CBT Model of Medically Unexplained Symptoms, Explained; CBT Trial for Q-Fever Fatigue

By David Tuller, DrPH

As I have recently written, four major clinical trials of CBT for so-called MUS have documented the opposite of what the investigators hoped to prove. In fact, the evidence from this research suggests that CBT is not an effective treatment for these conditions. That hasn’t stopped these investigators from claiming otherwise, of course. As my earlier post indicated, they have deployed a range of methodological, statistical and rhetorical strategies to obfuscate or downplay their poor results. Three of these studies were based at King’s College London, and one—the now-discredited PACE trial—at Queen Mary University of London.

A 2007 article in the journal Clinical Psychology Review“The cognitive behavioural model of medically unexplained symptoms: A theoretical and empirical review”–sheds some light on the background and possible genesis of these various trials. (One of the co-authors of the review was a lead investigator in all four of the MUS clinical trials.) This review outlined the rationale behind the CBT treatment approach to MUS, including specifically chronic fatigue syndrome (CFS) and irritable bowel syndrome (IBS), and suggested that these conditions are essentially self-sustaining. The review also helped clarify–for me—why I get a sense of déjà vu whenever I read another article from this group of investigators. Each one basically says the same thing, except with a change in the targeted condition.

Sure enough, the authors of the 2007 paper included a reference to a favorite meme among those who share their theoretical orientation: the “vicious circle” that purportedly leads to a downward spiral of increasing misery. Here’s how the authors presented this bedrock element of their construct: “The sine qua non of any CBT model [for MUS] is a vicious circle, the hypothesis that a self perpetuating interaction between different domains maintains symptoms, distress and disability. Irrespective of the symptom type…the CBT models of MUS, IBS and CFS propose a model of perpetuation that is, to borrow a term from systems theory and cell biology, autopoietic.”

These CBT models, noted the authors, provided guidance for studies of specialized interventions that could disrupt the “vicious circle” and, in doing so, lend support to the theory: “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. In as far as treatment studies are a test of this hypothesis, it has good but at present only indirect support. More direct research support, based on testing this theoretical model of MUS is needed.”

The review focused on various elements that have been hypothesized as part of the process of perpetuation, such as a heightened focus on symptoms and attributing illness to somatic causes. Illness related beliefs, the authors noted, have emerged as “potentially important factors” in the persistence of symptoms. The hypothesis, they explained, “is that the beliefs inform behaviour, in this case activity avoidance, which in turn affects physiology and symptoms, providing the rudiments of a vicious circle of symptom maintenance.”

This hypothesis about the generation of the “vicious circle” assumes that the beliefs lead to the symptoms, with behavior as the mediator. It is just as reasonable that causal relationships run in the other direction. Perhaps those with a serious illness–especially one characterized by the symptom of post-exertional malaise–accurately believe that activity is potentially harmful, and it should not be surprising that they would have poor outcomes.

In any event, as the 2007 paper suggested, the proof is in the pudding—that is, in the findings of trials designed to test interventions based on the CBT hypothesis. The authors’ appeal for more research appears to have been successful, given the quartet of big trials that have investigated this MUS model. For each study, the investigators published a protocol that cited supposedly promising preliminary data and presented the proposed research as necessary for obtaining authoritative evidence for clinical decision-making and public health policy.

And yet, when faced with null or minimal results, the investigators have not questioned their own attachment to their theoretical framework. Instead, they have found ways to minimize or ignore or argue away the reality of their findings—that CBT does not work as intended.

This is not how science is supposed to go. “Self-correcting” is supposed to mean what it says. But the onus isn’t really on “science.” It is on scientists. And if scientists refuse to self-correct no matter how poor their results, then what’s the point?


CBT Trial for Q-Fever Fatigue Syndrome Also Generates Inflated Claims

Looking beyond the UK, a 2017 study from a team of Dutch investigators–a large trial of CBT for prolonged fatigue after an acute bout of Q fever, a bacterial infection–could fit in with this MUS quartet. This phenomenon, known as Q-fever Fatigue Syndrome (QFS), afflicts a significant minority of patients after an acute bout of Q-fever. The Netherlands experienced a Q-fever epidemic in the late 2000s, providing an opportunity to conduct a large clinical trial of CBT for QFS. The investigators published a trial protocol in 2013. In 2019, they published both a long-term follow-up and a mediation analysis. The trial also tested a regimen of antibiotics.

(A robust critique of the trial, by Dr Mark Vink and Alexandra Vink-Niese, outlined serious methodological concerns that undermined the claims of CBT’s effectiveness. Lou Corsius also deconstructed the study effectively in two posts on his blog It’s About ME, here and here.)

The 2017 trial report in Clinical Infectious Diseases noted that the CBT approach to QSF was based on its purported success for CFS: “Cognitive-behavioral therapy (CBT), aimed at fatigue-related cognitions and behavior thought to perpetuate symptoms, can reduce symptoms and improve functioning in CFS. A considerable overlap in fatigue-perpetuating factors between QFS and CFS implies that CBT might also reduce fatigue severity in QFS.”

And here is how the intervention was described: “First, the model of fatigue perpetuating beliefs and behaviors is explained to patients. At the start of the therapy patients formulate their goals in behavioral terms. These goals usually include the resumption of work, hobbies, and other activities that imply that the patient is no longer severely fatigued and disabled, which is the goal of CBT for QFS…During the sessions, the therapist elicits and challenges patients’ non-accepting and catastrophising beliefs with respect to fatigue. Additionally, patients are taught how to distract their attention from their fatigue.” 

The primary outcome was fatigue at 26 weeks—the end of the treatment period. (The study used a different fatigue scale than the one used in the PACE trial, the Chalder Fatigue Scale.) The difference between the mean scores of the CBT group and the placebo group on the fatigue scale was modest–less than what was deemed a “clinically meaningful improvement” on the measure for an individual. Even these small reported benefits wore off by the one-year follow-up study, as reported fatigue increased again after treatment among those who had received CBT.

The study also measured activity levels objectively by having participants wear actometers for a period of time at the beginning and end of treatment. For some reason, this measure was not officially a primary or secondary outcome but was included as a potential mediating variable. In any event, the intervention did not lead to benefits in physical activity levels, despite the subjective reports of improvement in fatigue. The investigators did not report those disappointing data until the mediation analysis paper published two years after they had already claimed success.

Even after the subsequent null results for the primary outcome in the one year follow-up study, the investigators continued to argue that CBT should be recommended. Doubling down on their approach, they called for research into ways to maximize the supposed early benefits, perhaps with “booster sessions” of CBT. Given the anemic results of this major trial, others might reach a different conclusion.





5 responses to “CBT Model of Medically Unexplained Symptoms, Explained; CBT Trial for Q-Fever Fatigue”

  1. CT Avatar

    The 2007 paper by Deary, Chalder and Sharpe claims that a 2000 paper by Nimnuan, Hotopf and Wessely highlights that a patient’s poor relationship with their GP can lead them to consult more and report more symptoms. Unless my eyes deceive me, the 2000 paper did not highlight that at all. Rather, its authors appear to suggest that doctors’ negative feelings towards patients makes the doctors more likely to suspect their patients of having medically unexplained symptoms (MUS) and more likely to misdiagnose them with it.

    The way this group of CBT proponents receive the same scientific literature as the rest of us but interpret it quite differently reminds me of the “telephone game”. Are they metaphorically hard of hearing? Could they have illness related beliefs that impact on their ability to interpret/relay the literature correctly?

  2. Alicia Butcher Ehrhardt, PhD Avatar
    Alicia Butcher Ehrhardt, PhD

    They can’t get more research funding which pays their salaries if they don’t show positive results.

    THAT is the vicious circle.

  3. Milo Avatar

    Thank you for laying it all out David. Patients get sucked into programs not knowing the harm that is done to them- they are desperate and want to please their doctor because it means they have someone to go to if and when things get worse. It is happening all around the world, not just in the UK. Il is insidious and because it is introduced as “education” patients don’t even flinch.

  4. Lou Corsius Avatar
    Lou Corsius

    Contrary to what the authors claim, there is no beneficial effect as a result from CBT. The initial fatigue severity score in the CBT-group on CIS F (31,6) was very poor in itself. As we have shown in part 1 of this comment, a score of 31.6 is still far above a score of <27 that is used in other fatigue researches.[iv] (a higher score is worse)The fact that the SIP 8 scores at Endpoint were still very high, indicates that patients did not function very well. The therapy has not been beneficial at all. These scores are still high enough for the patients to be included in research again.At follow-up the score on CIS F had got worse for the CBT-group. There is no longer a statistically significant difference between the CBT-group and the Placebo-group on CIS F. When we look closer we can conclude that the Placebo-group scored better than at Endpoint and now scores even better than CBT at Follow-up.We see the same effect on the SIP 8 scores: the scores of the CBT-group got worse, while the Placebo-group scores ameliorated. The SIP 8 scores in the Follow-up group are now better than those of the CBT-group.The better results in the Placebo group, although not yet statistically significant, confirm earlier findings from the scores of CFS patients. In the study by Bazelmans et al[v] the authors phrased it as follows: for functional impairment, the effect was opposite to what was expected. Looking at the improvement and the better scores on SIP 8 and CIS F in the non-intervention/Placebo group in the Qure study, this is once more an indication that CBT with a graded activity protocol is impeding the naturally-occurring recovery process.The findings in this follow-up study confirm the concerns of patients suffering with QFS as well as with CFS: the CBT treatment with a graded activity protocol will lead to deterioration because patient learn during this treatment not to trust on the signs of their body that they have overstretched their possibilities. In the long end this will lead to deterioration. Although authors in several studies contest this conclusion and claim CBT treatment with graded activity is safe, these findings seem to confirm the experiences of the patients.

  5. Barbara McMullen Avatar
    Barbara McMullen

    Why isnt’ there a rule that all the analysis based on all the data be published in the same article at the same time? Why are they allowed to get away with publishing little bits years later?