By David Tuller, DrPH
UPDATE: Physical therapist Zachary Grin, whose criticism prompted this follow-up post, has posted a response, so I have highlighted it right below this introduction. He continues to believe I misrepresented the study. I continue to disagree. Prognostic factors, as he notes, are baseline characteristics. (In this case, “baseline” means mid-illness, so these are not presented here as pre-existing conditions.) Despite the statement that the study focuses on the “prognostic” value of these factors, the investigators throughout–and not just in the title and abstract–refer to the “impact,” the “effects” and the “influence” of anxiety and depression on “outcomes.” To me this usage is indisputably causal and can only be read as referring to whether anxiety and depression directly make things worse—that is, whether they perpetuate the illness. If that is not what the authors meant, as Dr Grin maintains, they have used language in a very sloppy and inaccurate manner and should clean it up. To Dr Grin, this language is consistent with suggesting just an association between a prognostic factor and the outcome and should be interpreted solely in that manner—which I think contradicts the plain meaning of the words.
But readers can certainly judge for themselves. In any event, I thank Dr Grin for his interest in engaging in this discussion.
Dr Grin’s Comment:
I am pleased that you found my criticisms on your interpretation interesting enough to warrant a follow-up post and I welcome the opportunity to further elaborate. The study explicitly set out to identify a potential “prognostic effect” of depression, anxiety, and personality disorders in patients with functional limb weakness/paresis (FND-par). The study did not aim to support whether these are perpetuating factors in FND-par.
A prognostic factor is a baseline characteristic of the patient that informs us about the likely overall *outcome* of the patient prior to treatment. A perpetuating factor contributes to the ongoing *presence or exacerbation of symptoms*. Identifying perpetuating factors can help select appropriate treatment targets but they do not provide the information needed to inform the patient’s overall prognosis.
It’s important to look beyond titles and summary conclusions when interpreting scientific research. The phrases “impact on the outcome” and “influenced outcomes” in the title and abstract conclusion, respectively, are both consistent with prognostic effects as prognosis directly informs expected outcome and perpetuating factors do not. The discussion section of the paper also explicitly states “no clear finding to suggest that the baseline presence of either depression or anxiety in absolute terms was correlated with clinical outcome”. Again, this is entirely consistent with their prognostic effect.
You stated that ”if the authors intended to make a subtle but clear distinction between the constructs of prognostic effect and perpetuation, they failed”. I would argue that the responsibility of clear interpretation also rests with those discussing the research, especially when their platform has a broad reach and an impact on public perception. While research findings can be subject to various interpretations, your interpretation appears to not acknowledge the basic differences between these two constructs and their use in clinical practice—the very reason for conducting such studies. It is your responsibility to understand these differences before disseminating your interpretation.
I stand by my initial criticism of your interpretation: your post misrepresented the study by conflating prognostic factors with perpetuating factors.
Zachary Grin, PT, DPT
**********
Zachary Grin, a physical therapist in New York City who specializes in treating patients with functional neurological disorder (FND), commented on my most recent blog after I posted a link to it on X, the social media platform formerly known as Twitter. As he noted in his series of tweets (or his series of Xs?), he found my post to be “very misleading.” Per his interpretation, the study I discussed was not investigating whether depression and anxiety were perpetuating factors in functional limb weakness, or “FND-par,” but whether they had prognostic value. (The study also examined personality disorders, but the authors ultimately concluded that there was insufficient data on them to draw conclusions.)
Dr Grin quoted this passage from the abstract to support his point: “The impact of these conditions on the prognosis of FND-par has not been systematically reviewed. The aim of this study was to identify a potential prognostic effect of comorbid depression, anxiety, and/or personality disorder on prognosis in patients with FND-par.” (The use of both “prognostic effect” and “prognosis” in the same sentence is clunky and seems redundant; perhaps the study, like so many academic papers, could have benefited from a rigorous copy-editing process.)
As Dr Grin further noted: “Prognosis and perpetuating factors are not the same thing and this study does not show anxiety and depression aren’t perpetuating factors. You are misrepresenting this study.”
I have written many posts about FND, and Dr Grin has often challenged them. As in this case, he has done so politely, which I greatly appreciate because it differs from how many people—including sometimes me–approach such discussions on X. Although we rarely concur on these matters, I have generally enjoyed exchanging views with him. In this case, I thought Dr Grin made an interesting point. And since I assume others might agree with him, it seemed worthwhile to write this follow-up to the post.
It is certainly true, as he noted, that some factors or indicators might have value in predicting outcomes—in other words, prognostic value—without being directly implicated in perpetuating the illness of concern. Perhaps the factors, whatever they might be, are associated with or accompany the mechanism that is actually perpetuating an illness without being directly implicated in that perpetuation.
The flaw in Dr Grin’s argument, as I see it, is that the description he cited of the study’s goals was not the only way in which the authors themselves framed their research. The headline itself, for example, did not mention “prognosis” or “prognostic effect.” It stated explicitly that the study is about the “impact” of anxiety, depression, and personality disorders “on the outcome of patients with functional limb weakness.” The abstract’s conclusion indicated that the study “found no evidence that depression or anxiety influenced outcomes”—again without mentioning prognosis or prognostic effect. Notwithstanding Dr Grin’s concerns, these phrases indicate in plain English—at least to me–that the study examined whether anxiety and depression were perpetuating factors.
In other words, if the authors intended to make a subtle but clear distinction between the constructs of prognostic effect and perpetuation, they failed. Perhaps they would agree with Dr Grin’s reading of what they wrote; if so, they should have done a better job of explaining themselves. While Dr Grin suggested that I was conflating these two constructs in a problematic manner, my response is that the authors themselves were the ones responsible for any such conflation. I was only quoting their words.
In his series of tweets, Dr Grin included the following: “I hope others actually read the paper rather than take your word for it.” I definitely agree! I think, if others actually read the study, they will find that it does indeed appear to state unequivocally that, according to the available data, anxiety and depression were not shown to be perpetuating factors—not just that they have no prognostic value. If that was not the intended meaning, I would advise the authors to submit a clarification or corrigendum. From my perspective, Dr Grin’s interpretation does not hold water.
If only I could read the whole paper to know what’s going on but I’m grateful to David for explaining it to me. I’d be interested to read the alternative explanation too but I don’t have access to that either. Quel dommage.
As is often the case with FND research, I wonder if this research was a patient research priority and if they were involved in the design of the research.
Even on Genetics 101 fundamentals some health officials and journalists do not want to entertain appropriate change (Good Nomenclature: A Matter of Life and Death – 6).
I am pleased that you found my criticisms on your interpretation interesting enough to warrant a follow-up post and I welcome the opportunity to further elaborate. The study explicitly set out to identify a potential “prognostic effect” of depression, anxiety, and personality disorders in patients with functional limb weakness/paresis (FND-par). The study did not aim to support whether these are perpetuating factors in FND-par.
A prognostic factor is a baseline characteristic of the patient that informs us about the likely overall *outcome* of the patient prior to treatment. A perpetuating factor contributes to the ongoing *presence or exacerbation of symptoms*. Identifying perpetuating factors can help select appropriate treatment targets but they do not provide the information needed to inform the patient’s overall prognosis.
It’s important to look beyond titles and summary conclusions when interpreting scientific research. The phrases “impact on the outcome” and “influenced outcomes” in the title and abstract conclusion, respectively, are both consistent with prognostic effects as prognosis directly informs expected outcome and perpetuating factors do not. The discussion section of the paper also explicitly states “no clear finding to suggest that the baseline presence of either depression or anxiety in absolute terms was correlated with clinical outcome”. Again, this is entirely consistent with their prognostic effect.
You stated that ”if the authors intended to make a subtle but clear distinction between the constructs of prognostic effect and perpetuation, they failed”. I would argue that the responsibility of clear interpretation also rests with those discussing the research, especially when their platform has a broad reach and an impact on public perception. While research findings can be subject to various interpretations, your interpretation appears to not acknowledge the basic differences between these two constructs and their use in clinical practice—the very reason for conducting such studies. It is your responsibility to understand these differences before disseminating your interpretation.
I stand by my initial criticism of your interpretation: your post misrepresented the study by conflating prognostic factors with perpetuating factors.
Zachary Grin, PT, DPT
I’ve no idea who’s right or wrong in this case but I’m certain of one thing – conflating FND with MUS is way worse than any conflation that may or may not have happened here.
I’ve been trying to work out who is right and who is wrong here. Where I’ve got to is this: From what I can see, I think the review probably only looked at baseline data for anxiety and depression, although that’s not 100% clear to me. That would point to it being about prognosis, not about perpetuating factors, since, for all we know, patients’ anxiety and depression could have miraculously disappeared during the follow-up period. (Who knows? Patients might have been so relieved to finally have nice, friendly doctors telling them that their symptoms were real and not in their heads that their anxiety and depression evaporated shortly after diagnosis.) I don’t think the authors have been careful enough in their use of language and the title and abstract conclusion don’t help with that. There’s nothing in the abstract conclusion to indicate that the depression and anxiety being referred to are the patient’s pre-existing anxiety and depression. The use of the word ‘concurrent’ rather than ‘pre-existing’ or ‘initial’ further compounds that, I think.
Furthermore, the main conclusion brings ‘treatment’ into it. Why would you bother to treat anxiety and depression if their presence at baseline has no impact on prognosis/outcomes? And why would you bother to do research into that treatment? As I noted above, for all we know, the patient’s anxiety and/or depression could have resolved once the patient felt confident that they were in good hands. Alternatively, these things could have got worse once the patient realized that, despite being in good hands, their symptoms weren’t improving. That might be good reason to treat, I suppose, but bringing up ‘treatment’ in the conclusion is odd and could confuse people given that nothing about it can be concluded from this study.
Thanks for the update, David, and for explaining it further to folks like me. I’m not sure what any of this means if the ‘baseline’ measures of depression and anxiety were taken mid-illness (I assume during treatment then?) rather than at the point of diagnosis (before any treatment had begun?). Also, given that the follow-up periods of the reviewed studies appear to have been of variable length, do these things combine to make the results interpretable at all?
ZG commented:
“It’s important to look beyond titles and summary conclusions when interpreting scientific research.”
I’d agree with that, but it’s also important to ensure that titles and conclusions accurately reflect the paper/study, don’t confuse the reader and aren’t open to misinterpretation. This is even more important in the area of medicine because readers – overworked doctors – are highly likely to be hard pressed for time and may rely on the title and abstract or may jump to the main conclusion to glean what the paper’s about. I think medical authors and journal editors therefore have a responsibility to ensure that the titles, abstracts and main conclusions are crystal clear and not open to misinterpretation.