By David Tuller, DrPH
*This is a crowdfunding month at University of California, Berkeley. If you appreciate my work and would like to make a donation (tax-deductible to US taxpayers) to the university in support of my position, here’s the link: https://crowdfund.berkeley.edu/project/46120
It must be tough for investigators when a major study seeking to assess the effectiveness of an intervention for a challenging condition yields null results. That’s what happened in 2019 with a trial of rituximab for ME/CFS, published in Annals of Internal Medicine. Findings from earlier research had suggested that rituximab, a drug used to treat autoimmune diseases, might have an impact on ME/CFS. However, the trial results did not support the hypothesis, forcing the investigators to revisit their notions about the mechanisms driving the disease.
Last year, a large trial for functional motor disorder (FMD), a subcategory of functional neurological disorder (FND), reported similarly disappointing news. In a paper in The Lancet Neurology, published in July, the investigators of the trial, nicknamed Physio4FMD, reported null results for specialized physiotherapy on the primary outcome–self-reported physical function at 12 months. While those who received the Physio4FMD intervention had slightly better scores on this measure than those who received treatment as usual (TAU), the results were neither statistically nor clinically significant.
(The lead author posted a thread about the findings here.)
In such cases, investigators are often in the somewhat thankless position of having to publish further analyses, trying to find some silver linings even though their intervention has already failed its most important test. Since last month, the Physio4FMD team has published two additional papers: a look at factors predicting outcomes, and a cost-effectiveness analysis. (I might get around to looking at those in a subsequent post.)
FND, formerly called conversion disorder, is the current term for a category of neurological symptoms that do not fall within established disease categories. The sub-group of functional motor disorder includes arm or leg weakness and paralysis, gait disorders, and the like. These conditions, whatever their cause, can be chronic, seriously disabling, and resistant to treatment. In the past, they were generally viewed as psychiatric conditions. In recent years, FND experts have categorized them as “brain network” disorders. In reality, their etiology remains unknown.
This is the second time in recent years that a high-profile FND treatment trial produced null results for its primary outcome. In 2020, the CODES trial for psychogenic non-epileptic or “dissociative” seizures, another subcategory of FND, reported that cognitive behavior therapy was no more effective than standard care in leading to seizure reduction at 12 months. In that case, FND experts argued after-the-fact that seizure reduction was the wrong primary outcome and that “quality-of-life” measures were more important.
Just as CODES was the largest trial of dissociative seizures, this new FND study–“Specialist physiotherapy for functional motor disorder in England and Scotland (Physio4FMD): a pragmatic, multicentre, phase 3 randomised controlled trial”–represented a first for the field. Noted the paper: “To the best of our knowledge, Physio4FMD is the first fully powered randomised controlled trial of a physical therapy-based intervention for functional motor disorder and is the largest randomised study of people with functional motor disorder published to date.”
The trial’s primary analysis included 241 participants from 11 hospitals in Scotland and England, with 138 assigned to the Physio4FMD intervention and 103 to TAU. The latter consisted of whatever treatment the participants received, or didn’t receive, after referral to the local National Health Service (NHS) neurological physiotherapy service. The intervention included nine sessions over three weeks, with a final session three months later. (Recruitment began in 2018 but was interrupted by the COVID-19 pandemic. The paper goes to substantial lengths to explain how the team addressed these challenging circumstances, including in the statistical analyses.)
As described in the paper, the Physio4FMD intervention sought to focus on the factors presumed to be driving the symptoms, such as paying excessive attention to symptoms, and had three broad goals: “to help patients understand their symptoms; to retrain movement with redirection of attention away from focusing on their body; and to develop self-management skills.” The approach had undergone extensive development in the years before the trial. As noted, “the protocol builds on expert consensus recommendations for physiotherapy for functional motor disorder and was tested with promising outcomes in a prospective cohort study and a randomised feasibility study.”
The trial was unblinded and relied on subjective outcomes—a study design that generates an unknown amount of bias, for any number of reasons. In such cases, modestly positive findings are as likely to reflect the bias inherent in the design as any genuine impact of the intervention.
**********
Poor results on the primary outcome
In any event, the intervention did not produce the expected results. The primary outcome, the SF-36, is a frequently used measure for self-reported physical function. As described in the paper, it “includes ten questions for participants to self-rate their degree of limitation when attempting vigorous activities (eg, running or lifting heavy objects), moderate activities (eg, moving a table or pushing a vacuum cleaner), carrying groceries, climbing stairs, walking various distances, washing, and dressing.”
Scores on the SF-36 range from 0 to 100. Higher scores represent better physical function. A score of 65 or below, for example, was considered disabled enough to be able to enter the PACE trial, which purported to prove that psycho-behavioral interventions could cure ME/CFS. In the Physio4FMD trial, the average scores at baseline were 26 and 31, respectively, for the intervention and TAU groups. That is very, very disabled. At 12 months, both groups averaged just over 37—still very, very disabled.
Besides not being statistically significant, the mean difference between the two groups at 12 months on the SF-36 was also, at 3.5 points, below the threshold considered clinically significant. On the SF-36, the threshold for a difference to be considered clinically significant is 10 points.
Among the many secondary outcomes, the Physio4FMD intervention arm scored better than the TAU arm on an overall rating of symptom improvement and on treatment satisfaction. But many other secondary measures had null results. As the investigators noted in the limitations section, given the number of secondary outcomes, some might have been found to be statistically significant by chance, and the analysis did not include the extra tests designed to minimize this possibility.
The rating of symptom improvement, called the clinical global impression of improvement scale (CGI-I), is much briefer than the SF-36. In the CGI-I, as the study explained, “participants rate their perception of improvement in answer to the question, ‘After physiotherapy, the problem with my movement is…’ with the responses either ‘much improved,’ ‘improved,’ ‘no change,’ ‘worse,’ or ‘much worse.’” The answers thus provide no indication of the respondent’s level of disability in relation to others—just in relation to their own prior subjective state.
Like the CGI-I, the SF-36 is self-reported and therefore subject to biases related to that status. Unlike the CGI-I, it covers a range of specific activities and requires the respondent to consider each one separately. With its 100-point scoring, the SF-36 allows for easy comparison of results with other populations. In the Physio4FMD study, no matter what participants reported on the CGI-I, they remained severely disabled overall, according to the primary outcome.
In summing up, the investigators concluded that,“taken together, the subjective improvements in symptom ratings along with the very high levels of satisfaction with treatment, suggest that specialist physiotherapy could be a valued and safe treatment option for some people with functional motor disorder.”
Suggesting that a treatment “could be” an option for “some” patients is not saying much. As for the reports on symptom improvement and treatment satisfaction, it shouldn’t be surprising that patients who receive care from compassionate clinicians are more likely to answer questionnaires positively than patients who don’t receive the same level of care. These responses should not therefore be interpreted to mean the intervention is effective–especially given the poor results for the more comprehensive and thorough assessment provided by the SF-36, the primary outcome.
The bottom line, per the SF-36 data, is that the patients in this trial remained extremely debilitated, whether they received the Physio4FMD intervention or physiotherapy at a local NHS service. That’s the take-home message here.