COgnitive behavioural therapy vs standardised medical care for adults with Dissociative non-Epileptic Seizures (CODES): a multicentre randomised controlled trial protocol

The overall aim of the CODES Trial is to evaluate the clinical and cost effectiveness of specifically adapted CBT (plus Standardised Medical Care-SMC) in comparison to SMC alone for outpatients with DS, within a pragmatic, multi-centre RCT.

Our primary objective is to evaluate the effectiveness of CBT (plus SMC) compared to SMC alone in reducing DS frequency (our primary outcome) at 12 months post randomisation.

Our secondary objectives are to evaluate:

The study is a multicentre, pragmatic, parallel group randomised controlled trial (RCT) of CBT plus SMC vs SMC alone. After an extended screening phase to identify eligible patients, those patients found to fulfil eligibility criteria and who consent to the RCT are randomised at a 1:1 ratio to either treatment arm. Patients will be assessed at baseline (pre-randomisation) and six months  post randomisation, with final outcome assessed at 12 months post randomisation.

The study will take place in neurology services and neuropsychiatry/mental health services in England, Scotland and Wales. The current clinical neurology services from which we will recruit patients for the screening phase are those secondary and tertiary clinical services seeing patients presenting with seizures that require diagnostic assessment and review; the clinical psychiatry services from which we will subsequently consent patients for the RCT are those that have interest and experience in treating patients with DS. To date we have approval to recruit participants from 25 neurology and 15 psychiatry services.

The target population for this pragmatic trial is adult outpatients with DS which persist following diagnosis by neurologists/epilepsy specialists. Since our study comprises more than one stage to ascertain eligibility, we describe inclusion and exclusion criteria for each stage below. A summary of the participant flow through the study is shown in Fig. 1.

In order to randomise 298 patients (see justification below) we need to identify a substantially larger pool of DS patients in neurology clinics, since in addition to other potential exclusion criteria, approximately 14 % of DS patients will be seizure free three months post diagnosis [29]. In order to allow for ineligibility of patients attending the psychiatry appointments, we envisage needing to recruit 501 patients who meet eligibility in the screening phase of the study.

Adults with DS will receive their diagnosis and an information leaflet on DS from a neurologist/epilepsy specialist.

Participants will be included in the in the screening phase if they:

Participants will be excluded from the screening phase if they:

If patients are deemed eligible for the screening stage of the study they will be given study information and, if interested, will be contacted by phone and/or letter by a research worker. Following further eligibility checks against the above criteria and explanation of the study they will then be consented to the screening phase of the study. In addition to the provision of demographic data, participants’ DS seizure occurrence (in terms of self-reported frequency and severity) will be monitored fortnightly using seizure diary completion. Data will be collected by a research worker by phone, email, text or post, depending on the participant’s preference.

Three months after receiving their diagnosis of DS, those meeting eligibility criteria so far will be reviewed by a liaison psychiatrist or neuropsychiatrist with an interest and expertise in DS, to carry out the final screen for eligibility for the RCT.

Participants will be included in the in the RCT if at the psychiatrist assessment visit and pre-randomisation:

Participants will be excluded from the RCT if, at the psychiatric assessment visit and pre-randomisation, they:

The psychiatrist will go through the diagnosis again, undertake a clinical assessment and provide them with a further, more extensive information booklet on DS. Those patients who have continued to experience DS in the previous eight weeks and who meet the further study criteria listed above will, with their agreement be contacted by a research worker and then, if willing, consent to take part in the trial, undergo a baseline assessment and then be randomised to receive either 12 sessions of CBT (plus a booster session) plus SMC or SMC alone.

Neurologists will be expected to assess the patient in their usual way to determine the nature of the patient’s seizures. In some cases it will be possible to make a secure diagnosis on the basis of the history, witness history and physical assessment. In other cases mobile phone footage, EEG or video EEG may be required to make the diagnosis. Where available and practical an EEG with concurrent video is the most reliable way to make the diagnosis but it is not mandatory. Likewise, neuroimaging is not mandatory and it is anticipated that this will be carried out only according to clinical need. Where video EEG telemetry is not available, we will accept a consensus diagnosis that either involves the agreement between two neurologists in the clinical service dealing with the patient or between the study neurologist and one neurologist within the project team.

Neurologists will not be expected to carry out a standardised psychiatric assessment. However, as with all patients attending a neurological service, if there are clearly recognisable psychiatric risks related to self-harm, harm of others or psychosis the neurologist will refer to the relevant psychiatric services, or ask the patient’s general practitioner (GP) to do so.

All neurologists will be expected to give patients a copy of an information booklet about Dissociative Seizures (see above). In addition to information about DS this will include direction to self-help information (e.g. www.​nonepilepticatta​cks.​info, www.​neurosymptoms.​org).

The neurologist will be expected to provide the following information:

There are other aspects of the initial neurological consultation that may vary according to the patient but could involve: i) explaining that antiepileptic drugs do not help DS, can have serious long-term side effects and should be withdrawn gradually; ii) explaining that talking treatments may be helpful for some people but the evidence is currently uncertain as to whether it is worthwhile; iii) providing explanations to family and friends about the diagnosis, and what to do when the patient has an attack; iv) providing general information about distraction techniques; and v) discussing driving regulations.

We are recommending that the neurologists offer at least one further neurology follow-up visit (although fewer or more are allowable) which may typically cover the following topics: i) overall general review of progress; ii) checking the patient’s understanding of the diagnosis and explaining it again if necessary; iii) supervision of AED withdrawal; iv) management of any comorbid physical conditions; reassessment of major psychiatric risk such as self-harm or psychosis; v) recommendations for antidepressant or anti-anxiety medication prior to the first visit with the psychiatrist if clinically indicated; and vi) completion of forms about driving or from the Department of Work and Pensions if requested by those agencies.

Our primary evaluation of treatment is at 12 months post-randomisation. However, we will also collect outcome measures at 6 months post randomisation and include these in our analyses.

Our primary outcome measure is monthly DS frequency operationalised as seizure occurrence over the previous four weeks. This is a discrete variable that comprises a count of seizures and therefore will reflect all participants’ outcomes, whether they improve or not during the study. Seizure frequency has been used as an outcome measure in other studies of psychological interventions for DS (e.g. [24, 25, 30, 37, 38]). This will be recorded by patients in seizure diaries, as has been done in other studies of psychotherapy for DS. We will collect seizure frequency data from the patients every two weeks by whichever means they find acceptable (diaries, text/phone/online). We will also request an overall self-report estimate of DS frequency in the previous four weeks from participants at baseline and the two follow-up time points, to allow for missing diary data.

Our study seeks to evaluate the effectiveness and cost-effectiveness of CBT+ SMC vs SMC alone in a number of domains. Thus we have selected a range of clinical and economic secondary outcome measures. These are listed in Table 1. In view, however, of the likelihood of psychiatric comorbidities moderating outcome in this patient group, we also include at baseline a screening measure of personality disorder, the Standardised Assessment of Personality Abbreviated Scale, Self Report version (SAPAS-SR) [39, 40] and a structured psychiatric screening instrument (the Mini International Neuropsychiatric Interview; MINI v6.0 [41]); the prerandomisation assessment will also include a single item measure of treatment preference and a measure of expectations of treatment outcome, both of which may also moderate outcome.

We will describe the target population in a number of ways. Demographic information (including age, gender, relationship status, presence/absence of dependants and/or a carer, ethnicity, postcode to evaluate Indices of Multiple Deprivation, employment status, receipt of state benefits, previous medical help sought for a mental health problem, previous diagnosis of epilepsy and current receipt of anti-epileptic drugs) will be collected at the beginning of the screening phase. This will be supplemented with information on self-reported pre-existing medical conditions at the pre-randomisation assessment so that reports of any newly-diagnosed reported medical conditions or symptoms following randomisation can be compared to these in order to identify adverse events. Information on potential adverse events will be collected systematically at 6 and 12 months post randomisation by the research workers. We will also record demographic data and the qualifications and clinical experience of the neurologists, psychiatrists and therapists in the study.

As indicated in Table 1, we will investigate potential mediators of the treatment effect by evaluating responses at baseline and the 6 and 12 month follow-up time points on the Beliefs About Emotions Scale [42], three locally-developed questions to measure avoidance of people, places and activities due to DS, and a single item scale for participants to measure their confidence in the treatment they have received; we will investigate whether a further item rating their confidence in their diagnosis of DS may act as a moderator and/or mediator of the treatment effect.

We based our power calculation on the effect size obtained in our pilot RCT study [24]. This data represented the largest study of this kind to date and importantly included a control group. Our previous pilot trial’s analysis, which controlled for pre-randomisation seizure frequency, reported a standardised effect size for the reduction in seizure frequency under CBT compared to SMC at the end of CBT treatment of Cohen’s d = 0.75 (log scale). Our previous study also detected a more conservative and moderate effect size of Cohen’s d = 0.42 on the log-scale at a follow-up time more directly comparable to the current 12-month post randomisation point (median seizure frequency in the CBT group: 12 at pre-randomisation, 1.5 at follow-up; median frequency in the SMC group: 8 at pre-randomisation, 5 at follow-up). We consider this effect to be clinically important and therefore base our power calculation on this moderate effect size. To detect an effect of d = 0.42 with 90 % power using a 2-sided t-test for log-frequencies at the 5 % significance level, we need 121 participants/group. The sample size must be inflated to allow for potential therapist effects within the CBT group, since each therapist will most likely treat several patients. Based on a typical therapist intraclass correlation coefficient of 0.02 [43] and 15 therapists delivering CBT (average workload 10 patients/therapist), 149 participants are needed per arm to achieve 92.6 % power. We will record pre-randomisation seizure frequencies and include this information as a covariate in the analysis model. This will increase the precision of our future intervention effect estimate. To account for this precision gain and the subsequent reduction in sample size requirement, we can apply a deflation factor to the estimated sample size [44]. We calculated the size of this deflation factor as 0.8367 based on a correlation of r = 0.42 between pre-randomisation and follow-up in DS frequencies [24]. Finally, in our pilot RCT, 7/66 patients were entirely lost at follow-up. We therefore need to inflate the sample size allowing for a more conservative rate of 17 % attrition at 12-month follow-up. Our final randomisation target is 298 participants (149 per arm).

Randomisation will be carried out by the King’s Clinical Trials Unit (KCTU: www.​ctu.​co.​uk) based at the Institute of Psychiatry, Psychology and Neuroscience, using a web-based system. Randomisation takes place once participants have consented to take part in the trial and baseline assessments have been undertaken. The unit of randomisation will be the individual participant. Stratified randomisation with randomly varying block sizes will be used to ensure 1:1 allocation within each of the locations of the neuro/liaison psychiatry clinics from which DS patients are recruited. Allocation will be concealed from the research workers who undertake trial data collection, and also from the trial statistician. The Trial Manager will receive notification of the allocation so that therapists can be informed that they need to arrange to see those participants requiring CBT. We will require research workers to indicate if they have become unblinded and when this occurred, so that outcome assessments can be undertaken by blinded assessors.

Statistical analyses of the primary and secondary outcomes will be conducted after the database has been locked, with no interim analyses. All analyses will adopt the intention to treat principle. Descriptive statistics will describe the characteristics of the initial pool of patients in the screening phase, and also those entering the RCT. For the primary seizure frequency outcome, treatment effectiveness will be assessed by estimating the incidence rate ratio (IRR), comparing the CBT + SMC and SMC arms at the 12-month follow-up time point. Generalised linear mixed modelling (GLMM) assuming a Poisson distribution, a log-link and allowing for possible overdispersion will provide this estimate. The dependent variable of this analysis model will be seizure frequency at 6- and 12 months post randomisation. The explanatory variables will be baseline seizure frequency, randomisation stratifier (liaison/neuropsychiatry clinics), and trial arm, time (6 or 12 months post-randomisation) and trial arm x time interaction. All the available post-randomisation data is modelled simultaneously to gain precision. The model includes the interaction term to allow for time-varying treatment effects. The model also contains participant-varying random intercepts to account for the correlation between the two repeated measures, and will consider including random effects to account for effects of the doctor delivering SMC and therapist-varying intercepts in the CBT arm to account for therapist effects. The models are estimated using maximum likelihood analysis and will allow for missing outcome data under the missing at random (MAR) assumption. The effect of departures from this assumption on results will be assessed using sensitivity analyses [45]. Analyses of secondary outcomes will use a similar approach (for example, continuous outcomes such as quality of life will be analysed using a linear mixed model).

We will undertake a cost-effectiveness analysis from a (i) health and social care and (ii) societal perspective (including lost productivity and informal care). The number and duration of CBT sessions will be centrally recorded and other service utilisation will be recorded with the Client Service Receipt Inventory [46] questionnaire at baseline, and at 6-and 12-month follow-ups. This will include primary and secondary care contacts, social care use, care from family members and medication. We will also record lost work time. Information from Hospital Episode Statistics will also be used to estimate hospital use.

Intervention costs for the CBT sessions will be based on factors such as salaries, overheads, training and supervision. Unit costs for other services will be obtained from national sources [47, 48]. Costs of lost work and informal care will be based on average wage rates but with alternative values used in sensitivity analyses. Costs will be combined with the primary outcome measure (change in DS frequency) and also QALYs generated from the EQ-5D-5L [49] using area-under-the-curve methods. In sensitivity analyses we will use the SF-6D, generated from the SF-12v2 [50], to derive QALYs via an algorithm developed by economists at the University of Sheffield [51]. If the intervention is less expensive and more effective than SMC then it will be ‘dominant’. If it is more expensive and more effective, incremental cost-effectiveness ratios will be constructed to show the extra cost incurred to achieve a one-unit reduction in DS frequency or one extra QALY. Uncertainty around cost-effectiveness estimates will be explored using cost-effectiveness planes (derived from incremental cost-outcome pairs from 1000 bootstrapped resamples) and cost-effectiveness acceptability curves (CEACs - derived using the net benefit approach). In addition to the use of the SF-6D, sensitivity analyses will also be conducted around the costs of the intervention, informal care and lost employment. There has been limited previous research in this area and this trial will provide evidence on the impact of CBT in DS patients over a one-year follow-up.

We will undertake a nested qualitative study to investigate the illness attributions, treatment preferences and experiences of trial participants. In-depth interviews with 20 patients selected across sites who have received CBT + SMC and 10 who have only received SMC will be undertaken to understand what they felt to be beneficial in terms of the interventions, and  what made it easy or more difficult for them to take part. We will also enquire about the extent to which they have been able to implement the content of the CBT, and their views of the content of the therapy. We will include participants who did not attend all the CBT sessions, as well as those who did. The interviews will take place in participants’ homes or will be office-based. The interviews will be digitally recorded and transcribed. Thematic Framework Analysis [52] will be carried out by at least two Research Workers under the supervision of an experienced qualitative researcher and rigour will be increased by them undertaking independent coding, followed by discussion meetings to agree a coding framework, to reduce bias in the interpretation of themes. Triangulation of the findings from the qualitative analysis with the results of the quantitative outcome measures will increase understanding of the trial process and may assist in understanding anomalies in outcomes.

We will collect data on paper source data worksheets. Data will then be entered onto the InferMed MACRO online data entry system, on a study specific database designed and hosted at the KCTU. The system is compliant with Good Clinical Practice and FDA 21 CFR Part 11. Randomisation and post-randomisation information will be accessed directly by the trial statistician using CTU systems.

In addition to a Trial Management Group (TMG), monitoring of the trial activities and recruitment will be undertaken by the Trial Steering Committee (TSC) and the DMEC. The DMEC will monitor recruitment into both the initial screening (phase 1) and randomisation (phase 2) stages of the trial on a monthly basis. Monitoring by the study team, the Trial Manager and the KCTU will ensure that the trial complies with Good Clinical Practice and maintains scientific integrity. The Trial Manager will monitor data collection procedures and undertake source verification checking against the paper data records at regular intervals.

We have developed a protocol for monitoring adverse events during both the screening and RCT phases of the study. While we may become aware of adverse events at any stage of the study, research workers will specifically enquire about these at the 6-and 12-month post randomisation follow-up points. We anticipate a high rate of these events, given the likely psychiatric comorbidity in this patient group, but we will distinguish between serious adverse events that are likely/unlikely to be due to the intervention in the RCT phase of the study. At the end of the study, three independent scrutineers, with clinical experience of working with DS patients, will assess whether events were serious or non-serious.

NRES Committee London-Camberwell St Giles. Reference number 13/LO/1595.

Ongoing (recruiting participants).