Introduction

Background to this project

In 2012, a call was issued by the National Institute for Health Research (NIHR) Health Technology Assessment (HTA) programme for a multicentre, two-arm randomised controlled trial (RCT) to answer the question ‘What is the clinical and cost-effectiveness of cognitive–behavioural therapy (CBT) for psychogenic non-epileptic seizures (PNESs)?’ (see Appendix 1). The call specified that the CBT should be tailored for adult PNES patients whose condition persisted beyond diagnosis by a neurologist or neuropsychiatrist. The trial would be conducted in outpatient or community healthcare settings.

The decision to include patients with comorbid epilepsy would be considered by the applicants; consideration would also be given to psychiatric comorbidities. CBT would be compared with standard medical care, which was to be defined by the research team; however, it was recognised that this might include a patient information leaflet. Potentially important outcomes were identified as seizure frequency and severity, as well as duration of seizure freedom. Other outcomes identified were psychiatric symptomatology, psychosocial functioning, quality of life, healthcare utilisation, cost-effectiveness and socioeconomic factors (i.e. disability and return to work). The minimum follow-up duration was set as 1 year post randomisation.

The call acknowledged that although psychotherapy is deemed the first-line intervention for PNESs, there is limited evidence of its effectiveness, despite reports that CBT has been shown to be successful in treating other somatoform conditions. A pilot RCT demonstrated that a version of CBT designed specifically for patients with PNESs was able to reduce seizure frequency over standard care.1 The call highlighted the requirement for a fully powered RCT to evaluate the clinical effectiveness and cost-effectiveness of CBT tailored for PNESs. The current study is a response to the call made by NIHR, proposing the expansion of the earlier pilot1 into a multicentre RCT recruiting patients from neurology services. The call recognised that PNESs are often referred to by other names. We adopted the term dissociative seizures (DSs) for reasons discussed in the next section (see Dissociative seizures). The trial was subsequently named ‘COgnitive behavioural therapy vs. standardised medical care for adults with Dissociative non-Epileptic Seizures: a multicentre randomised controlled trial (CODES)’.

Dissociative seizures

Dissociative seizures are paroxysmal behavioural events that resemble epileptic seizures or syncope; they are, however, not a result of epilepsy or any other medical condition. They are widely understood as involuntary episodes that arise via dissociative mechanisms and as a disorder sitting at the interface between neurology and psychiatry.1–3 DSs are currently classified as conversion (functional neurological symptom) disorder4 and as a dissociative neurological symptom disorder.5 For the purposes of the current study, DSs were defined as events leading to a transient loss of consciousness or apparent altered responsiveness. Such episodes would be clinically incompatible with recognised neurological or general medical conditions, and not better explained by another physical or psychiatric disorder. Symptoms or deficits would also cause significant distress, psychosocial impairment or warrant medical evaluation.

Dissociative seizures are also referred to as ‘pseudoseizures’, ‘functional seizures’, ‘psychogenic non-epileptic seizures’, ‘non-epileptic events/spells’ or ‘non-epileptic attack disorder’ among other names. Currently, the most prominent term used is PNESs, although it may have negative connotations for some patients,6,7 with the term ‘psychogenic’ potentially implying that the attacks are ‘deliberate’ or ‘imagined’ despite the intentions of healthcare professionals (HCPs). We chose the term ‘dissociative seizures’ in line with the International Classification of Diseases and Related Problems, Tenth Revision, classification that was current at the time that the project started,8 as it encourages a positive diagnosis and discussion of an aetiological mechanism.

The prevalence of DSs is estimated as 2–33 per 100,000 people,9 although more recent assessments place this estimate closer to 50 per 100,000 people.10 A recent UK study2 documented the incidence rate as 4.9 per 100,000 people per year. Women form around three-quarters of all patients with DSs,11 with onset typically occurring in the late teens/twenties,12 although DSs can occur in both the old and the young.13–15 It is thought that around 12–20% of patients presenting at epilepsy clinics may have DSs.16

Video electroencephalography (EEG) is the gold standard diagnostic technique for DSs, with the key signs17 being the absence of typical epileptic EEG abnormalities and the presence of an intact alpha rhythm. However, a careful history alone should usually provide sufficient grounds to suspect DSs.18 It is estimated that about 22% of DS patients have comorbid epileptic seizures,19 which may create problems in diagnosis and management.18

Up to 80% of patients with DSs have a history of other functional somatic symptoms and disorders, including other functional neurological symptoms.20 There are high levels of psychiatric comorbidity in patients with DSs,21–23 including maladaptive personality traits, post-traumatic stress disorder (PTSD), anxiety and depression, with patients demonstrating less adaptive coping styles.24,25 It has also been reported that patients with DSs might have a slightly higher risk of mortality; however, this is not directly related to the seizures.26

A recent meta-analysis has shown a clear association between the presence of childhood stressors and the presence of adulthood stressors in patients with DSs, with an odds ratio of 3.1 [95% confidence interval (CI) 1.7 to 5.6].27 In addition, the perceived impact of traumatic events appears greater.28 Individual studies have shown an association between history of trauma and DSs, with 44–100% of patients reporting a history of trauma and 23–77% of patients reporting being physically and/or sexually abused.29 This suggests that these experiences may be potential risk factors for DSs for a significant proportion of patients. The frequency of childhood sexual abuse is higher among females with DSs,30 and it has been reported that, for example, in one DS population studied, 41% of females had suffered from sexual abuse.31 This is a much higher rate than the estimated rate of sexual abuse (15–25%) in the general female population.32 Reports of sexual abuse increase the likelihood of a diagnosis of DSs by approximately threefold.28,33 Importantly, however, a diagnosis of DSs is not dependent on a history of abuse.22 In addition, patients with DSs report more life events than those with epilepsy and motor conversion disorder in the 12 months prior to DS onset,34,35 and it has been considered that repeated adverse experiences over the longer term may be as relevant as acute life events in the development of DSs.22

In addition to trauma, two-thirds of patients with DSs report significant problems in family and social environments.31 When compared with patients with epilepsy, marital and family problems have been reported to be more prevalent in patients with DSs.36 Individuals with DSs have reported viewing their family as having a dysfunctional communication style,37 which may contribute to DS symptomology through distress, criticism and a tendency to somatise.38 As a result, individuals with DSs sometimes perceive their families as lacking in commitment and support.36

Quality of life in individuals with DSs is lower than in those who are diagnosed with epilepsy,39 which may in part be because of the association between quality of life and depression, dissociation, somatic symptoms, escape–avoidance coping strategies and family dysfunction.40 Patients with DSs can experience very restricted lives and demonstrate high levels of avoidance behaviour41–43 owing to fears of having a seizure. Individuals with DSs may also experience high levels of stigma related to their seizures,44 which can lead to individuals isolating themselves to avoid any adverse social reactions.45,46 In addition, family members can influence quality of life in patients with DSs. Those who have a family environment characterised by high levels of criticism and lack of interest and support have lower health-related quality of life (HRQoL).47

In a study of longer-term outcome of 50 patients that was undertaken by retrospective analysis after an average of 2 years, over half of the patients were found to be in either a poor or a very poor state because of a combination of physical, psychological and social issues.48 In addition to a potentially low quality of life, around 70% of individuals with DSs have poor long-term outcomes, including chronic disability and welfare dependence.49 Studies suggest that around half of patients are receiving or dependent on disability/state benefits during follow-up.49,50 In conjunction with this, it has been surmised51 that the societal costs of having DSs can be considerable and can, as a result of rates of unemployment and disability, approximate the costs associated with intractable epilepsy. There is also some evidence that disability and welfare dependence can continue even if patients become seizure free.52,53

The most consistent predictive factor of poor outcome, defined in different ways in different studies, for people with DSs appears to be the duration of the symptoms: the longer the duration, the more likely the negative outcome.48,54,55 Further factors that negatively predict outcomes include receiving social security benefits;56 the presence of previous psychiatric diagnoses,56 with evidence specifically relating to the presence of depression56,57 and anxiety;56 poor psychopathology inventory scores;49 and a difficulty in forming relationships.58 Personality factors may also be associated with different patterns of outcome.49,57,59 Factors predictive of good outcomes have variously been found to include employment;2,60 higher educational achievement49,61 and intelligence quotient;62 low somatisation scores;49 and being accompanied to the first clinic visit.61 There have also been indications, of varying strength, that hyperkinetic DS semiology is associated with a less good outcome than hypokinetic DS semiology.48,49,54,61

Health economic aspects of dissociative seizures

There is a lack of research investigating the healthcare costs incurred by people with DSs. However, the literature often refers to the prevalence of unnecessary tests and treatments that patients can undergo that are expensive and potentially harmful.63 A study in the USA64 reported the pre-diagnosis cost of this patient group (excluding diagnostic tests) as potentially exceeding US$25,000 per patient, based on an average of 6 years to arrive at the correct diagnosis. A similar investigation in Ireland65 reported an annual pre-diagnosis cost of €5429.30, assuming that it takes an average of 5 years to reach a DS diagnosis. A correct diagnosis of DSs appears to lead to a decrease in health service use and a subsequent reduction in costs.66,67 Although it is possible that patients with DSs may later develop other medically unexplained symptoms,20 which may also have financial implications, there is no evidence to predict whether or not this likelihood is reduced by psychological interventions.

Treatment for dissociative seizures

Although some research in the last 10 years has investigated pharmacological treatment for DSs using antidepressants,68,69 the treatment of choice in both clinical and research settings remains psychotherapy.70,71 There has been some research on the beneficial use of psychodynamic therapy,72,73 group psychotherapy74 and psychoeducational approaches.75,76 However, CBT has the most substantial body of data suggesting effectiveness in treating a range of somatoform disorders,77,78 although effect sizes have been noted to be small to medium.78 More specifically, with respect to treating patients with DSs, the evidence for DS reduction comes from small, open-label studies and pilot RCTs.79,80 Carlson and Nicholson Perry79 reviewed 13 studies of mixed quality and somewhat differing interventions that demonstrated that 47% of patients were reported to be seizure free at the end of the psychological intervention; they also reported that, considering data available from 10 studies, 82% of patients had shown at least a 50% reduction in DS occurrence at the end of the intervention.

Martlew et al.80 reviewed data from eight open-label studies and four RCTs, and concluded that the strongest evidence for DS reduction at that point came from a pilot RCT.1 This RCT, based on a previous single case study81 and an open-label study,82 was undertaken at just one clinical site and compared a manualised CBT treatment for DSs plus standard medical care with standard medical care alone. Standard medical care was provided within a specialist tertiary neuropsychiatry service. A total of 66 patients were allocated at random to two equally sized treatment arms. Patients who were allocated to receive CBT plus standard medical care were offered 12 sessions of DS-specific CBT (plus up to two booster sessions), including handouts supporting the session content. The structure of the treatment package was described.1,83,84 Intention-to-treat analyses showed a significant reduction in DSs within the CBT arm at the end of treatment; at follow-up, the CBT arm tended to have fewer seizures than the arm receiving standard medical care only. In addition, at follow-up the CBT arm tended to be more likely to be seizure free for the last 3 months of the study than the standard medical care arm. Both trial arms showed a degree of improvement in some measures of health service use, as well as on the Work and Social Adjustment Scale; however, anxiety, depression and employment status showed no change. Although this study provided preliminary evidence of the efficacy of CBT for this patient group, it lacked independent outcome assessors and demonstration of effectiveness across multiple sites and practitioners.

Additional evidence for the efficacy of a largely CBT-informed approach in the treatment of DSs has been found within a small, four-arm, pilot RCT.68 Thirty-eight patients (of whom 34 provided outcome data) from three sites were randomised to four treatment arms: 12 sessions of CBT-informed psychotherapy (CBT-ip) (n = 9), 12 sessions of CBT-ip plus flexible low-dose sertraline (n = 9), sertraline alone (n = 9) and treatment as usual (n = 7). CBT-ip was heavily informed by principles of CBT but also included other therapeutic techniques derived from mindfulness, dialectical behaviour therapy and psychodynamic approaches. The study was not powered to allow between-condition comparisons for the primary outcome of DS frequency; therefore, comparisons were made within treatment allocation. The two treatment arms incorporating CBT-ip demonstrated significant seizure reduction and improvement in functioning and scores on symptom scales, whereas the sertraline-alone arm showed a trend towards DS reduction but no change in secondary outcomes. No significant changes were found in the treatment-as-usual arm. There were also significant reductions in clusters of DSs in the two trial arms delivering CBT-ip, although the extent of the reduction depended on how clusters were defined.85

Nonetheless, despite showing potential efficacy of a CBT approach to the reduction of DS occurrence, neither pilot RCT was an adequately powered effectiveness study that provided a sufficiently robust evidence base on which to make treatment recommendations.

Although the literature seems to suggest that psychotherapy (and, potentially, specifically CBT) should be the treatment of choice for patients with DSs, there is no uniform, recommended care pathway for these patients in the UK.71 Furthermore, the availability of psychiatric and psychological services for the assessment, diagnosis and management of DSs is highly variable, despite the potential impact of DSs on patients, their families and society.70 Mayor et al.71 found that 15% of HCPs who responded to their survey indicated that they had nowhere to refer their patients; only one-third indicated that they would be able to refer patients with DSs for psychotherapy. Such psychotherapy might be further limited, given that these professionals reported that under half of their patients would be offered at least one psychotherapy session. Thus, although evidence exists for the benefit of psychotherapeutic input, the lack of an evidence-based care pathway and of evidence from larger robust treatment trials means that many patients may not be funded by their Clinical Commissioning Groups to receive psychotherapy. In addition, the geographical distribution of the patients means that in outside specialist centres there may be limited knowledge or willingness to enable these patients to be seen, resulting in inequalities in healthcare provision. A stronger evidence base would provide the basis on which policy changes to the service provision for DS patients could be made. This need is highlighted by the National Institute for Health and Care Excellence86 and the Scottish Intercollegiate Guidelines Network,87 which indicate the need for psychiatric and psychological input for DS patients. In addition, the International League Against Epilepsy,88 US National Institutes of Health89 and the National Institute for Neurological Disorders and Stroke90 have all identified the need to develop effective methods for treating DSs.

Summary and methodological rationale

The CODES trial was designed to evaluate the clinical effectiveness and cost-effectiveness of CBT for DSs within a care pathway involving neurology, neuropsychiatry and psychotherapy. It offers a potential template for future services and the commissioning of DS treatment. In addition, it can form the basis for the more extensive training of therapists to work with patients with DSs and support the importance of psychiatrists in treating this patient group, who often present with complex mental health needs.70 Preliminary evidence of efficacy had been obtained through a proof of principle RCT1 and, therefore, this study was the next step in examining the clinical effectiveness and cost-effectiveness and generalisability of CBT as an intervention for DSs through a pragmatic, adequately powered, multicentre RCT.

Research objectives

The main aim of this research, as determined by the NIHR HTA programme commissioning brief, was to evaluate the effectiveness of a psychological intervention (DS-specific CBT) plus standardised medical care (SMC) (i.e. CBT + SMC) compared with SMC alone in improving DS control and a range of psychosocial outcomes and in reducing health service use and costs.

The primary objective was to assess the effectiveness of CBT + SMC compared with SMC alone in reducing monthly DS frequency at 12 months post randomisation.

Secondary objectives were to assess the effectiveness of CBT + SMC compared with SMC alone at 12 months post randomisation in relation to:

• reductions in DS severity
• improvements in seizure freedom, psychosocial and psychological well-being and HRQoL
• participants’ global clinical improvement [Clinical Global Impression of Change (CGI)]
• participants’ satisfaction with treatment
• reductions in health service use
• cost-effectiveness of CBT + SMC compared with SMC alone.

In addition, we sought to characterise:

• patients’ subjective experiences of either the CBT or SMC treatment
• subjective experiences of HCPs (neurologists, psychiatrists and CBT therapists) when delivering SMC or CBT (as relevant)
• treatment fidelity of the manualised CBT treatment and the implications for roll-out in the NHS.

The completion of the CODES trial involved a number of stages in accordance with Medical Research Council guidelines,91 and these included our earlier work.1,81,82 We refined our CBT manual and patient handouts based on our experience from our previous RCT.1 We then trained CBT therapists) to deliver the intervention. We also developed our protocol and materials for neurologists and psychiatrists involved in diagnosing and treating patients in the study, with service user (SU) input into patient materials, and trained the research workers who would assess participants. At an early stage in the project we developed and published both our protocol for this complex intervention study84 and our statistical analysis plan (SAP)92 prior to undertaking any data analysis. We described the participants initially recruited to the study in neurology/specialist epilepsy services and then those participating in the RCT. We undertook an analysis of the fidelity with which the CBT was delivered, and conducted process analyses in the form of in-depth qualitative interviews with patients in both treatment arms. We also explored the experiences of neurologists, psychiatrists and CBT therapists delivering care during the study. We collected and analysed clinical effectiveness and cost-effectiveness follow-up data for both trial arms, with a view to disseminating the results and implications of the analyses once the study had been completed.