By David Tuller, DrPH
Had things gone as planned, the PACE trial should have been able to serve as proof that so-called medically unexplained symptoms (MUS)—in this case what the investigators referred to as chronic fatigue syndrome–could be successfully treated with psychological and behavioral therapies. The Lancet published the first PACE results, which reported benefits from cognitive behavioural therapy and graded exercise therapy, in early 2011. The timing dovetailed with the National Health Service’s expansion of a program called Improving Access to Psychological Therapies (IAPT).
Launched in England in 2008 to increase links between medical care and mental health services for those identified as suffering from depression and anxiety disorders, IAPT’s mandate next expanded to include other groups of patients, specifically those suffering from either long-term conditions, such as cancer, and from conditions identified as part of the MUS category. Many academic promoters of IAPT are colleagues of key PACE investigators. They are also fellow proponents of the CBT/GET paradigm for the illness, or cluster of illnesses, variously referred to as CFS, myalgic encephomyelitis, ME/CFS and CFS/ME.
Unfortunately for these folks, many segments of the international scientific community have rejected the PACE findings after public debate over the trial’s many documented missteps. In 2017, the US Centers for Disease Control and Prevention dropped its recommendations for CBT and GET after having already removed references to PACE from its website. Two years ago, BMC Psychology published a widely read paper rebutting the main PACE claims, based on raw trial data that were released only after a patient waged a lengthy battle for access under the freedom-of-information law. (I was a co-author of that paper.) Such developments have made it challenging to cite PACE as authoritative support for IAPT expansion.
Not to worry! Those in favor of IAPT expansion are busy demonstrating that their preferred approach can address other kinds of MUS. Last fall, Lancet Gasteroenterology and Hepatology published a 24-month follow-up study of a clinical trial of CBT for irritable bowel syndrome (IBS), another condition lumped into the MUS category under the IAPT framework. The 12-month results of the study–called Assessing Cognitive behavioural Therapy for IBS, or ACTIB–were published earlier last year in Gut, a BMJ journal. While the investigators did not engage in some of the unusual methodological and ethical lapses documented in PACE, they nonetheless failed to create a compelling case for these CBT interventions.
The study included 558 participants with “refractory” IBS symptoms, defined as “ongoing symptoms after 12 months despite [the patient] being offered appropriate medications and lifestyle advice.” The investigators randomized them to three groups, with all receiving “treatment-as-usual” (TAU)—in other words, the standard care that had failed to resolve their complaints. Participants in two of the groups also received CBT–in one group through telephone sessions with a therapist, and in the other through a lower-intensity web-based program involving a lesser amount of therapist contact.
In subjective questionnaire responses, both CBT groups showed more improvement than the TAU-only group. That trend was sustained, to a lesser degree, in the 24-month follow-up. The investigators concluded that “increasing access to CBT for IBS could achieve long-term patient benefit.”
Let’s stipulate that stress reduction, however achieved, is generally a positive goal and a benefit to health. If the biochemical correlates of stress make people worse over the long-term no matter what their state of health or sickness, then strategies that might lower or minimize stress, including possibly CBT and other forms of psychotherapy, could easily lead to bonafide reports of improvements.
The IBS trial, like PACE and related studies, was unblinded and relied on self-reported outcomes. As has been argued repeatedly in the PACE debate, this study design creates a significant possibility of generating misleading results due to bias, even more so when a therapeutic relationship between clinician and patient is involved. Since it is difficult if not impossible to accurately measure the extent of such bias in any particular study with these characteristics, findings should be interpreted with caution. Investigators in the IBS field obviously confront a deficit of reliable biomarkers. But objective indicators such as a reduction in health care visits or fewer days off from work–vetted by official records–could support the argument that self-reported changes in symptoms are genuine and not an artifact of bias.
It is especially important to keep this factor in mind when the reported benefits are relatively small, such as those in this study. One of the two primary outcomes, for example, was the IBS Symptom Severity Scale,* a 500-point instrument based on patients’ answers to five questions about the intensity and frequency of their symptoms—abdominal pain, bloating, etc. Scores of less than 75 represent normal bowel function, with increasing numbers representing mild, moderate and severe symptoms. A change of 50 points or more on the scale is considered clinically significant. *[In this paragraph, I had mistakenly identified the instrument as the “Severe Severity Scale” instead of the “Symptom Severity Scale.”]
Of the 70% of participants who supplied data at 12 months and the 58% who did so at 24 months, more individuals in the CBT arms than in the TAU arm had reductions in their scores of at least 50 points. But their experience tells us nothing about the final results of the significant minority that did not provide outcome data. Perhaps participants who found the whole exercise to be a waste of time, or worse, were less willing to provide follow-up data, either at the 12-month or 24-month point.
With an “intention-to-treat” analysis designed to account for missing data, the investigators compared the means of the severity scale results in the three arms. At 12 months, the mean score was lower for both CBT interventions than for TAU—61.6 points lower for telephone-delivered CBT and 35.2 points for web-based CBT. The first number is a bit over the 50-point threshold for what is considered a clinically significant change, the second a bit under. At 24 months, the respective mean scores in the two groups were 40.5 and 12.9 lower than the TAU score–less than 50 points in both cases. (In fact, the 24-month finding for web-based CBT was not statistically significant.) In other words, at both 12 months and 24 months, the reported benefits of CBT over TAU on this scale could be considered modest at best.
These reported improvements would seem to fall within the range that might be expected from the potential bias inherent in this study’s design. When some trial participants know they are getting something extra and other trial participants know they are not getting something extra, and both groups know the something extra is supposed to improve their health, it isn’t surprising that afterwards those who got something extra could be more likely to report improved health than those who got nothing extra.
But perhaps anything extra would do the trick, even more so if it involved contact with a therapist and/or web-based engagement and entertainment. Perhaps an intervention of streaming Harry Potter movies and discussing quidditch strategy with a phone therapist who alternated between the voices of Hermione and Hagrid would have a similarly beneficial effect as CBT– especially when compared to getting nothing. We just don’t know.
In this case, the researchers did not seek to craft a control group whose experience would simulate the intensity or frequency of the CBT interventions. As awful as PACE was, at least the adaptive pacing therapy arm paralleled the CBT/GET arms in terms of possible therapist contact and interaction, if not in the content. The IBS study contained no comparable control group—just people receiving regular care, which had already proven ineffective. These participants were offered the opportunity to access the web-based program after the first 12 months, although without any clinical support. This arm was thus randomized to what was essentially a “waiting list” condition—a frequent phenomenon in psychological studies.
As often happens in CBT studies, the intervention was adapted to the circumstances, making it harder to determine what aspect, if any, produced an effect. In this case, the CBT included advice on “developing stable healthy eating patterns,” per the Gut paper. In an illness like IBS, it is conceivable that sustained guidance on food and diet alone could lead to improved digestion and an easing of symptoms, but the study did not feature a comparison arm focused solely on that sort of intervention. Perhaps the reported improvements, if real, had more to do with suggestions about “healthy eating patterns” than with other elements of the CBT.
In presenting the findings, the Gut article noted that “psychological therapies and particularly face-to-face cognitive–behavioural therapy (CBT) can reduce symptom scores and improve quality of life by targeting unhelpful beliefs and coping behaviours.” In the context of this research, “reduce symptom scores” sounds about right, given that reducing symptom scores in a study prone to bias is not necessarily synonymous with actually reducing symptoms. In any event, a Cochrane review of psychological therapies for IBS, one of the cited references for the statement about symptom-reduction and improved quality-of-life from psychological therapies, questioned whether the reported benefits were meaningful.
The Cochrane review concluded the following: ”Psychological interventions may be slightly superior to usual care or waiting list control conditions at the end of treatment although the clinical significance of this is debatable. Except for a single study, these therapies are not superior to placebo and the sustainability of their effect is questionable.”
As with research into the CBT/GET treatment paradigm, leading members of the UK academic and medical establishments have produced an IBS study that has generated data of limited value while undoubtedly costing a lot of money. Major public health policy should ideally not be based on findings from open-label trials relying solely on self-reported outcomes, even more so when therapeutic relationships are involved and when there is no adequate control group. In this case, the investigators have documented the following: Compared to offering IBS patients nothing, offering them something leads to a few changes in how they answer questionnaires. Perhaps that represents some genuine improvement; perhaps not. It is hard to tell from the study.