Functional Seizures: When a Diagnosis Becomes an Explanation

Author profile image FND Nope Moderator
0 comments

In July 2014, investigators involved in the CODES research programme produced a patient factsheet explaining dissociative seizures, their proposed causes, diagnosis and treatment. The document described seizures as episodes in which the brain effectively switches off through dissociation, encouraged patients to understand and accept the diagnosis, and outlined psychological treatment as a route towards recovery.

Six years later, the same research programme published the results of a major randomised controlled trial investigating cognitive behavioural therapy (CBT) for dissociative seizures. The trial failed to demonstrate a statistically significant benefit on its predefined primary outcome: seizure frequency at 12 months. As previously examined by FND Nope in our article on David Tuller's investigation, researchers subsequently emphasised favourable secondary outcomes and questioned whether seizure frequency had been the appropriate primary measure.

Reading the original factsheet alongside the subsequent trial raises a question extending beyond treatment effectiveness. How much of what patients were told represented established evidence, and how much reflected a theoretical framework that had yet to be adequately tested?


What Patients Were Told Before the Trial

The Dissociative Seizures factsheet, dated 16 July 2014, was designed for use by neurologists and prepared by CODES Trial investigators, including Jon Stone, Markus Reuber, John Mellers, Trudie Chalder, Alan Carson and Laura Goldstein. It presented dissociative seizures as genuine episodes resembling epilepsy but arising through a different neurological process. The proposed mechanism was dissociation, described as an unconscious switching-off response comparable to daydreaming, becoming detached during frightening experiences or mentally withdrawing from unpleasant situations.

By page 8, the document moved from explanation to treatment, stating that understanding and accepting the diagnosis helped many people recover. It recommended psychiatric or psychological involvement and discussed the potential importance of psychological factors. The document also acknowledged that neither clinicians nor patients necessarily identified relevant psychological factors. On page 9, it conceded that further research was needed to establish whether specific talking treatments worked.

That qualification matters. The factsheet did not explicitly claim that CBT had already been proven effective by the forthcoming CODES trial. Nevertheless, it presented patients with a coherent explanatory framework in which dissociation, psychological processes, worry, learned responses and treatment through greater understanding were closely connected. Whether that framework accurately explained the individual patient's seizures remained a separate question.


Then Came the CODES Results

The CODES randomised controlled trial, published in The Lancet Psychiatry in 2020, enrolled 368 participants. Of these, 186 received CBT alongside standardised medical care, while 182 received standardised medical care alone. The predefined primary outcome was monthly dissociative seizure frequency at 12 months.

At that point, the median monthly seizure frequency was four in the CBT group and seven in the comparison group. Although the numerical difference favoured CBT, the adjusted incidence rate ratio was 0.78, with a 95% confidence interval of 0.56โ€“1.09 and a p-value of 0.144. The trial therefore did not demonstrate a statistically significant between-group difference on its primary outcome. This does not prove that CBT has no effect, but the trial did not establish the benefit it was primarily designed to test.

Several secondary outcomes favoured CBT, including psychosocial functioning, longer seizure-free periods, general impressions of improvement and treatment satisfaction. Those findings deserve accurate reporting, but they answer different questions. Feeling improved, reporting greater satisfaction or functioning better socially does not necessarily establish that seizure frequency has been significantly reduced. The trial's secondary p-values were not adjusted for multiple testing, making their interpretation particularly important.

This distinction became central to David Tuller's May 2025 investigation. Tuller examined how CODES investigators subsequently placed greater emphasis on favourable secondary findings and questioned the suitability of seizure frequency as the primary endpoint. As discussed in our earlier article, this was particularly contentious because seizure reduction had been selected prospectively and had also featured in the programme's earlier research.

The criticism is not that secondary improvements are meaningless. It is that a trial cannot retrospectively establish success on its original research question by elevating different outcomes after the primary analysis fails to demonstrate a significant benefit. If investigators believe they selected an inappropriate endpoint, that is a methodological limitation requiring acknowledgement and potentially further research, not a reason to reinterpret the original result as though the primary objective had been achieved.


But Did the Trial Establish the Proposed Mechanism?

There is a more fundamental limitation. CODES was a treatment trial, not a direct experimental validation of dissociation as the cause of the participants' seizures. Even a successful CBT trial would not, by itself, establish that every participant's episodes arose through the proposed dissociative mechanism. Psychological interventions can influence distress, coping, behaviour and symptoms without necessarily identifying the underlying cause of a condition.

The original factsheet, however, describes dissociation as the process responsible for the seizures. It compares severe attacks involving collapse, shaking and unresponsiveness with ordinary experiences of attentional absorption or detachment. These analogies may communicate a theoretical model, but an analogy is not a demonstration that the same mechanism produces the different phenomena.

The document also describes attacks associated with stress, hyperventilation, pain, relaxation or no identifiable trigger. Such associations may be clinically relevant, but they do not independently establish dissociation as the causal explanation. If almost any triggering circumstance can be incorporated into the same model, the question becomes what observations would actually distinguish that model from competing explanations.

The subsequent failure to demonstrate a significant primary treatment effect does not disprove dissociation. It does, however, leave the original causal claims requiring independent evidence. The treatment trial cannot supply that missing demonstration, and favourable secondary outcomes cannot establish a neurological mechanism that the trial was not designed to measure.


Diagnostic Confidence and the Limits of EEG

The 2014 factsheet claims that specialists can correctly diagnose attacks in at least eight out of ten cases using descriptions, rising to nine out of ten when video recordings are available. These figures are presented without sufficient methodological detail for patients to assess the reference standards, populations or alternative diagnoses considered. The factsheet also illustrates epileptic seizures with abnormal EEG activity and dissociative seizures with normal EEG activity.

Yet the International League Against Epilepsy published a staged diagnostic framework in 2013, distinguishing possible, probable, clinically established and documented psychogenic non-epileptic seizures. Diagnostic certainty was therefore already recognised as dependent on the quality and extent of available evidence before the CODES factsheet was produced.

Furthermore, a normal scalp EEG does not universally exclude epilepsy. A study comparing simultaneous scalp and intracranial EEG recordings found that scalp EEG detected 100 of 172 recorded seizures in a selected group undergoing invasive epilepsy investigation. Detection varied substantially by seizure type, with focal aware and subclinical seizures particularly likely to lack a detectable scalp correlate. These findings cannot be generalised to every seizure presentation, but they demonstrate why an absence of recorded epileptic activity must be interpreted cautiously.

The CODES trial itself did not require video-EEG confirmation for every participant. Its published baseline data indicate that 195 of 368 participants, approximately 53%, had diagnoses supported by EEG and video. Where that evidence was unavailable, the trial permitted alternative expert clinical assessment. This was an explicitly defined eligibility procedure, but it is important when interpreting the diagnostic certainty of the trial population. The study was not restricted to patients whose typical episodes had all been documented by video-EEG.

None of this means the participants were necessarily misdiagnosed. It means the available evidence varied, and that classification, exclusion of epilepsy and demonstration of a dissociative mechanism remain distinct questions. A treatment trial conducted within an established diagnostic framework does not independently validate every assumption underlying that framework.


The Safety Assurances Are Another Problem

Page 10 of the CODES factsheet makes exceptionally strong claims about seizure safety. It advises that emergency medical attention is unnecessary, asserts that severe harm has not been recorded and states that breathing cannot stop during a dissociative seizure. Such categorical assurances are troubling in material intended to guide patients, relatives and carers, particularly when future episodes may differ from those previously assessed.

A 2023 systematic review and meta-analysis published in Epilepsia examined injuries associated with functional seizures. Across 13 studies involving 1,673 individuals, the pooled lifetime prevalence of seizure-related injuries was 25%, with a 95% confidence interval of 19โ€“32%. The certainty of this estimate was rated very low, and it included injuries of differing severity. It therefore cannot establish that severe injuries occur in one quarter of patients. Nevertheless, it documents that injuries do occur and undermines any general impression that functional seizures are inherently incapable of causing physical harm.

More strikingly, the 2020 CODES trial itself recorded adverse events, including injuries attributed to dissociative seizures. Its serious-adverse-event table also included events classified as related to dissociative seizures. These observations do not establish that every serious event represented direct physical injury from an attack, but they make blanket assurances of harmlessness difficult to defend. The trial's own safety reporting demonstrates why individualised risk assessment matters.

The factsheet also acknowledges that some patients with dissociative seizures have a history of epilepsy. The CODES trial permitted participants with comorbid epilepsy only if they had been free of epileptic seizures for 12 months. An existing functional diagnosis therefore cannot justify assuming that every subsequent episode is non-epileptic or harmless. New, prolonged or substantially different attacks, breathing difficulties, serious injuries and abnormal recovery may require urgent medical assessment.

These safety findings are separate from the trial's treatment results, but they expose another weakness in the original patient information: reassurance was sometimes expressed as certainty where clinical caution would have been more appropriate.


What the CODES Story Actually Demonstrates

The sequence is revealing. In 2014, patients were given an explanatory account centred on dissociation, psychological influences and the possibility of recovery through understanding and treatment. In 2020, the research programme's major CBT trial failed to demonstrate a statistically significant improvement on its predefined primary measure of seizure frequency, although several secondary outcomes favoured treatment. In 2025, David Tuller drew attention to the subsequent emphasis on those secondary outcomes and the questioning of the original endpoint.

Alongside that history, diagnostic research demonstrates important limitations in the interpretation of seizure presentations and scalp EEG, while subsequent injury research challenges the categorical reassurance offered in the original factsheet. These are not interchangeable criticisms, but they point towards the same broader problem: confidence in a clinical framework can extend beyond what particular studies have actually established.

The CODES trial did not prove that dissociation causes functional seizures. It did not establish a statistically significant reduction in seizure frequency from CBT at 12 months. Its positive secondary findings cannot be discarded, but neither can they erase the primary result or validate the proposed causal mechanism. The original factsheet's safety claims also require qualification in light of documented injuries and the need to recognise medical emergencies.

The issue is not whether patients may benefit from psychological support or whether clinicians can recognise presentations associated with functional seizures. It is whether patients are clearly told which conclusions have been demonstrated, which remain provisional and which are theoretical explanations. A diagnosis should not be treated as proof of its proposed cause, and a treatment framework should not be treated as vindicated merely because some outcomes improve.

Patients were given a confident explanation before the definitive treatment trial was completed. When that trial failed to demonstrate a significant benefit on its primary outcome, the appropriate response was to acknowledge the limitation, not to shift attention away from the question the trial was designed to answer. Diagnostic confidence, theoretical explanation and treatment success are three different claims. Each requires its own evidence.


References