By David Tuller, DrPH
Apparently, after 30+ years of promoting cognitive behavior therapy for chronic fatigue syndrome (so-called), Professor Trudie Chalder has discovered something new: Patients worry a lot! And those with worse symptoms tend to have higher worry levels! She believes CBT for CFS can be tweaked to make a difference. She presents this information in a new study, published in the journal Behavior Therapy–yet another in which she and her colleagues disavow the possibility of inferring causality but use causal language anyway.
The article, posted a few days ago, is called “Generalised worry in patients with Chronic Fatigue Syndrome following Cognitive Behavioural Therapy–a prospective cohort study in secondary care.” Sten Helmfrid made me laugh with this tweet about it: “Professor Chalder concludes that patients with #ME/CFS worry a lot and suggests that CBT should focus on the issue. One of the reasons that patients worry is the large number of substandard CBT papers that are pumped out. I don’t think CBT is the solution!”
The study involved a cohort of 470 people with CFS [I am using this name in this post because the study uses it] attending a specialty care clinic between 2007 and 2014 and receiving CBT. Participants completed self-report measures, including a measure of “generalised worry,” at baseline, discharge from treatment, and follow-up at three months and six months. More than 70% met the study’s criteria for “generalised worry” at baseline.
After CBT, the average worry score marginally improved, although there was a huge participant drop-off at six-month follow-up–less than half returned questionnaires. Not surprisingly, higher worry scores were associated with worse scores for fatigue, anxiety, and work and social adjustment. The authors recommend that CBT for CFS should specifically target “generalised worry” alongside whatever else it is supposed to be doing.
The article is based on the questionable notion that it is possible to tease out excessive or abnormal levels of “generalised worry” from people’s experience of having a chronic disabling disease–especially one that is routinely misinterpreted by people like Professor Chalder as psychologically and behaviorally driven. To measure the core construct, Professor Chalder and her colleagues used an instrument called the Penn State Worry Questionnaire (PSWQ), which asks respondents to consider 16 statements characterizing their level of worrying.
Here are some examples: “Many situations make me worry,” “I am always worrying about something,” and “I worry all the time.”
It stands to reason that having CFS or any major illness or disability would impact people’s responses to many if not all of these questions about how much they worry. To address concerns about the appropriateness of the instrument for this study population, the authors note that “the PSWQ is considered a valid and reliable measure of problematic worry for patients with CFS.”
First, what does that mean? For CFS patients specifically, what would be defined as “problematic worry” as opposed to appropriate or non-problematic worry, and on what grounds? And no citation or reference is provided for the assertion. Who is it that considers the PSWQ to be a “valid and reliable measure” of this construct for this patient group, and based on what data?
A search of PubMed for the PSWQ plus “chronic fatigue syndrome” or “myalgic encephalomyelitis” yields no results. Yet an ingenious sleuth in this field alerted me to one such study—“Hyperventilation in patients with chronic fatigue syndrome: The role of coping strategies.” Was this 2007 paper in Behaviour Research and Therapy what Professor Chalder and her colleagues had in mind?
Given that the authors of this 15-year-old article included the PSWQ among their multiple outcomes, I suppose one could argue they considered it to be “a valid and reliable measure” in this context. But they did not discuss why they considered the PSWQ to be appropriate for this particular population, they did not mention the construct of “problematic worry,” and they only briefly mentioned the results for this scale. It would therefore not be accurate to suggest that they validated the scale for use with a CFS sample.
Beyond that issue, the new study has no comparison arm and is not designed to allow for causal inferences. In the limitations section, the authors acknowledge this: “There was no control group and we did not randomize our sample, therefore it is not possible to draw conclusions about the effectiveness of CBT for CFS, or generalised worry.“
Yet the paper includes language implying causation. The discussion includes this: “This prospective cohort study aimed to explore the prevalence of generalised worry in people with CFS and examined whether worry changed as a result of CBT specifically tailored to target fatigue and social functioning.”
If the study design does not allow for conclusions to be drawn about the effectiveness of CBT for CFS or “generalised worry,” as the paper argues, then how could the research assess whether “worry changed as a result of CBT”? It cannot. It is measuring associations only, not whether one thing changed “as a result of” another.
In a 2020 paper also on clinical outcomes of CBT, Professor Chalder, Professor Sir Simon Wessely, and colleagues engaged in similar linguistic slippage, noting the inability to make causal assumptions while simultaneously making causal assumptions. That published was published by the Journal of the Royal Society of Medicine, whose editor has since been named to head BMJ. Psychology professor Brian Hughes and I published a rebuttal to that paper in the Journal of Health Psychology after the JRSM rejected it.
As in that paper, the authors are contradicting themselves on the question of causality. Either they didn’t notice, or they didn’t care. Apparently this little detail also slipped by the journal’s peer reviewers and editors. These lapses are troubling–but par for the course with this cabal.