Moderators of cognitive behavioural therapy treatment effects and predictors of outcome in the CODES randomised controlled trial for adults with dissociative seizures

Supervised machine learning compared to large language models for identifying functional seizures from medical records

OBJECTIVE

The Functional Seizures Likelihood Score (FSLS) is a supervised machine learning-based diagnostic score that was developed to differentiate functional seizures (FS) from epileptic seizures (ES). In contrast to this targeted approach, large language models (LLMs) can identify patterns in data for which they were not specifically trained. To evaluate the relative benefits of each approach, we compared the diagnostic performance of the FSLS to two LLMs: ChatGPT and GPT-4.

METHODS

In total, 114 anonymized cases were constructed based on patients with documented FS, ES, mixed ES and FS, or physiologic seizure-like events (PSLEs). Text-based data were presented in three sequential prompts to the LLMs, showing the history of present illness (HPI), electroencephalography (EEG) results, and neuroimaging results. We compared the accuracy (number of correct predictions/number of cases) and area under the receiver-operating characteristic (ROC) curves (AUCs) of the LLMs to the FSLS using mixed-effects logistic regression.

RESULTS

The accuracy of FSLS was 74% (95% confidence interval [CI] 65%-82%) and the AUC was 85% (95% CI 77%-92%). GPT-4 was superior to both the FSLS and ChatGPT (p <.001), with an accuracy of 85% (95% CI 77%-91%) and AUC of 87% (95% CI 79%-95%). Cohen's kappa between the FSLS and GPT-4 was 40% (fair). The LLMs provided different predictions on different days when the same note was provided for 33% of patients, and the LLM's self-rated certainty was moderately correlated with this observed variability (Spearman's rho2: 30% [fair, ChatGPT] and 63% [substantial, GPT-4]).

SIGNIFICANCE

Both GPT-4 and the FSLS identified a substantial subset of patients with FS based on clinical history. The fair agreement in predictions highlights that the LLMs identified patients differently from the structured score. The inconsistency of the LLMs' predictions across days and incomplete insight into their own consistency was concerning. This comparison highlights both benefits and cautions about how machine learning and artificial intelligence could identify patients with FS in clinical practice.

Predictors of Treatment Success of Psychotherapy in Functional Disorders: A Systematic Review of the Literature

OBJECTIVE

Functional disorders (FDs) benefit from psychotherapy. However, the determinants predicting their efficacy remain largely unexplored.

METHODS

A systematic literature review was conducted. PubMed, Web of Science, Embase, Cochrane collaboration and grey literature were screened from inception to November 2024. Randomized controlled trials on predictors of success of psychotherapy for FDs (e.g., somatoform disorders, irritable bowel syndrome [IBS], chronic fatigue syndrome/myalgic encephalomyelitis [CFS/ME], fibromyalgia [FM]) in adults (i.e., ≥ 18 years of age) were included. The review yielded 24 eligible studies and included 3382 participants. A standardized quality assessment via ROB-2 Tool was performed. PRISMA guidelines were followed.

RESULTS

Most studies applied CBT-based interventions (n = 19), mainly face-to-face, with some internet-based (n = 5), while fewer used emotional-based (n = 4), mindfulness-based (n = 3), psychodynamic (n = 1) or operant behavioural therapy (n = 1). The primary factors identified as predictive of treatment success in FM and somatization were the intensity of experienced pain. Moreover, the presence of mental disorders, i.e., depression and anxiety disorders, emerged as predictors for a range of disorders including FM, IBS, somatization disorder, hypochondriasis, medically unexplained symptoms and dissociative seizures. Symptom severity was recognized as a predictor across various FDs with findings indicated that severe severity could predict treatment outcomes.

CONCLUSION

The body of research concerning predictors of treatment success in the context of FDs can help clinicians identifying appropriate psychotherapy trajectories.

TRIAL REGISTRATION

Not applicable. PROSPERO no. CRD42022379791; OSF (https://osf.io/8q7z9).

Reflections on the CODES trial for adults with dissociative seizures: what we found and considerations for future studies

The COgnitive behavioural therapy versus standardised medical care for adults with Dissociative non-Epileptic Seizures multicentre randomised controlled trial is the largest, fully-powered study to test the clinical and cost-effectiveness of a psychotherapeutic intervention in this population. We also explored predictors or moderators of outcomes and investigated mechanisms of change in therapy. In this current review of findings, we discuss issues related to the design of the trial and consider the study's nested qualitative studies which were undertaken not only to shed light on the original research questions but to provide insights and recommendations for other researchers in the field of functional neurological disorder. Finally, we consider issues relating to the possible clinical application of our study findings.

How does cognitive behavior therapy for dissociative seizures work? A mediation analysis of the CODES trial

BACKGROUND

We compared dissociative seizure specific cognitive behavior therapy (DS-CBT) plus standardized medical care (SMC) to SMC alone in a randomized controlled trial. DS-CBT resulted in better outcomes on several secondary trial outcome measures at the 12-month follow-up point. The purpose of this paper is to evaluate putative treatment mechanisms.

METHODS

We carried out a secondary mediation analysis of the CODES trial. 368 participants were recruited from the National Health Service in secondary / tertiary care in England, Scotland, and Wales. Sixteen mediation hypotheses corresponding to combinations of important trial outcomes and putative mediators were assessed. Twelve-month trial outcomes considered were final-month seizure frequency, Work and Social Adjustment Scale (WSAS), and the SF-12v2, a quality-of-life measure providing physical (PCS) and mental component summary (MCS) scores. Mediators chosen for analysis at six months (broadly corresponding to completion of DS-CBT) included: (a) beliefs about emotions, (b) a measure of avoidance behavior, (c) anxiety and (d) depression.

RESULTS

All putative mediator variables except beliefs about emotions were found to be improved by DS-CBT. We found evidence for DS-CBT effect mediation for the outcome variables dissociative seizures (DS), WSAS and SF-12v2 MCS scores by improvements in target variables avoidance behavior, anxiety, and depression. The only variable to mediate the DS-CBT effect on the SF-12v2 PCS score was avoidance behavior.

CONCLUSIONS

Our findings largely confirmed the logic model underlying the development of CBT for patients with DS. Interventions could be additionally developed to specifically address beliefs about emotions to assess whether it improves outcomes.

[Conclusion instead of exclusion-The clinical diagnosis of functional movement disorders]

Functional neurological movement disorders are common in neurological practice and lead to a high degree of impairment and chronification. Affected patients usually receive a diagnosis with considerable delay and often do not get disease-specific treatment. The reasons for this delay are related to extensive diagnostic measures to exclude other nonfunctional neurological diseases. As a consequence, functional movement disorders are typically communicated as diagnoses of exclusion, which makes it difficult for patients to understand and accept the diagnosis. This is particularly unfortunate, because in the majority of patients the diagnosis can be made with confidence based on clinical features, i.e., inconsistency and incongruence. The clarification of the symptoms and the resulting treatment options should be supplemented by patient-friendly explanations of the pathophysiological basis of the disease. In this way, patients are enabled to understand and accept the diagnosis. Moreover, it can put an end to the search for a diagnosis, which can sometimes take decades, and paves the way for treatment. Thus, the diagnosis by exclusion itself becomes the starting point for treatment and can itself have a therapeutic effect.

An update on psychogenic nonepileptic seizures

PURPOSE OF REVIEW

The understanding of psychogenic nonepileptic seizures (PNES) has advanced steadily over recent decades. This update summarizes new insights from the last three years.

RECENT FINDINGS

The process of diagnosing PNES has shifted from the exclusion of epilepsy to one based on the recognition of typical clinical features. While the diagnosis cannot rely on any single feature in isolation, a range of semiological features characterising PNES are now recognised and a number of studies hint at the potential for machine learning and AI to improve the diagnostic process. Advances in data processing and analysis may also help to make sense of the heterogeneity of PNES populations demonstrated by recent studies focussing on aetiology and patient subgroups. It is now clear that PNES are associated with high rates of mental and physical comorbidities and premature death, highlighting that they are only one manifestation of a complex disorder extending beyond the nervous system and the seizures themselves.

SUMMARY

PNES are now understood as a manifestation of dysfunction in interacting brain networks. This understanding provides an explanation for the psychopathological and semiological heterogeneity of PNES patient populations. New insights into medical comorbidities and increased rates of premature death call for more research into associated pathological processes outside the nervous system.

Using Verbally-Reported and Video-Observed Semiology to Identify Functional Seizures

Diagnosis of functional seizures, also known as psychogenic nonepileptic seizures, starts with a clinical interview and description of the seizures. A targeted approach to this evaluation can provide valuable information to gauge the likelihood of functional seizures as compared with other similar conditions including but not limited to epileptic seizures. This review focuses on the use of patient and witness descriptions and seizure videos to identify patients with probable functional seizures. Particular emphasis is given to recognizing the limitations of the available data and the influence of health-care provider expertise on diagnostic accuracy.

Improving the treatment of functional seizures through a public specialist outpatient clinic

OBJECTIVES

We performed an audit of the first 12 months of clinical operations to assess the feasibility of a newly established public outpatient clinic for the assessment and treatment of functional (psychogenic nonepileptic) seizures (FS).

METHOD

Clinical notes for the first 12 months of the FSclinic weresystematicallyreviewed with data compiled onreferral pathways, clinic attendance, clinical features, treatments, and outcomes.

RESULTS

Of eighty-two new FS patients referred to the clinic, over 90% attended. Patients were diagnosed with FS after comprehensive epileptological and neuropsychiatric review, mostly with typical seizure-like episodes captured during video-EEG monitoring, and most accepted the diagnosis. Most had FS at least weekly, with little sense of control and significant impairment. The majority of individuals had significant psychiatric and medical comorbidity. Predisposing, precipitating, and perpetuating factors were readily identified in >90% of cases. Of 52 patients with follow-up data within12 months, 88% were either stable or improved in terms of the control of their FS.

CONCLUSION

The Alfred functional seizure clinic model, the first dedicated public outpatient clinic for FS in Australia, provides a feasible and potentially effective treatment pathway for this underserved and disabled patient group.

Speech and language therapy: A treatment case series of 20 patients with functional speech disorder

BACKGROUND

While expert consensus recommendations support the use of speech and language therapy (SLT) for patients with functional speech disorder (FND-speech), there are limited published data on clinical outcomes.

OBJECTIVE

To retrospectively report the treatment outcomes and clinical characteristics of patients with FND-speech that attended outpatient SLT as part of a multidisciplinary program for functional neurological disorder (FND).

METHODS

In this case series, we included adult patients with FND-speech that consecutively participated in outpatient SLT at our institution between October 2014 and September 2021. Baseline demographic and neuropsychiatric characteristics were extracted from the medical records, along with data on FND-speech phenotypes, number of treatment sessions received, and clinician-determined outcomes. Only descriptive statistics were used to report findings.

RESULTS

Twenty patients met inclusion criteria; ages ranged from 21-77, with a mean of 51.6±16.2 years. 85% of the cohort presented with mixed FND-speech symptoms. Patients attended a range of 2-37 visits, with an average of 9.2±8.0 visits over 4.4±3.5 months. At the last treatment session, 3 patients were asymptomatic, 15 had improved, and 2 had not improved; 8 individuals that improved received video telehealth interventions.

CONCLUSION

This case series lends additional support for outpatient SLT in the assessment and management of individuals with FND-speech, and may help clarify patient and provider treatment expectations. Additional prospective research is needed to investigate baseline predictors of treatment response and further define the optimal frequency, intensity, duration, and clinical setting for SLT delivery in this population.

Toward a possible trauma subtype of functional neurological disorder: Impact on symptom severity and physical health

BACKGROUND

As a group, individuals with functional neurological disorder (FND) report an approximately 3-fold increase in adverse life experiences (ALEs) compared to healthy controls. In patients with FND, studies have identified a positive correlation between symptom severity and the magnitude of ALEs. While not all individuals with FND report ALEs, such findings raise the possibility of a trauma-subtype of FND.

OBJECTIVE

This study investigated if patients with FND, with or without probable post-traumatic stress disorder (PTSD) and/or significant childhood maltreatment, differed in their symptom severity and physical health.

MATERIALS AND METHODS

Seventy-eight patients with FND were recruited (functional seizures, n = 34; functional movement disorder, n = 56). Participants completed self-report measures of symptom severity [Somatoform Dissociation Questionniare-20 (SDQ-20), Screening for Somatoform Disorders: Conversion Disorder subscale (SOMS:CD), Patient Health Questionniare-15 (PHQ-15)], physical health [Short Form Health Survey-36 (SF36-physical health)], childhood maltreatment [Childhood Trauma Questionnaire (CTQ)], and PTSD [PTSD Checklist-5 (PCL-5)]; a psychometric battery of other common predisposing vulnerabilities was also completed. To adjust for multiple comparisons, a Bonferroni correction was applied to all univariate analyses.

RESULTS

Patients with FND and probable PTSD (n = 33) vs. those without probable PTSD (n = 43) had statistically significant increased scores on all symptom severity measures - as well as decreased physical health scores. In secondary post-hoc regression analyses, these findings remained significant adjusting for age, sex, race, college education, and: pathological dissociation; alexithymia; attachment styles; personality characteristics; resilience scores; functional seizures subtype; or moderate-to-severe childhood abuse and neglect scores; SOMS:CD and SDQ-20 findings also held adjusting for depression and anxiety scores. In a separate set of analyses, patients with FND and moderate-to-severe childhood abuse (n = 46) vs. those without moderate-to-severe childhood abuse (n = 32) showed statistically significant increased SDQ-20 and PHQ-15 scores; in post-hoc regressions, these findings held adjusting for demographic and other variables. Stratification by childhood neglect did not relate to symptom severity or physical health scores.

CONCLUSION

This study provides support for a possible trauma-subtype of FND. Future research should investigate the neurobiological and treatment relevance of a FND trauma-subtype, as well as continuing to delineate clinical characteristics and mechanisms in individuals with FND that lack a history of ALEs.