informed decisions based on the existing scientific literature. Instead, it has chosen to publish papers that present a limited view of the illness, the research, and the experiences of physicians who care for these patients.
Readers expect the Lancet to provide reasonably impartial and authoritative information on all diseases. However in terms of CFS, it has failed to live up to its reputation.
Since this analysis, the situation has not changed. The Lancet carry a purely descriptive account summarizing the report from a Working Group to the Chief Medical Officer (7) but there was no original research on neurological, virological or immunological aspects of the illness. However, a paper on the efficacy on CBT was published in 2001 (8).
The editor also chose to publish an inaccurate and rather mischievous letter from Mouterde (9), which claimed that ME only existed in the UK and that it might therefore be an invention of British culture and media. Yet, a check using Medline would have revealed recent articles on CFS from Belgium, Netherlands, Italy and Spain. Three letters noting the existence of ME in Europe were published, but there was a notable delay (10-12).
With thanks to my colleagues who wish to remain anonymous.
Effects of the bias, or what readers may not know about CFS.
The lack of balance disempowers doctors who rely on journals like the Lancet to provide them with high quality medical information. While the journal has focused its attention firmly on the psychiatric\cognitive-behavioural aspects of the illness, experts around the world have been discussing the limitations of the current case definitions and the issue of subgroups among CFS patients. How many readers of the Lancet will be aware of this? Indeed, how many are will be aware of the evidence undermining the psychiatric theories?
Take the much cited cognitive-behavioural model developed by Professor Wessely (King's College) and Dr. Sharpe (now at Edinburgh). This assumes that whilst CFS/ME may be triggered by an external agent such as a pathogen, it is primarily perpetuated by psychosocial factors and/or physical inactivity. The proponents of this theory have claimed that the resulting deconditioning induces fatigue at increasingly lower levels of exertion which is then wrongly attributed to ‘physical’ disease. Patients become helpless, hopeless and depressed, which further exacerbates the fatigue and so on. This model erroneously assumes that all patients share the same maladaptive beliefs and behaviours. It therefore ignores individual differences, as well as all the research which shows this explanation only applies to a small proportion (one quarter) of this patient population13. As the following illustrates, failure to consider these variables has given a totally misleading view of the illness and its treatment.
The Lancet readers are not aware of this editorial preference, so many will assume that the published articles were judged primarily on the basis of the ‘best evidence’. As a result, they may interpret the absence of information on the somatic aspects of CFS as reflecting a general absence of research or scientific knowledge. The readers therefore will not know that there are subgroups of CFS patients and that the psychological model may not be the only explanation of this illness. Indeed, we are aware that due to this paucity of knowledge, some physicians have offered their CFS patients advice and treatment options which are neither appropriate nor helpful.
Copyright Dr. Ellen Goudsmit. 2003. ©
All rights reserved. This article
may not be reproduced without
Be sure to see the many other valuable articles at our Main M.E. Home Page