By David Tuller, DrPH
For years, experts in functional neurological disorder (FND) have cited a seminal study in their field to claim that the diagnosis was the second-most-common presentation at outpatient neurology clinics, with a prevalence of 16%. This claim was, and is, categorically untrue. The Scottish Neurological Symptoms Study (SNSS), which yielded multiple papers about a dozen years ago, actually found that 209 out of 3781, or 5.5%, were identified as having conversion symptoms. (Conversion disorder is the old name for FND.) At that rate, it was the eighth-most-common presentation in the SNSS, not #2.
This discrepancy was pointed out to me a year or so ago by a valued source. I blogged and posted on social media about the issue. I also wrote—twice–to a major journal seeking a correction in one such paper. I received no response. Two weeks ago, I sent a letter seeking a correction to another major journal, NeuroImage: Clinical. Several colleagues—from Berkeley, Columbia, Johns Hopkins, University College London, etc—co-signed the letter.
The authors have now agreed to correct the false statement that the SNSS found FND to be the second-most-common presentation. The correction is inadequate in some respects. Nevertheless, I hope it means that they will at least no longer cite the SNSS to claim FND is the #2 presentation at outpatient neurology clinics, with a prevalence of 16%. I also hope the FND experts themselves will now initiate corrections in the dozens of papers that have included these untrue data points. It will be a tedious process for all involved if our group has to send out letters to journal editors about every single one of these papers.
I need to say I was impressed with the speed with which the journal handled our request. Within five days of sending our initial letter, I heard back from an editor, Anna Bankhoff, who was seeking our group’s input about proposed language for a correction. I attributed this quick action to the integrity of the journal; the authors themselves have previously exhibited little interest in responding to these concerns and cleaning up the mess. The authors presumably understood that a correction in one paper meant it would be hard to object to requests for corrections in every other paper with the same misinformation.
This was the sentence in question: ““This renewed interest [in FND] parallels recognition that FND is the 2nd most common outpatient neurologic diagnosis, with many patients remaining chronically symptomatic and incurring high healthcare costs.” The reference was to Stone et al, a 2010 paper arising from the SNSS.
The initial proposal from the authors was to remove the #2 claim and simply say that FND was a “common” presentation. We responded, as a group, that we found that solution inadequate. Here’s what we wrote:
Dear Dr Bonkhoff–
I have consulted with my colleagues, and first we would like to thank the journal and the authors for promptly acknowledging that Perez et al includes an error requiring a corrigendum. I think it’s fair to say that this sort of swift action does not happen frequently in academic publishing, so it is very much appreciated.
As regards the proposed wording, we believe it is definitely a step in the right direction but not in itself sufficient. The original sentence compared FND to other diagnoses by calling it the second-most-common presentation while inaccurately citing findings from the Scottish Neurological Symptoms Study (SNSS). To undo the misimpression generated by this claim, the replacement sentence should also place FND in the context of other diagnoses.
From our perspective, calling it “common” and leaving it at that is not enough. First, the understanding of “common” is dependent upon the circumstances. Readers of the article could easily assume “common” means 20% or 30% or 40% of patients–not 5.5%. Second, the proposed phrasing doesn’t indicate that FND was not, after all, the #2 diagnosis in the SNSS but down the list at #8—after headache, epilepsy, peripheral nerve disorders, miscellaneous neurological disorders, multiple sclerosis/demyelination, spinal disorders, and Parkinson’s disease/movement disorders. Therefore, referring to FND as “common” would not fully address the problem created by the original misstatement in the first place.
(As a side note, the phrase “outpatient neurologic diagnosis” is ambiguous and could easily be misinterpreted to mean that these diagnoses were made in primary care and not in secondary care settings–i.e. neurology clinics.)
Another consideration: Two of the co-authors of Perez et al, Professors Alan Carson and Jon Stone, are also co-authors of a new paper called “Functional neurological disorder is common in patients attending chronic pain clinics,” published on May 23rd in the European Journal of Neurology. Citing the SNSS findings, this article reported that “the prevalence of typical FND in patients attending neurology outpatient clinics is 5.4%.” (Since this “typical FND” group included 209 out of 3781 patients, as we noted in our original communication with you, it is unclear why the figure given is 5.4% and not 5.5%.)
Given this most recent definitive statement about the SNSS, it seems reasonable to suggest that Perez et al make a similarly clear account of the prevalence of “typical” FND. We recommend one of the following options to replace the original sentence:
“This renewed interest parallels recognition that FND is a relatively common diagnosis in neurology outpatient clinics, being found in 5.5% of referrals, with many patients remaining chronically symptomatic and incurring high healthcare costs.” (Or 5.4%, if our calculation is somehow wrong.)
“This renewed interest parallels recognition that FND is a relatively common diagnosis in neurology outpatient clinics, being found almost as often as multiple sclerosis and Parkinson’s, with many patients remaining chronically symptomatic and incurring high healthcare costs.”
Thank you again for the opportunity to comment on this proposed correction involving the erroneous citation in Perez et al. We look forward to the resolution of this matter.
Best–David
David Tuller, DrPH
Senior Fellow in Public Health and Journalism
Center for Global Public Health
School of Public Health
University of California, Berkeley
**********
The journal responds, and we respond again
That letter was sent last Monday. This morning, I received a response from the editor-in-chief, Dr Andrew Zalesky, that the authors had revised the correction and that it would be published as soon as possible. As in the previous proposal, the correction includes the statement that FND is “common” but then, in an additional paragraph, notes that ranking diagnoses in order is challenging because different studies yield different rates, depending on quality and methodology. The correction then cites some other research to bolster their claim that FND is “common.” (More on that additional research in a later post.) To me, the authors seem to be trying to absolve themselves of responsibility for making a bogus claim by suggesting that the entire effort to assess prevalence is fraught with issues. That might be true–but this response raises the obvious question of why these experts have spent the last decade making declarative assertions that FND is the #2 diagnosis.
Below is the response I sent to Dr Zalesky:
Dear Dr Zalesky–
Thank you for the response. We very much appreciate the speed with which you and Dr Bonkhoff have addressed this matter.
The decision by top investigators to repeatedly disseminate untrue information about a seminal study in their field of expertise has led to significant confusion about the actual prevalence rates of FND–not least, undoubtedly, among clinicians who end up seeing and treating some of these patients. We continue to believe that the word “common” on its own is liable to be misunderstood. We also continue to believe that the most appropriate method of correcting this error would be to provide the actual information from the study whose findings were misrepresented. Nonetheless, we appreciate that the authors have acknowledged their error and we accept that the process has now reached its conclusion.
As suggested by the persistent claim that the Scottish Neurological Symptoms Study (SNSS) found FND to be the #2 presentation at neurology clinics, leading experts at times seem to perceive FND even in the absence of the positive clinical signs that are now required for a diagnosis–or at least they are willing to include such patients in their reported prevalence rates. The phenomenon of interpreting (or misinterpreting) data to maximize apparent prevalence rates of illness has a name in public health and epidemiology. It is called “diagnosis creep.”
To put this correction in perspective, it is also worth noting another recent reference—an accurate one–to the prevalence of FND found in the SNSS. Professor Jon Stone, a co-author of Perez et al, is also the lead author of Stone et al, the 2010 study from the SNSS cited in Perez et al. In addition, Professor Stone is one of two co-authors of the section about “conversion disorder” on the medical education site UpToDate. (Conversion disorder is the former name for what is now called FND. It is unclear why UpToDate is still using this outdated term.) A note at the bottom of this section indicates it was updated in June of 2022.
In the epidemiology sub-section, the article states that “the point prevalence of conversion symptoms in clinical settings ranges from 2 to 6 percent.” The section cites three studies, including Stone et al. Here’s what it notes about the latter: “A prospective study of 3781 neurology outpatients found that conversion disorder was present in 6 percent.” This statement is obviously inconsistent with any claim that the study found FND to be the second-most-common presentation, which itself is based on the assertion that the prevalence was 16%.
It remains perplexing that experts in this field, including co-authors of the SNSS, have had no hesitation in tripling that study’s reported prevalence rate by lumping in many other participants who might have had “functional” disorders but did not meet the specific diagnostic criteria for FND—as if the criteria were optional or irrelevant. They clearly know that the SNSS did not in fact find FND to be the #2 diagnosis, as evidenced by the various correct citations of the data. Yet this false claim now peppers the medical literature.
We hope FND experts take this correction to heart and exercise more caution in citing the SNSS in future papers. The many co-authors of Perez et al, a virtual who’s who of leaders in the field, might also consider initiating similar corrections in every article of theirs that includes the same mis-citation.
Thank you again. Now that this issue is settled, I plan to blog about it. (I posted the original letter we sent but have refrained from further public comment while the question was being debated.) I will forward the link to any posts that I write.
Best–David
David Tuller, DrPH
Senior Fellow in Public Health and Journalism
Center for Global Public Health
School of Public Health
University of California, Berkeley
**********
Some further thoughts on this matter…
As noted in my response to Dr Zalesky, we still believe the use of “common” on its own is problematic—it could mean anything to anyone. The authors clearly do not want to be held to the 5.5% figure from the SNSS, even though that is the largest and most authoritative study on the issue. Instead, they prefer to leave the meaning of “common” vague, allowing readers to interpret that how they will. Presumably they believe that the higher rates are more accurate than 5.5%; I assume that’s why they have made this claim repeatedly. But even if that is their belief, it does not warrant the distortion of findings from the SNSS.
It takes an impressive amount of chutzbah to routinely mis-cite data in this way; it is hard to imagine that Berkeley graduate students in my department would get away with such misrepresentation in their dissertations without serious repercussions. It is also perplexing that the entire field of FND has accepted these mis-citations without apparent question. For top medical journals to have published dozens of papers in the last decade with these indisputably untrue claims represents a failure of the peer review process. I hope this situation triggers some soul-searching on the part of authors, peer-reviewers and journal editors about the proper way to cite study findings—although I won’t hold my breath on that score.