Trial By Error: Trio of Trials Shows Limits of CBT for Medically Unexplained Symptoms

Lancet Psychiatry recently published the results of a high-profile trial of cognitive behavior therapy as a treatment for so-called dissociative seizures, also known as psychogenic non-epileptic seizures. The trial, nicknamed CODES, found that CBT had no impact on seizure frequency–the primary outcome. The average number of seizures per month dropped in both the treatment and comparison groups, with no statistically significant difference between them.

With this study, British investigators from the so-called biopsychosocial (BPS) camp have completed an impressive trifecta of major, presumably definitive trials of CBT as a treatment for so-called €œmedically unexplained symptoms€ (MUS). The others include the PACE trial for chronic fatigue syndrome, with key results published in 2011 and 2013, and a trial for irritable bowel syndrome, with results published last year.

As a group, this trio of trials demonstrates the ineffectiveness of cognitive behavior therapy as a treatment for the kinds of MUS targeted by the UK National Health Service’s Improving Access to Psychological Therapies (IAPT) program. The metastasizing IAPT program began a decade ago to boost mental health care for those suffering from depression and anxiety. It has since expanded to offer psychological treatment to patients with long-term conditions, such as cancer and multiple sclerosis, and those diagnosed with MUS.

In implementation guidelines, IAPT splits MUS into sub-categories, including chronic fatigue syndrome, irritable bowel syndrome, and “not otherwise specified.€ The latter would presumably include functional neurological disorders (FNDs), of which dissociative seizures are an example. The implementation guidelines recommend specialized variations of CBT as treatments for MUS.

Northwestern University law professor Steven Lubet and I have recently published a commentary on the need for humility in making categorical statements regarding conditions of unknown etiology. And a recent series of articles on the advocacy site Opposing MEGA has highlighted estimates of high misdiagnosis rates for various subgroups of MUS, although proponents of the construct tend to downplay or ignore these data.

In any event, the concerns raised would matter less if the evidence indicated that an intervention like CBT were an effective treatment for the conditions in question. So let’s review.


As is well-known, the PACE trial reported benefits from CBT that turned out to be largely illusory, according to reanalyses of the study’s main findings. (I was a minor co-author on the main paper debunking the findings.) The lead investigators weakened their outcome measures for improvement and recovery, providing inadequate and even laughable reasons for having abandoned their own protocol methodology.

In any event, the international community has spoken on PACE. An open letter to The Lancet, signed by more than 100 scientists, clinicians and other experts, denounced the study’s €œunacceptable methodological lapses,” and the US Centers for Disease and Control and Prevention has removed recommendations based on the trial.

Last year, a team from King’s College London published two papers from the largest trial of CBT for irritable bowel syndrome, with 558 participants. Specifically, the trial tested a web-based CBT program that the investigators had spent many years developing and piloting. The first paper, in a BMJ journal called Gut, featured the 12-months results. The second, in Lancet Gastroenterology & Hepatology, included the 24-month results.

As I have reported at length, the web-based CBT program produced statistically significant but clinically insignificant benefits over treatment-as-usual in reducing IBS symptom severity at 12 months. At 24 months, any benefits were neither statistically nor clinically significant. Yet a start-up company based in San Francisco licensed the product from King’s College London and has falsely promoted it as effective in reducing IBS symptom severity. (A course of telephone-based CBT produced slightly better results at 12 months, although by 24 months these results were also clinically insignificant. In any event, the web-based version is what King’s College London licensed as a commercial product.)

The web-based program produced modest improvements in scales measuring more generic domains, such as work and social adjustment. It is not surprising that a course of CBT, whether in person or online, might lead to an improvement in reported ability to adapt to or cope with an illness, especially in an open-label trial relying on subjective outcomes. But that does not mean it is accurate or ethical to present or market this intervention as a treatment for the illness itself, especially given, in this case, the poor results for the core outcome of symptom severity.

Finally, we have CODES. While the primary outcome yielded null results, the investigators reported positive findings in nine out of 16 secondary outcomes. Most of these secondary outcomes assessed subjective generic domains, such as social adjustment, emotional states, quality of life, and attitudes toward treatment; subjective outcomes in open-label trials are highly susceptible to bias. After correcting for multiple comparisons, only five of the 16 secondary outcomes were statistically significant. (More on these secondary outcomes in another post.) And yet, on the basis of these secondary outcomes, King’s College London issued a deceptive press release touting the success of the trial.

A commentary published alongside the trial report, written by another expert in the FND field, adopted the convenient position that seizure frequency wasn’t the best primary outcome for dissociative seizures anyway. The author did not mention that CODES investigators had been promoting seizure frequency as the best primary outcome since at least 2010, when they published a pilot study of their CBT program.

For what must be a range of reasons, MUS experts maintain a persistent faith in CBT’s healing powers as a treatment rather than as helpful but limited adjunct support. Their own research demonstrates that this faith is not warranted. The body of evidence from these three studies is consistent: CBT is not an effective treatment for chronic fatigue syndrome, irritable bowel syndrome, or dissociative seizures, although it might help some patients feel a bit better in some ways.

Comments are closed.

Scroll to Top