A psychologist speaks on CBT for CFS

Deary, V., Chalder, T and Sharpe, M. The cognitive behavioural model of medically unexplained symptoms: A theoretical and empirical review. Clinical Psychology Review, 2007 Jul 17; [Epub ahead of print].  doi:10.1016/j.cpr.2007.07.002

The article is a narrative review of the theoretical standing and empirical evidence for the cognitive behavioural model of medically unexplained symptoms (MUS) in general and for CFS and irritable bowel syndrome (IBS) in particular. A literature search of Medline and Psychinfo from 1966 to the present day was conducted using MUS and related terms as search terms. All relevant articles were reviewed. The search was then limited in stages, by cognitive behavioural therapy (CBT), condition, treatment and type of trial.

Evidence was found for genetic, neurological, psychophysiological, immunological, personality, attentional, attributional, affective, behavioural, social and inter-personal factors in the onset and maintenance of MUS. The evidence for the contribution of individual factors, and their autopoietic interaction in MUS (as hypothesised by the cognitive behavioural model) is examined. The evidence from the treatment trials of cognitive behavioural therapy for MUS, CFS and IBS is reviewed as an experimental test of the cognitive behavioural models. We conclude that a broadly conceptualized cognitive behavioural model of MUS suggests a novel and plausible mechanism of symptom generation and has heuristic value. We offer suggestions for further research.  

Extracts: “The sine qua non of any CBT model is a vicious circle, the hypothesis that a self perpetuating interaction between different domains maintains symptoms, distress and disability.... Rygh et al. (2005) suggested that central mechanisms such as vigilance and attention, or the effects of anxiety, depression or stress, may dampen inhibition of these pathways, lowering the threshold yet further. This could lead to normally benign sensations being experienced as pain, leading in turn to further sensitisation and vigilance... Rief and Barsky (2005) acknowledged Dantzer's research but noted that whilst there is some evidence that immune changes can induce behaviour change, it is “unclear whether this causality can also be bi-directional and whether it contributes especially to the development and maintenance of somatoform symptoms in humans.” They observed that the biological findings in patients with MUS are highly varied and do not yet present a consistent and coherent picture. Whilst there is some evidence of increased autonomic arousal and of delayed recovery of the stress response, these findings vary between conditions and between different phases of the same condition. Therefore rather than anchor their explanation to a particular bodily system, Rief and Barsky (2005) proposed a more generic model in which symptoms arise through a two stage process of generation and selection. At the first stage, bodily symptoms may be generated by multiple determinants including over-arousal, chronic stressors, HPA activity, sensitisation etc. At the second stage a hypothetical filter system selects symptoms for conscious attention. This selective process will in turn have multiple determinants: health anxiety, depression, lack of distraction, uncertainty as to origin of symptom etc. These contribute to “faulty filtering” and increased symptom perception. Brown (2004) has suggested a similar model which emphasises the roles of attention, mis-attribution and mis-interpretation in the maintenance of MUS... A cognitive bias develops for symptoms, further amplifying them which serves to further sensitise “the neural loops supporting cognitive rumination… pain and illness lead to more pain and illness” (Ursin, 2005). These models are notably similar to Barsky and Borus's (1999) somatosensory amplification and Rief and Nanke's (1999) cognitive-psychobiological framework...” On the model itself: “An innate tendency to somatopsychic distress and ease of distress sensitisation, combined with childhood adversity, increase both the amount of symptoms experienced and lowers the threshold for their detection. Life events and stress lead to physiological changes which produce more symptoms and set up processes of sensitisation and selective attention. This further reduces the threshold of symptom detection. Lack of explanation or advice increases anxiety, symptoms and symptom focus. Stress cues become associated with symptoms through classical conditioning. Avoidance of symptom provocation, and symptom-led activity patterns, lead to further sensitisation through operant conditioning. The prolonged stress of the illness experience itself further activates physiological mechanisms, producing more symptoms, sensitisation, selective attention and avoidance. The individual can thereby become locked into a vicious cycle of symptom maintenance... The CBT model of CFS (see Suraway et al., 1995; Deary & Chalder, 2006) hypothesises that in vulnerable individuals, such as those who are over-active or under-stress, CFS is precipitated by life events or viruses leading to an autopoietic cycle in which physiological changes, illness beliefs, reduced and inconsistent activity, sleep disturbance, distress, medical uncertainty and lack of guidance interact to maintain symptoms. The evidence supports some of the individual dots in this picture but not yet the lines between them.... The nature of individual differences in susceptibility to MUS would also merit further attention. High neuroticism may offer an underlying common mechanism for distress sensitivity and intolerance which lowers the threshold for symptom detection (both mental and physical), and leads to increased propensity to conditioned responses, more attention to threat stimuli and more avoidant coping. The neuroticism concept captures many of the factors hypothesised to be at work in MUS. This would suggest two further lines of research. Firstly, more examination of the physiological, cognitive and behavioural markers of N might give us more insight into the processes at work in MUS. Secondly, longitudinal studies in N might give us a clearer idea of what mediates and moderates development of MUS in vulnerable individuals. The CBT model of MUS offers a previously undescribed illness mechanism maintaining a distinct group of disorders that we might call autopoietic conditions. The fundamental hypothesis underlying the model is that symptoms are maintained by a self perpetuating, multi-factorial cycle. Treatment is aimed at elaborating the unique inter-play of factors in any given patient and dismantling the autopoietic mechanism by making changes in target areas... Distress intolerance is as a key theme in the CBT model of MUS maintenance. Symptoms are perceived as aversive/threatening, which triggers a physiological response, which serves to maintain avoidance, symptom focus and symptoms. One hypothesis would be that developing the ability to tolerate symptoms whilst not letting them dictate behaviour would be a promising line of enquiry. The so called third wave of CBT, particularly Acceptance and Commitment Therapy (see Hayes, Strosahl,&Wilson, 1999), with its emphasis on distress tolerance and goal maintenance would seem well suited as a paradigm to apply to these conditions, as would the related intervention of mindfulness.”


[Ed. Note: This review relies on a reductionistic approach, i.e. the view that an illness is not a disease unless it has an identified physical cause, plus the assumption that if no such cause has  been identified in 20 years, there may not be a physical cause to be identified. To put it another way, here absence of evidence is taken as evidence of absence.

Given the definition of CFS selects a  heterogeneous population, there is unlikely to be a single cause. It is this which has led to inconsistent findings in biomedical research and the uncertainty regarding aetiology helps to feed the CBT theory.  Aside from the lack of a physical cause argument, the model also continues to rely on  the assumption of the ‘universal response’, which is invariably unhelpful (e.g. avoidance behaviours), coupled with  maladaptive personality traits (e.g. perfectionism, neuroticism). Thus there is no recognition of the research indicating the use of appropriate, adaptive coping strategies (e.g. Saltzstein et al 1998), let alone the fact that many patients engage in pro-active coping such as pacing (original version). Pro-active coping appears to be linked with positive outcomes in the same way that the strategies described in the paper are often associated with negative ones (Aspinwall and Taylor 1997). Furthermore, there is no recognition of individual differences in the appraisal of threats or traits such as optimism and resilience, and the model cannot explain the outbreaks (Raised N scores in the chronic phase after the Royal Free outbreak can be explained in terms of the somatic items common to both neuroticism and the illness itself. Research has shown that N scores are no different to those found in MS). 

Another  factor which makes a reappearance is the  notion of misattribution, which  assumes that the evidence of  ongoing pathology documented in ME and CFS is unreliable (e.g. Innes 1970, Lane et al 2003, Natelson et al 2005, Chia and Chia 2007), in contrast to the findings from personality studies (e.g. where only some met the CDC criteria, cf. White and Schweitzer 2000). There is also a notable reliance on generalisations, as before. The fact that one study found that people with IBS engaged in ‘all or nothing’ coping (p. 7) does not mean that this strategy is used in the same way by other patient groups, and the authors' reference to a “clinically prevalent belief” that this is the case in relation to CFS  is anecdotal. That is the kind of evidence they reject when offered by those outside the CBT school.

This more  sophisticated model still does not explain some of the neurological signs documented in ME, or the abnormalities found in other subsets. Raised levels of cytokines are linked to low cortisol and ‘burn-out’, but not to the evidence of ongoing infection (cf. Lane et al 2003, Chia and Chia 2007). However, the inconsistent findings relating to CFS is a problem which we must acknowledge, hence I will  focus in more detail on the strength of the psychological arguments. 

The carefully linked associations between different components of the model are elegant and seemingly persuasive unless the reader is familiar with the literature and is aware, for instance, that some references are speculative articles (e.g. Barsky and Borus), or reviews (Henningsen et al 2003) rather than reports of original research. In my view, the paper trivialises some of the symptoms, and given the paucity of evidence, overstates the significance of attention and attributions. Research suggests that most patients appear to have realistic attributions, and members of the CBT school themselves have found psycho-social attributions in their own studies. When reading this, I wondered if they assume these to be as faulty as somatic attritutions?

One question which was not answered here and in the first version is how a virus might trigger the illness, while attention, beliefs and low cortisol suddenly cause the exact same symptoms following the acute phase. This assumption that patients cannot tell when the disease has passed and psychological factors take over requires evidence of a personality which does not recognise differences in internal cues.

The model also fails to explain the links with exertion 24 hours to five days following activity, as indicated using objective measures, i.e. the evidence shows that the muscles are weak, and the symptoms are not solely a reflection of an (unfounded) fear of  exacerbations, cf Paul et al 1999. moreover, fatigue after five days is difficult to attribute to deconditioning.

This model recognises the complexity of CFS to a far greater degree than the first version but I remain unconvnced. It appears to have the subtext: patients are not to be believed, they misunderstand and exaggerate. This brings us back to the explanations of psychiatrists such as McEvedy and Beard and Barsky and Borus. They all clearly had feeble females in mind: poorly educated, gullible and without sense.  Though the authors cite a person questioning the direction of causality, the rest of the paper leaves readers with the clear message that all the associations are invariably one way.  Thus the psychological mechanisms cause the immunological and HPA abnormalities, the anxiety and depression  reflect the “distress proneness”,  and so on.  References to support the general use of avoidance coping are missing, possibly because avoidance behaviour in CFS tends to be limited cf Sisto et al 1998. The vicious circle model predicts that the symptoms would get progressively worse but in most patients with CFS, they do not. I am also left with the question which element of the model explains the improvements noted in a significant proportion of patients over time? Do people suddenly become less perfectionist? Do they become less sensitive to stress? The natural course of the condition is not discussed, and therefore requires no explanation. This is part of the modus operandi of the CBT school, (they never discuss findings such as McGarry et al 1994 and Paul et al 1999 for the same reason), and the new, more complex model, merely continues where the old, disproven one, left off. 

Other means of diverting attention and avoiding difficult questions involves recategorising findings. The Wallman et al paper on pacing is classed as a study of  GET even though the protocol permitted patients to stop the activity if they felt unwell, which is contrary to the principle behind GET, where patients are encouraged to continue and tolerate symptoms, at least for a few days. This is in line with the ‘no pain, no gain’ philosophy underlying  this intervention. However, I welcome the recognition that the treatment effects associated with CBT are “modest at best” (p.9) and the acknowledgement that links between the various components are far from solid. But the main impression of this paper is one of  missing information.   The careful selection of patients, e.g. using broad criteria, and the lack of measures of immune function etc are not mentioned as  factors which may have confounded results. The bias and selectivity reflect a lack of respect for the scientific process, e.g. appraising findings which do not support the model, discussing the evidence for alternative views and most of all, an accurate discussion of the expereinces and views of those who suffer from the conditions. These criticisms do not  preclude the possibility that there is a subset of patients within the CFS population whose symptoms are perpetuated largely by fear etc, and I accept that some patients may believe the misinformation on the internet and subsequently develop maladaptive beliefs, adding to the stress and undermining the HPA axis etc  as a result.

I wish  to encourage a greater tolerance of alternative views and more objectivity. I believe that these are essential to help us understand the aetiology of CFS, and to improve the standard of patient care. People require a range of opinions, hence this post. Please note that the above criticisms of specific arguments should not be taken as reflecting a personal preference for any specific theory.]

---------------------------------------------------------------------
Ellen M. Goudsmit CPsychol CSci AFBPsS

Tags: CFS, CBT, psychiatry, feeble, women, objective