By David Tuller, DrPH
In February, I wrote a post tracking how a core finding from Bermingham et al, a 2010 study, has been misrepresented repeatedly in claims about the costs to the National Health Service of so-called “medically unexplained symptoms.” The misrepresented finding has been cited by proponents of an NHS effort to divert people labeled as having MUS away from specialist medical care and toward psychological interventions. This approach to MUS is part of the expansion of an NHS program called Improving Access to Psychological Therapies.
Putting aside for a moment legitimate concerns about MUS as a clinical construct, it is troubling that untrue information about NHS costs has been widely disseminated and cited in the context of public health policy debates and decision-making on the issue. That’s why I have started pressing some of those involved to correct these mis-citations of Bermingham et al. This task is more challenging than it should be–even when the mistake is indisputable, as it is in these instances.
In Bermingham et al, distinctions between the concepts of MUS and “somatisation” have essentially been erased. The first paragraph of the abstract addresses the prevalence and burden of “medically unexplained symptoms.” By the abstract’s second paragraph, the focus has switched, without explanation or comment, to the economic costs of what it refers to as somatisation. The paper notes that the unexplained symptoms “are thought to represent a somatic manifestation of psychological distress, or abnormal activation of sensory pathways.” This quick elision between MUS and somatisation is unwarranted.
One of the paper’s key findings is that the costs incurred in treating those of working-age in England identified as having MUS—that is, those found to have been somatising–accounted for 10% of what the NHS spent on that demographic group. Yet a series of MUS thought leaders have cited Bermingham et al while declaring something different: that these costs accounted for 10 % of the total NHS budget (or 11%, in one case). The effect of this error is to more than triple the apparent economic impact of MUS reported by Bermingham et al.
In other words, these investigators have misinterpreted a seminal study in their field of expertise while citing inflated numbers that help support their argument. This surprising collective lapse in scientific literacy should raise questions about the credibility of related pronouncements on public health policy initiatives, such as the expansion of IAPT. (It should be noted that the figure from Bermingham et al has been referenced accurately in other instances, including by some who have also mis-cited it.)
In seeking corrections of misstatements of Bermingham et al’s results, I am not endorsing the study’s methodology or calculations. While Bermingham et al has been an influential piece of research, the data have serious limitations that have often been overlooked. Nor am I endorsing the study’s apparent premise that all cases of MUS involve somatisation.
From the perspective of Bermingham et al, MUS is a useful, valid and discrete clinical category. The corollary is that conditions and symptoms currently unexplained by medical science are not just poorly understood at present but are inherently unexplainable in pathophysiological terms and therefore represent the process of somatisation–or at least can appropriately be treated that way for the sake of analysis.
Strict adherence to this approach appears to violate common sense. Throughout history, sickness and death have been attributed to a wide range of unverifiable external and internal phenomena–witchcraft, demonic possession, miasmas, wandering wombs, planetary movements, hypothesized psychological processes (penis envy, etc). Many or most must have sounded reasonable at the time. Yet evidence of organic causation of formerly unexplained or mysterious illnesses has repeatedly emerged to debunk previous theories and certitudes.
Members of our species can obviously suffer from a range of discomforts and pains that elude the current reach of medical care and practice. It is unclear how these lacunae in knowledge across multiple scientific domains can be convincingly invoked as evidence for a distinct and stable object of study called MUS. Do those who hold this view believe we have reached the end of advances and discoveries in the diagnosis and treatment of complex diseases?
For argument’s sake, let’s stipulate that some people might be somatising their “psychological distress,” as Bermingham et al posits. How can it be determined who is somatising and who is suffering from something for which an underlying organic cause might yet be found? And if unexplained suffering is routinely interpreted as somatising based on the absence of a currently proven underlying organic cause, how does that interpretation itself impact the likelihood that possible underlying organic causes will be researched, much less discovered?
Every human sensation, it seems, could be identified as somatisation or an example of MUS in the absence of a currently proven underlying organic cause. Bermingham et al lists dozens of conditions across 19 medical specialties in which, it is claimed, “somatisation plays a part.” The possibility that some of these conditions might in future be fully explained through identifying an underlying organic cause does not seem to have been considered. It is not self-evident that lumping them all together as if they could be assessed and managed as part of a unified category makes sense.
In an interesting finding, most of the costs related to somatisation in Bermingham et al were not attributed to those identified as having somatisation disorder. In the study, a significant majority of those who were somatising were found to have “subthreshold somatisation” instead. Here’s how the paper describes it: “The term ‘subthreshold somatisation’ is used to refer to any somatoform disorder in which symptoms are not sufficiently numerous or diverse to qualify for a diagnosis of somatisation disorder…‘Somatisation disorder’ is used to describe patients who meet full criteria for the disorder.”
Ok, then. Is that clear? Most of those represented as somatisers and included in this study about the costs of somatisation did not actually have the associated disorder. Instead, they were experiencing “subthreshold” levels of somatisation. So whatever somatisation disorder is, most of the information analyzed in Bermingham et al is about something else. Got it.
Under the circumstances, the reported findings in Bermingham et al could be framed another way. According to the data, the cost of treating people of working age identified as having somatisation disorder accounted for less than one billion pounds–between 3-4 % of what the NHS spent on that demographic group. When phrased like that, things sound a bit different.
Why does any of this matter? Under current and developing NHS practice, people who present with complaints of a symptom or symptoms that some general practitioner or other clinician cannot explain in medical or organic terms could be identified as having MUS. Even if, per Bermingham et al, they are experiencing only “subthreshold” levels of somatisation, they could be tagged as prime candidates for psychotherapy under the IAPT program and discouraged from accessing specialist care.
In the UK, GPs are being educated about MUS and encouraged to spot presumed somatisers and curtail extensive searches for medical explanations of their complaints. This approach has been framed by proponents as both a win for public health and an efficient way to manage medical resources. Given the suspect quality of the research I have seen in this domain so far, I have reservations about MUS and somatisation and the various claims being made.
In any event, whatever its merits, Bermingham et al has been routinely cited in research involving MUS costs. And if the findings are to be cited, they should at least be cited accurately. That means if they have not been cited accurately, they must be corrected.
In that spirit, below is my recent letter to Professor Helen Payne at the University of Hertfordshire’s School of Education–with a short follow-up for clarification.
Dear Professor Payne—
On July 1st, I cc-d you on a letter I sent to Professor Carolyn Chew-Graham of Keele University. That letter involved Professor Chew-Graham’s repeated mis-citation of a 2010 study, Bermingham et al, about the costs to the National Health Service of so-called “medically unexplained symptoms.” In the letter, I indicated that I planned to contact other investigators who have made the same mistake.
Bermingham et al reported that the costs of treating working-age people in England diagnosed with MUS accounted for 10% of NHS spending for that age group. Yet in many instances, investigators have cited the study in declaring something different–that these costs accounted for 10% of all NHS spending. This inflated estimate has been referenced in discussions of public health policy, such as the expansion of the NHS program called Improving Access to Psychological Therapies.
By way of introduction, I am a senior fellow in public health and journalism at the Center for Global Public Health at the University of California, Berkeley. For the last few years, I have examined and critiqued some of the research in the ME/CFS field, including the PACE trial. I have chronicled my findings in more than 150 posts and articles on Virology Blog, a widely read science site hosted by Professor Vincent Racaniello, a microbiologist at Columbia University.
More recently, I have been reviewing some of the literature on MUS as well as on the expansion of the IAPT program to those diagnosed with ME/CFS, irritable bowel syndrome, and other chronic and complex illnesses. As with the studies on ME/CFS, investigators seem to be making some unwarranted claims. The frequent mis-citation of Bermingham et al is a key example.
I first alerted Professor Chew-Graham to this salient error in a letter I sent her in January. Since I did not receive a response, I appealed directly to the British Journal of General Practice, seeking a correction in a 2017 editorial for which she was the lead author. That correction has finally been posted, accompanied by an appropriately prominent note. In this month’s letter to Professor Chew-Graham, I urged her to follow up by also correcting an article about MUS that she wrote the same year for a Keele University blog.
A similar mis-citation of Bermingham et al has appeared as well in two of your own publications. A 2017 paper in Health Research Policy and Systems included the following: “These conditions cost £18 billion in 2008 according to Bermingham et al, and £3.1 billion of this to the NHS, which is approximately 10% of the NHS budget.” This statement is not an accurate description of Birmingham et al and needs to be corrected.
In a paper the following year in Frontiers in Psychology, you and a co-author wrote: “At approximately £11.64 billion it cost around 10% of the whole English National Health Service (NHS) budget in 2015/16.” Although the previous sentence cites Bermingham et al, no reference is specifically provided for the estimate of £11.64 billion in costs. Rather, this figure appears to have been back-calculated by taking 10% of that year’s total NHS budget, not the amount spent on the working-age population. In any event, the statement needs to be corrected.
Thank you for your quick attention to this matter. The development of sound public health policy depends on the integrity and robustness of the published literature. That is why this indisputably wrong but potentially influential citation of Bermingham et al needs to be corrected promptly in the various venues in which it has appeared.
Given the significance of the issue, I am cc-ing several people on this e-mail. They are: editors at Health Research Policy and Systems and Frontiers in Psychology, the two journals in question; Professor Chew-Graham and others who have made the same mistake; Sarah Bermingham, the lead author of Bermingham et al; four physicians involved in the effort by the National Institute for Health and Care Excellence to create new ME/CFS guidelines, as well as two patient representatives on the NICE committee; Carol Monaghan MP, who has expressed concern about the poor quality of ME/CFS research; and Professor Racaniello, who hosts Virology Blog. (I also plan to post the letter.)
David Tuller, DrPH
Senior Fellow in Public Health and Journalism
Center for Global Public Health
School of Public Health
University of California, Berkeley
As soon as I posted the letter on Facebook, a reader noted a point of confusion about Professor Chew-Graham’s misstatements. I re-read what I’d written about that. It wasn’t explicitly wrong. But it wasn’t exactly right, either. Since I was trying to hold others to account, I decided I needed to clarify the issue for Professor Payne and the other recipients. Here’s the follow-up I sent:
Dear Professor Payne–
I would like to clarify one point in the letter I sent earlier. The phrasing suggested that Professor Chew-Graham had twice misrepresented Bermingham et al to mean that costs for MUS among the working-age population accounted for 10 % of total NHS spending. This is not the case. In fact, her editorial in the British Journal of General Practice, since corrected, initially stated that these costs accounted for 11% of total NHS spending, not 10% of total NHS spending.
I should have made that point explicitly clear. I apologize for not having done so.