The Diagnosis of the Gaps: FND and the Problem of Phenotype Overlap

Quackademic

03 Oct 2026 0 comments

I want to start with something that could easily get lost in criticism of this paper: I have enormous respect for Todd Davenport and the other authors of Long COVID Is Not a Functional Neurologic Disorder. Published in the Journal of Personalized Medicine in 2024, the paper addresses an important problem, particularly the risk of people with poorly understood post-infectious disease receiving a functional diagnosis before the underlying illness is adequately understood.

The authors themselves acknowledge that the rate of FND misdiagnosis in Long COVID is unknown and that no carefully designed studies have systematically tested the hypothesis that Long COVID is a functional disorder. They also argue that patients can have FND alongside other diagnosable conditions and warn against treating an entire presentation as functional merely because functional features may be present.

There are nevertheless parts of the paper that deserve closer scrutiny. Some claims about what is common, rare or typical rely heavily on clinical experience rather than direct comparative evidence, and some of the distinctions drawn between FND, ME/CFS and Long COVID are less clear than they initially appear. But after pulling those distinctions apart, I think simply calling the disputed claims "misinformation" misses the more interesting problem. Some apparent disagreements are not really arguments about facts at all. They are arguments about definitions.

The answer changes when the definition changes

The clearest way to see the problem is to start with the authors' own comparison. Below is Table 1 from the paper, reproduced as published, with colour added only to make the Yes, No, Common and Sometimes entries easier to scan.

Table reproduced from Davenport TE, Blitshteyn S, Clague-Baker N, et al. Long COVID Is Not a Functional Neurologic Disorder. Journal of Personalized Medicine. 2024;14(8):799. View the original Table 1.

The structure of the table is revealing. The first section compares relatively broad clinical phenomena such as pain, dizziness, neuropathic features, recurrent flu-like symptoms, dysautonomia, fatigue and impaired sleep. The second section changes character. Instead of comparing presentations such as weakness, tremor, seizures or gait disturbance, the rows become functional leg weakness, functional seizures, functional tremor, functional dystonia and other explicitly functional phenomena.

That matters because the diagnostic interpretation has now been built into the feature being compared. If I ask whether functional tremor occurs in FND, I have already supplied much of the answer. The more useful comparison is whether tremor can occur in each condition, and then what evidence allows us to distinguish the mechanisms producing it. The same applies to weakness, gait disturbance, sensory disturbance, cognitive symptoms and many of the other phenomena in the table.

The table also moves between different conceptual levels. Pain, dizziness and fatigue are symptoms. PEM is a defined clinical phenomenon. Dysautonomia can refer to objectively demonstrable autonomic dysfunction. An abnormal sleep study is a test finding. Functional tremor and functional seizures are already diagnostically classified phenotypes. These are not equivalent kinds of things, yet the table places them in the same column and reduces them to Yes or No.

This is not merely semantics. Consider "neuropathic features". If that means objectively demonstrated neuropathy, then FND does not create neuropathy. If it means burning, tingling, numbness or altered sensation resembling neuropathy, functional sensory symptoms can look similar. The same problem occurs with dysautonomia. Objectively demonstrated autonomic dysfunction such as POTS or autonomic neuropathy is one question; whether a patient can present with symptoms that may be interpreted through a functional framework is another. A single Yes or No cannot answer both questions.

Dizziness exposes the problem particularly well

Dizziness is interesting because Table 1 gives FND a straightforward "No". Yet the paragraph immediately introducing the table says that people with "functional disorders" often exhibit chronic dizziness and chronic vertigo, alongside sensory disturbance, weakness, balance difficulty, pain, fatigue, sleep impairment, urinary and gastrointestinal symptoms and cognitive dysfunction. The next sentence switches specifically to "people with FND". At minimum, the terminology is muddy.

There is also a recognised functional dizziness phenotype. Persistent postural-perceptual dizziness, or PPPD, is classified by the Bárány Society as a chronic functional vestibular disorder characterised by persistent dizziness, unsteadiness and non-spinning vertigo. Whatever boundary one chooses between FND and the wider family of functional disorders, an uncomplicated claim that a functional presentation cannot look like dizziness is difficult to maintain.

PEM shows why definitions matter

Post-exertional malaise makes the definition problem even clearer. NICE describes ME/CFS-type PEM as worsening after activity that is often delayed, disproportionate to the activity and associated with prolonged recovery. Under that strict definition, saying another disease does not have ME/CFS-type PEM may be perfectly defensible. Change the question to whether another disease can produce post-exertional symptom worsening that resembles PEM, however, and the answer can change.

Multiple sclerosis provides a useful control. Exercise and increases in body temperature can temporarily worsen existing neurological symptoms through Uhthoff's phenomenon. That does not mean MS and ME/CFS share the same mechanism, nor does it turn Uhthoff's phenomenon into ME/CFS-type PEM. It simply demonstrates that "Does MS have PEM?" and "Can MS produce something that looks like post-exertional symptom exacerbation?" are not the same question.

The same distinction matters for FND. A functional presentation may worsen with activity, exertion, attention, fatigue or symptom provocation. At the level of the patient's description, that can enter the same initial territory as "I get much worse after doing things". That does not establish ME/CFS-type PEM. It demonstrates why the word PEM, PEM-like or simply post-exertional worsening can materially alter what is being compared.

Definition is not a minor detail in a comparison like this. Change the operational definition and you can change the truth value of the cell.

Unreferenced is not the same as false

Another distinction matters when criticising Davenport et al. A statement without an adequate reference is not automatically false. The paper makes several observations explicitly on the basis of clinical experience, including comments about the typical motor examination in Long COVID, the character of give-way weakness and the authors' impression that a significant number of patients with Long COVID are being misdiagnosed with FND. Those observations may ultimately prove correct, but clinical experience is not the same thing as quantified comparative evidence.

I would therefore separate criticism into three categories. There are unsupported or weakly supported claims, where the available evidence does not establish the strength of the statement being made. There are actual errors or internal inconsistencies, where a claim conflicts with evidence or another statement in the paper. Then there are definition-dependent claims, where the answer changes according to exactly what the authors mean by the feature being compared. The third category may be the most interesting because it reveals a problem extending beyond this particular paper.

What happens if we remove the diagnostic adjectives?

As a thought experiment, we can reconstruct the comparison around what a clinician or patient initially encounters: the presentation. I have also added MS as a useful control because it is an established neurological disease that overlaps with many of these symptoms. The question is deliberately different from Davenport's. Instead of asking whether another disease literally produces functional tremor, functional weakness or functional seizures, I am asking whether the presenting phenotype can occur in each condition or whether a sufficiently similar functional presentation exists that the initial complaint can plausibly enter functional diagnostic territory.

There is one important rule in this reconstruction: comorbidity does not count. A person with FND can also have POTS, neuropathy, epilepsy or another disease, but that does not make those conditions manifestations of FND. Likewise, a person with MS, ME/CFS or Long COVID can have another disorder. The question is whether the presenting phenotype itself has a functional analogue or interpretation sufficiently similar to create genuine diagnostic overlap.

There is also an important distinction between FND and the wider family of functional disorders. Sometimes the analogue sits directly inside FND, such as functional weakness, functional tremor, functional seizures or functional sensory symptoms. In other cases it may sit in neighbouring functional territory, such as functional dizziness, functional gastrointestinal presentations or functional urological presentations. These categories are not interchangeable and I am not claiming that they share one pathology or mechanism. The point of the exercise is to ask how much presenting clinical territory already has some functional interpretation available.

For that reason, Yes* does not mean "FND causes this disease process". It means that a sufficiently similar presentation exists within functional diagnostic territory that the initial phenotype can plausibly be interpreted as functional. That distinction is central to the exercise rather than a limitation of it.

Yes* means that a sufficiently similar presentation exists within functional diagnostic territory to create phenotypic overlap. It does not mean that the same pathology or physiological mechanism is present, nor that every functional analogue listed here is a subtype of FND. Can occur means the feature is recognised or reported but is not sufficiently characteristic for an unqualified Yes. Not characteristic does not mean impossible. Unclear means the available evidence or definition does not justify forcing the presentation into either Yes or No. Comorbidity is not counted. This reconstructed table is a conceptual exercise, not a diagnostic tool.

Why is the functional column so full?

The concentration of Yes* entries is deliberate. It does not mean that FND causes neuropathy, autonomic neuropathy, epilepsy, multiple sclerosis or the pathology underlying Long COVID or ME/CFS. Nor does it mean that every functional disorder is FND. It means that functional medicine collectively contains explanations for a surprisingly large amount of the phenomenology through which neurological and systemic disease can initially present.

Some examples are straightforward because the analogue sits directly inside FND. Weakness has functional weakness. Tremor has functional tremor. Abnormal posturing can resemble functional dystonia. Jerks can be functional. Gait disturbance has functional gait disorder. Numbness and altered sensation have functional sensory presentations. Seizure-like episodes have functional seizures. Cognitive, speech, swallowing and visual complaints also have recognised functional presentations. In these cases the overlap is not hypothetical: the functional category is explicitly named around the phenotype.

Dizziness provides another route into functional territory. FND does not create vestibular damage merely because someone feels dizzy, but PPPD is explicitly classified as a chronic functional vestibular disorder characterised by dizziness, unsteadiness and non-spinning vertigo. A patient presenting with persistent dizziness therefore already occupies territory for which a functional diagnosis exists, even though another patient with superficially similar dizziness may have a completely different mechanism.

Autonomic-type symptoms show why the distinction between disease and presentation matters. POTS, autonomic neuropathy and other objectively demonstrated autonomic disorders are not FND. But patients do not necessarily arrive with a completed autonomic diagnosis. They may arrive with dizziness, light-headedness, palpitations, shaking, weakness, altered breathing, sweating, presyncope-like sensations and exercise intolerance. Those complaints overlap with symptoms that can receive functional explanations. The Yes* therefore does not mean "FND causes POTS". It means the presenting territory can overlap before the mechanism has been established.

The same applies to neuropathic symptoms. FND does not cause small-fibre neuropathy, large-fibre neuropathy or nerve damage. But a patient does not necessarily arrive saying "I have biopsy-proven small-fibre neuropathy". They may arrive with burning, tingling, numbness, pain or altered sensation. Functional sensory symptoms can occupy the same descriptive territory. Calling the row "neuropathy" would justify a No. Calling it "neuropathic-like sensations" can justify a Yes*. The definition changes the answer.

Post-exertional worsening is deliberately included for the same reason. Formally defined ME/CFS-type PEM is more specific than simply becoming worse with activity. It can involve delayed worsening, disproportionate symptom exacerbation and prolonged recovery. FND does not therefore acquire ME/CFS-type PEM merely because symptoms worsen after exertion. But patients and clinicians encounter the presentation before its mechanism and classification are necessarily settled. Weakness, fatigue, dizziness, movement symptoms or other functional complaints can worsen during or following activity and therefore resemble the broader complaint of post-exertional worsening. The Yes* is not saying these phenomena are equivalent. It is showing how changing from strict PEM to PEM-like presentation changes the apparent boundary.

Bladder and gastrointestinal symptoms make the taxonomy even more obvious. FND, functional urological disorders and disorders of gut-brain interaction are not interchangeable diagnoses. Functional urinary retention is not functional tremor, and functional dyspepsia is not FND. But that is precisely the point. The question here is not whether everything functional belongs inside FND. The question is whether a functional interpretation is already available for the presenting complaint. In both urology and gastroenterology, the answer can be yes.

This is also why comorbidity cannot be used to manufacture overlap. A person with FND who also has POTS does not establish that POTS is functional. A person with FND and epilepsy does not make epileptic seizures functional. A patient with FND and neuropathy does not make neuropathy functional. The relevant observation is narrower: before the underlying process has necessarily been established, much of the presenting phenomenology already has a functional analogue.

The striking feature of this table is not that FND somehow causes everything. It is that functional explanations are available for an extraordinary amount of presenting phenomenology.

That distinction makes the table more interesting, not less. If weakness can be functional weakness, tremor can be functional tremor, seizure-like episodes can be functional seizures, sensory disturbance can be functional sensory symptoms, cognitive complaints can be functional cognitive symptoms, persistent dizziness can be framed functionally and bladder or gastrointestinal complaints can enter neighbouring functional diagnostic categories, then a remarkable amount of clinical territory is already covered.

The result is almost the inverse of the impression created by the original table. Davenport et al. produce a conspicuous block of No, No, Yes answers once categories such as functional tremor, functional weakness and functional seizures appear in the first column. Remove the diagnostic adjective and compare what initially presents to the clinician, and the functional column becomes crowded. That does not make those presentations functional. It demonstrates why the ability of a presentation to fit somewhere within functional phenomenology is, by itself, a weak discriminator.

Nor does this demonstrate that Davenport's original answers are all false. Some are perfectly defensible under the definitions being used. The point is that much of the apparent separation depends on what question the table is actually asking. "Does MS cause functional tremor?" and "Can MS cause tremor?" are different questions. "Does FND cause neuropathy?" and "Can functional sensory symptoms resemble neuropathic symptoms?" are different questions. "Does FND cause POTS?" and "Can symptoms initially attributed to POTS-like autonomic dysfunction enter functional diagnostic territory?" are different questions. A table can produce very different answers depending on which level it chooses.

Why does FND keep appearing in these differentials?

The immediate reason FND appears in Davenport et al. is that the paper is specifically arguing that Long COVID should not broadly be considered FND. But I think there is a deeper reason FND repeatedly appears around new, rare and difficult-to-diagnose neurological illness. FND itself covers an enormous amount of neurological phenomenology: weakness, tremor, dystonia, jerks, gait disturbance, sensory symptoms, seizure-like episodes, cognitive symptoms, speech disturbance, visual symptoms and more. The wider functional family extends the available functional interpretations further still.

This matters because patients do not generally arrive with the mechanism of their illness already established. They arrive with phenotypes. Weakness. Tremor. Dizziness. Fatigue. Pain. Numbness. Jerks. Falls. Cognitive difficulty. Bladder symptoms. Gastrointestinal symptoms. Exercise intolerance. Some of these may ultimately be explained by FND, some by another functional disorder, some by established neurological disease and some by a disease that has not yet declared itself clearly. The phenotype comes before the explanation.

When an underlying disease is early, rare, atypical, fluctuating, poorly measured or simply not yet understood, there may therefore already be a functional explanation capable of accommodating much of what the patient is experiencing. That does not make the presentation functional. It means functional attribution remains available across an unusually broad amount of uncertain clinical territory.

The diagnosis of the gaps problem

This is where I think the discussion becomes epistemological rather than merely diagnostic. Modern FND practice explicitly rejects diagnosing FND simply because investigations are normal or because another disease has not been found. FND is supposed to be a rule-in diagnosis based on positive clinical evidence. That is an important distinction, but it does not completely answer the harder question created by the breadth of functional phenomenology.

If functional explanations collectively cover such a large proportion of presenting phenomenology, what prevents some patients whose underlying cause has not yet become apparent from being captured inside that framework?

I think of this as the diagnosis of the gaps problem. I do not mean the crude caricature "we don't know, therefore FND". Medicine will always contain gaps: early disease, rare disease, atypical presentations, imperfect tests, disorders without established biomarkers and entirely new illnesses whose mechanisms have not yet been characterised. The problem is subtler. If functional categories already provide plausible interpretations for much of what appears inside those gaps, then compatibility with a functional explanation is remarkably easy to achieve.

That changes the evidential question. Showing that weakness resembles functional weakness, tremor resembles functional tremor or a seizure-like event resembles a functional seizure establishes compatibility. It does not, by itself, establish discrimination. The important question is what evidence reliably separates the functional explanation from the other processes capable of producing a similar phenotype.

This is especially important because Davenport et al. themselves describe precisely this kind of diagnostic territory. They discuss sensory abnormalities that may instead reflect small-fibre neuropathy, and spells in Long COVID that may represent presyncope or syncope, sometimes with convulsive activity, rather than functional seizures. The paper's concern is therefore not abstract: different mechanisms can produce superficially similar clinical events.

The common phenotype hypothesis

This also intersects with what I have called the common phenotype hypothesis: the possibility that at least some presentations grouped under functional diagnoses represent common clinical phenotypes that can be reached through heterogeneous mechanisms. A phenotype can be recognisable without its cause being settled. Weakness is real before its mechanism is known. Tremor is observable before its mechanism is known. A seizure-like event exists before its mechanism is known. The act of recognising the phenotype and the act of explaining it are not the same thing.

If that possibility has any validity, rare and incompletely understood disease is precisely where diagnostic caution becomes most important. A patient with a common phenotype produced by an uncommon mechanism may naturally resemble patients in a much larger functional category. The rarer or less measurable the underlying process, the greater the danger that resemblance itself begins doing more diagnostic work than it should.

This does not prove that FND diagnoses are generally wrong, nor does it make FND meaningless. It changes the question. Instead of asking whether a patient's presentation can be functional, the scientific burden moves towards the specificity, reliability, boundary conditions, longitudinal stability and falsifiability of the positive evidence used to conclude that it is functional rather than another process producing a similar phenotype.

Perhaps both sides are looking at the wrong level

Davenport and colleagues are concerned about inappropriate functional attribution in people with Long COVID and other complex illnesses. Critics are justified in questioning a comparison where categories such as "functional tremor" and "functional seizures" partly determine their own answers. Both observations can be correct while the deeper problem remains unresolved.

What struck me most while reconstructing the table was how difficult it became to maintain absolute boundaries once the diagnostic adjectives were removed. That does not make Long COVID FND, ME/CFS FND or MS FND. Nor does the existence of functional gastrointestinal, urological or vestibular diagnoses make those conditions FND. It demonstrates something narrower and, I think, more important: functional diagnostic territory covers an extraordinary amount of presenting phenomenology.

The interesting question is therefore no longer simply, "Can this look functional?" Across a remarkable range of presentations, the answer can be yes. That is exactly why the question is not very discriminating. The harder question is what evidence allows us to conclude that this particular presentation is functional rather than another process producing the same phenotype, especially when that process is rare, early or not yet understood.

The deeper problem exposed by this exercise is not whether every cell in Davenport's table is right or wrong. Once the diagnostic adjectives are removed, the apparent borders become much less tidy. Functional explanations remain available across weakness, movement, sensation, cognition, seizures, dizziness and neighbouring bodily symptom domains. That makes compatibility relatively easy to demonstrate, but compatibility is not discrimination. The harder question is what evidence reliably distinguishes a functional explanation from another process producing the same phenotype, particularly when that process sits in one of medicine's remaining gaps. That is where the real diagnostic problem begins.

References

Davenport TE, Blitshteyn S, Clague-Baker N, et al. Long COVID Is Not a Functional Neurologic Disorder. Journal of Personalized Medicine. 2024;14(8):799. Full text.

Aybek S, Perez DL. Diagnosis and management of functional neurological disorder. BMJ. 2022;376:o64. Article.

Staab JP, Eckhardt-Henn A, Horii A, et al. Diagnostic criteria for persistent postural-perceptual dizziness (PPPD): Consensus document of the Bárány Society. Journal of Vestibular Research. 2017;27(4):191–208. Full text.

National Institute for Health and Care Excellence. ME/CFS: diagnosis and management (NG206). NICE. 2021. Guideline.

McWilliams A, Reilly C, Gupta J, et al. Pain and functional neurological disorder: a systematic review and meta-analysis. Journal of Neurology, Neurosurgery & Psychiatry. 2024. Full text.

Espay AJ, Aybek S, Carson A, et al. Functional Neurological Disorder: New Phenotypes, Common Mechanisms. 2022. Full text.

Panginikkod S, Rayi A, Cabrero FR, Rukmangadachar LA. Uhthoff Phenomenon. StatPearls. Full text.