Self-Management education for adults with poorly controlled epILEpsy [SMILE (UK)]: a randomised controlled trial

Seizures

Seizure frequency and QoL are linked. PWE are likely to have similar QoL to the general population when seizures are controlled. 23 QoL is inversely associated with the amount of time since last seizure (i.e. ‘seizure recency’). 24 Given this association, even a minor improvement in seizure frequency may have a beneficial effect on health-related quality of life (HRQoL)25 and vice versa.

Seizures are associated with physical consequences that can negatively affect QoL. For instance, sustaining a serious injury during a seizure can be associated with long-lasting repercussions. Thus, minimising seizure frequency is also likely to reduce the ongoing consequences of seizure-related injury.

Antiepileptic drugs

The primary means of treating epilepsy is with AEDs; however, patients may also experience debilitating side effects from their medication. Adverse effects of medication significantly reduce QoL. 26 Apart from seizure frequency, adverse side effects from AEDs may be the biggest predictor of poor QoL in PWE. 24

Failure to comply with treatment regimens (i.e. ‘non-adherence’) also has serious implications for PWE. Non-adherence may result in increased seizure frequency, decreased QoL and missed employment. 27 Results of one AED adherence study showed that depression is associated with medication non-adherence. 28 This suggests a complex association between medication use, QoL and psychosocial well-being. These factors could be targeted by self-management intervention.

Psychological comorbidity

Psychosocial consequences may be as important as the physical consequences of epilepsy. Although treatment success is often reliant on biomedical factors, there are also important psychosocial issues to consider when managing the condition. 21 There is an elevated risk of poor mental health following epilepsy diagnosis. 29 PWE are more likely to experience higher rates of depression,30 anxiety31 and suicidal ideation. 32 Depressive symptoms can diminish QoL33,34 and influence seizure frequency. 35,36 Moreover, some forms of psychological conditions frequently co-occur,36 such as anxiety and depression, having a further impact on seizure frequency. Given the complex relationship between seizure frequency and psychological well-being, interventions that address both treatment targets seem worthwhile.

Stigma and resilience

Stigma is a major factor that influences QoL in PWE. 37,38 Although public perceptions about epilepsy have changed,38 negative beliefs and lack of knowledge continue to perpetuate stigma and discrimination. 39 In a survey of attitudes of the UK general public, participants felt epilepsy was embarrassing, frightening and meant being unable to drive or participate in employment. 40 Participants attributed seizures to stress, intoxication and mental health problems. These findings illustrate some of the prevalent, general attitudes and misconceptions about epilepsy. Despite epilepsy being a neurological condition, psychological and social factors are often attributed to its development,41 which may exacerbate feelings of ‘differentness’ and isolation in PWE.

Scambler and Hopkins42 divided stigma into felt versus enacted. Enacted stigma is discrimination against PWE based on a belief that the attributes of PWE are undesirable or unacceptable. Low employment rates among PWE may partly be a manifestation of enacted stigma. 38 Felt stigma is linked with the fear and shame tied to being ‘epileptic’. 42 Felt stigma is associated with low self-esteem, anxiety and depression. 38 Dilorio et al. 43 suggest that people with frequent seizures may internalise felt stigma and have unhelpful perceptions about their treatment. On the other hand, resilience can help to combat the diminished QoL associated with epilepsy diagnosis. 24 Those involved with epilepsy care can help PWE to combat stigma by giving them information and support strategies for overcoming stigma and discrimination. 38

Mastery and control

The term mastery is used to describe a state of confidence in which a person feels able to independently overcome the challenges with which they are faced. 44 Mastery is used in combination with self-efficacy to reduce stress and develop positive new behaviours. 45 In the literature, the term ‘self-mastery’ has been used interchangeably with self-efficacy. Evidence suggests when a person feels more confident in their ability to cope and manage their illness effectively, they are more likely to put self-management behaviours into practice. 46 Therefore, a sense of mastery and control is important for PWE as they have the potential to have an impact on seizure control and QoL. 47

Age and socioeconomic factors

The effect of epilepsy-related challenges on QoL can be related to the person’s life stage. 48 In a focus group study of UK adolescents, epilepsy diagnosis had an effect on QoL and identity formation. 49 In a study from the USA, seizure frequency was less concerning for older adults and their QoL than maintaining ‘normalcy’ in social and emotional function. 50 Other lifestyle variables such as socioeconomic status are also important for QoL in PWE. For instance, low socioeconomic status is related to an elevated risk of developing epilepsy,51,52 frequent hospital visits53 and attrition from epilepsy care,54 all of which can be associated with QoL. In interpreting the results of an intervention aiming to improve QoL and reduce seizure frequency, consideration of these factors will be important.

Summary of quality of life research

Living with epilepsy can influence QoL in PWE. In the present study, we aim to test the effect of an intervention designed to improve psychological, physical and social consequences of epilepsy. In considering the complexity of the above factors in managing this condition, the benefits of ‘self-management’ have been proposed in the literature. This addition to epilepsy treatment will be discussed in the remainder of this chapter.

What is self-management?

There is no widely accepted description of what self-management means, but the Department of Health has used the 2002 definition from Barlow et al. :55

Self-management refers to the individual’s ability to manage the symptoms, treatment, physical and psychosocial consequences and lifestyle changes inherent in living with a chronic condition.

Barlow et al. 55

The notion that self-management encompasses all of the aspects of living with a chronic condition is summarised in the following:

Self-management is defined as the tasks that individuals must undertake to live with one or more chronic conditions. These tasks include having the confidence to deal with medical management, role management and emotional management of their conditions.

Adams and Corrigan. p. 5756

Corbin and Strauss identified these tasks (i.e. medical, role and emotional management) as the key aspects of disease self-management. 57 It is axiomatic that those living with a long-term condition (LTC) must be self-managing their condition. Thus, interventions that focus on self-management are really directed at supporting people to manage their LTC optimally (i.e. supported self-management interventions).

To facilitate the core tasks of self-management described above, Lorig and Holman58 proposed the acquisition of five core self-management skills including ‘problem-solving, decision-making, appropriate resource utilisation, forming a partnership with a health care provider, and taking necessary actions’. For a schematic representation of how this might be incorporated into an intervention, refer to Appendix 2.

Extensive qualitative research has identified a number of common themes relating to self-management support among individuals living with a wide range of LTCs. These common themes include:

• the need for collaborative relationships with health-care professionals (HCPs)

• the need for support from HCPs regarding information and education

• medication adherence issues

• the need for emotional and peer support

• the need for carers to balance support and independence

• the individuality of each person’s experience of illness

• the importance of psychological support to help with adjustment for some people with LTCs. 59

The first two items on this list are reported in most studies looking at the experience of people living with any LTC. 57

What are self-management interventions?

Self-management interventions teach behaviours that seek to alleviate the consequences of a chronic illness. 60 The objectives are to facilitate patients taking an active role in their own health care by encouraging autonomy and providing accurate information on symptom management. 61

Recent work has attempted to identify the different activities that could be involved in supporting self-management (Box 1). 62 This taxonomy, derived from an expert advisory workshop and a systematic overview of the literature, describes only the potential components of self-management interventions as described in the literature. 59 Efficacy of these components in different LTCs has not been established. How, for example, does self-management support work?

Self-management courses have been successfully implemented in the UK for other LTCs such as diabetes mellitus and arthritis [e.g. Dose Adjustment For Normal Eating (DAFNE),63 Diabetes Education for Self-Management for Ongoing and Newly Diagnosed (DESMOND),64 X-PERT65]. 55 DESMOND, for instance, is a group education programme based on psychosocial theories of learning and self-mastery as a basis for behaviour change. 66 The diabetes mellitus course is generally 1 or 2 days long and consists of non-didactic teaching methods. 66 Increased self-confidence and empowerment have been reported following these targeted self-management education courses. 64,65,67 Thus, previous research suggests that short-term self-management programmes can be of benefit in UK health-care settings and diabetes mellitus courses are now offered to all people with diabetes mellitus in the UK free of charge.

Why self-management for people with epilepsy?

The National Institute for Health and Care Excellence (NICE) recommends that PWE are given adequate, structured information, and empowered to be successful in managing their condition. 68 Past studies indicate that PWE want to know more about how to effectively manage their condition. 69,70

Although PWE might wish to speak with someone other than their doctors about managing their epilepsy;70 many do not know who to ask or lack the confidence to do so. 21 The level of information needed to enhance seizure control and facilitate patient empowerment may require more detail than can be delivered through traditional outpatient consultation. 71 Interventions that aim to improve knowledge, confidence and empowerment may encourage PWE to seek support from a wider range of services to help manage their condition. At present, there are no standardised group education programmes available for PWE in the UK.

A Cochrane review of epilepsy interventions found that there is a potential benefit of self-management groups, although it is essential that further empirical evidence is generated. 72 The review was later updated, indicating that evidence exists to support the benefit of self-management and epilepsy nurse specialist (ENS) involvement in interventions. 73 However, the review stressed the limitations of findings to particular settings and called for improved service models for widespread intervention delivery. In order to test the usefulness of self-management interventions in helping improve QoL for PWE across the UK, further research is needed. 74,75

What evidence exists on self-management for people with epilepsy?

Epilepsy is a condition that requires targeted behaviour and management (such as medication adherence and avoidance of seizure triggers) to alleviate epilepsy-related consequences. For PWE, self-management strategies are defined as the behaviours that minimise frequency and severity of seizures76 and improve overall QoL. 77 Good self-management strategies may also help PWE to potentially feel more confident about health-care decisions and overcome psychosocial consequences of living with epilepsy. 76

At the time of developing the research protocol for this project, there was little evidence on the efficacy of group self-management programmes for PWE. Two key studies on group-based interventions were highlighted during trial design. The interventions that informed development of the SMILE (UK) trial are described below.

Sepulveda Epilepsy Education

The Sepulveda Epilepsy Education (SEE) programme was developed with an aim of meeting the complex psychosocial and educational needs for a wide range of PWE in the USA. 2 Helgeson et al. 2 evaluated the effect of the 2-day SEE intervention on psychological and physical outcome measures (i.e. self-efficacy, AED adherence). With a sample size of n = 38 (18 treatment, 20 waiting list control), limited statistically significant differences were found between groups at follow-up (Table 1). Nevertheless, compared with controls, the treatment group showed a significant decrease in seizure-related fear, hazardous self-management behaviour and misconceptions about epilepsy. The study also showed a statistically significant decrease in self-rated depression in the treatment group (p < 0.0007) immediately after participating in the course; however, this did not persist at their 4-month follow-up. No other statistically significant differences were observed between groups at the last follow-up, 4 months after enrolment (i.e. seizure frequency, self-efficacy).

Modular Service Package for Epilepsy

The Modular Service Package for Epilepsy (MOSES) intervention was originally developed in Germany. 78 After programme development, MOSES was tested in three German-speaking countries in a RCT. 1

The programme was adapted for both in- and out-patient contexts, and the course content was covered over 2 consecutive days. Participants were recruited from 22 specialist outpatient clinics and allocated to either the treatment or the waiting list control group receiving treatment as usual (TAU). Their sample contained 242 adults with epilepsy aged > 16 years with no other major comorbidity (see Table 1).

The final follow-up was carried out at 6 months. At this time point, course attendance had a positive effect on epilepsy-related knowledge, overall coping ability, seizure frequency and medication use. Thus, the authors concluded that MOSES is effective and reduces seizure frequency. They did not find, however, any effect of the intervention on a generic measure of QoL, self-esteem and other aspects of coping with epilepsy (i.e. information seeking). 1 They used the Short Form questionnaire-36 items during the trial, which may not have adequately detected the effect of the education programme. An epilepsy-specific measure of QoL [e.g. Quality Of Life In Epilepsy 89 (QOLIE-89) or Quality Of Life In Epilepsy 31 (QOLIE-31)] may be more sensitive in future research. 1

Overall, the effect of the MOSES programme on seizure frequency was encouraging and MOSES has now been routinely offered across Germany for approximately 15 years. Its success in practice led us to hope that it may be appropriate in UK health settings. Thus, MOSES was selected to trial for use in the UK.

These two seminal studies offered important conclusions about the benefits of self-management education in group settings. However, these studies lacked:

• a sample size adequate to detect significant differences in QoL with enough statistical power while minimising type 1 errors (false positives)

• a long-term follow-up period (e.g. at least 1 year)

• a process evaluation of self-management courses using qualitative methods to gain the patient’s perspective

• a health economics evaluation of the cost-effectiveness of self-management courses

• an assessment of implementation fidelity to determine whether or not the intervention was delivered according to protocol.

The SMILE (UK) programme was designed to address these topics.

Adapting the participant workbook and teaching manual for the United Kingdom

The SMILE (UK) research team obtained course material from the MOSES group. Most content had been translated from German except for personal testimonies from PWE. Personal testimonies were translated into English within the research department. Each SMILE (UK) collaborator edited and revised one module of the patient workbook. Some module titles were modified for ease of understanding.

Changes were made to several sections. For example, the section on ‘Famous PWE’ was modified to include celebrities and public figures that would be recognisable to people living in the UK. The section on ‘Networks’ was also changed to reflect local support network information. Epilepsy Action contributed as subeditors during this part of the intervention adaptation. They reviewed the use of English throughout the patient workbook to reduce the required reading age and improve accessibility for a wide range of audiences. Current information on antiepileptic medication was adapted from information available from the Epilepsy Society. Local regulations (e.g. regarding insurance and driving) were changed to reflect UK legislation.

Overview of the SMILE (UK) intervention

The SMILE (UK) course is a group-based, interactive education programme based on MOSES, developed to provide strategies for individuals living with poorly controlled epilepsy in the UK. 3 SMILE (UK) was designed to be delivered in 16 hours over 2 days.

The number of people per group was intended to range from 6 to 12 participants (including carers or family). All participants assigned to the treatment group were given the option to have a carer or significant other accompany them to the course if they wished. This was encouraged to make the participant feel supported, but also to add another perspective to group conversations (i.e. family or carer perspective). The group number was selected based on the original MOSES research78 and because it might have helped allay anxiety about speaking in front of a large audience and would thus be small enough to share personal experiences comfortably.

Two trained HCPs acted as group facilitators during the course, wherein attendees progressed through set modules and participated in group discussions. A workbook was used throughout the course so participants could become familiar enough with it to be able to use it at home. The programme covered the same nine modules offered by MOSES (i.e. living with epilepsy, epidemiology, basic knowledge, diagnostics, therapy, self-control, prognosis, psychosocial aspects, and network epilepsy)78 renamed for SMILE (UK) (see Table 3).

Teaching manual

A teaching manual was developed based on the MOSES ‘trainer workbook’. 78 For copyright reasons, specific content will not be provided here (please see Acknowledgements for more information). Content can be requested from the MOSES group. The aim of this resource is to guide facilitator teaching of the SMILE (UK) course. In the preface section of the book, it is suggested that course content may be negotiated between participants and facilitators in the future. The teaching manual is divided into two sections.

Participant workbook

Participants were provided with a workbook at the beginning of the SMILE (UK) course and encouraged to use it as a home reference after the course had finished. Each of the nine modules (Table 3) begins with a summary page (about the topic, aims and contents), and includes note-taking space, interactive questions (e.g. what does my doctor need to know about my seizures?) and a final summary of key points (see Acknowledgements for more information about copyright).

Not all workbook material is covered during the 2-day programme, so facilitators suggest that participants refer to the book in their own time and share content with family and friends. At the end of each module, there is a series of questions that encourage further consideration of issues raised and further teaching session. Ample time is provided for sharing problems and solutions.

Course slides

Facilitators are provided with a Microsoft PowerPoint file to show slides during the course. In the original MOSES study, course facilitators were encouraged to use overhead transparencies. We opted for PowerPoint slides based on these transparencies adapted for a UK audience. Slides correspond to participant workbook content and provide supplementary information.

Flip chart

A freestanding flip chart with large paper sheets was set up in full view of course participants. Throughout each day, discussion statements were written on the flip chart and participants were asked to respond during a group discussion. For example, participants were given ‘dot’ stickers on the first morning and these were used throughout the course as a method to compare participants’ views on certain topics. Questions were asked about how participants felt regarding certain statements and they were asked to place a sticker on a scale of responses (Figure 1). The array in the placement of stickers along the scale was then used as a way to initiate discussion on the topic.

FIGURE 1.

Example of reflective exercise using stickers on a flip chart. Participants were asked to place a sticker (represented by a green circle) on the scale according to how they felt.

Facilitator training

The chief investigator and co-investigators of the study underwent several stages in recruiting and training of SMILE (UK) facilitators. LR was responsible for recruitment of facilitators and staff.

The first step involved the engagement of London- and Kent-based ENSs. Requests to hospitals were generally received well, although were dependent on resource availability. Several centres had staff who expressed enthusiasm to teach, but had workloads that prevented their involvement with the study.

After an initial group of prospective facilitators had applied, the trial chief (LR) and a co-investigator (AJN) held interviews to assess each candidate’s suitability for SMILE (UK) facilitator training. A final group of facilitators was selected (see Appendix 5). In total, there were eight female and three male facilitators; seven were EEG technicians and four were ENSs. Four facilitators had a Bachelor of Science-level degree and seven had a Master of Science in epilepsy or epileptology. The average length of experience in epilepsy care for the group was 20 years (range 5–40 years).

SMILE (UK) facilitator training course

Three clinicians/researchers from the MOSES group came to London to conduct a 2-day facilitator training course held on 10 and 11 June 2013 (for topics covered during training, see Appendix 6). The study chief investigator (LR) and a co-investigator (LHG) attended the training course as well.

The MOSES trainers went through the SMILE (UK) course material, showed a facilitator training digital versatile disc (DVD), and went through group-based exercises with trainee facilitators. Attendees were provided with a ‘teaching manual’, which is organised similarly to the patient’s workbooks, with indications on the techniques to use such as mind maps and writing out participant answers on a flip chart or slides. Attendees were able to ask questions on why activities were undertaken in a certain way and learn from the experiences of the MOSES facilitators who were undertaking the course with patients in Germany. Strategies were shared for enhancing participant engagement and understanding, such as focusing on topics in the workbook relevant or of interest to course participants (e.g. if participants already use a seizure diary, move on to another topic), or asking someone to clarify what they mean (e.g. ‘Can you please elaborate on that?’ or ‘Could you give an example?’). When teaching SMILE (UK), facilitators were encouraged not to follow all workbook content sentence by sentence. All sessions during the 2-day facilitator training were video-recorded so that facilitators could review them. On conclusion of the SMILE (UK) training course, all facilitators received a certificate of completion. Throughout the SMILE (UK) study period, there was attrition of facilitators owing to work commitments in their clinics. New facilitators received training by watching the video of the 2-day training session and sitting in on an active SMILE (UK) course.

Methods

Results and discussion

Many themes emerging from the pilot study were similar to those from the process evaluation presented in Chapter 9. For these common themes, data will be presented together. In the section that follows, we will present the results relevant to the intervention that helped us to shape the SMILE (UK) course for the RCT. Full characteristics of volunteers from the pilot are also presented in Chapter 9. Briefly, six females and four males participated, aged from 21 to 60 years, having had epilepsy for 8 to 52 years.

Stage 1

Individuals having attended a neurology clinic were sent a letter from the local clinical PI along with information about the study in a patient information sheet (see Appendix 9). At this stage, patients had the chance to opt out from the next stage of the study. They had a 3-week period to return their opt-out slip in a pre-paid envelope. As this stage did not involve screening medical notes, patients contacted may not have necessarily met the eligibility requirements. The only requirement at this stage was that they had attended the neurology clinic within the past year.

Stage 2

At the second stage of recruitment, medical notes were screened by clinicians at the local hospital and a second invitation was sent to eligible participants along with the patient information sheet. Once again, patients had the opportunity to opt out from further contact from the research team. Again, at this stage, patients had 3 weeks to return their opt-out slip. Patients who did not opt out of either stage were contacted by the research team to invite them to enrol in the study.

Stage 3

At this point of contact, the research team telephoned eligible participants who had not opted out to invite them to join the study. Research workers confirmed eligibility criteria of interested individuals and arranged a face-to-face meeting to explain the study in more detail, obtain informed consent and conduct a baseline assessment. Participants gave written consent themselves. All meetings were arranged at participants’ homes or public locations convenient for them.

Eligibility criteria

Inclusion criteria:

• confirmed epilepsy diagnosis (all epilepsy-related conditions and seizure types included)

• current AED prescription

• aged ≥ 16 years

• ability to give informed consent, contribute during groups and complete questionnaires in English

• have experienced two or more seizures in past 12 months (self-reported).

Exclusion criteria:

• only having seizures that are psychogenic or non-epileptic in origin

• only experiencing seizure related to acute neurological illness or substance misuse

• diagnosis of a serious psychiatric disorder or terminal illness

• participation in other epilepsy-related research.

Informed consent

Written informed consent was obtained during a face-to-face meeting between the participant and a research worker, arranged either at their home or somewhere convenient for them. The study was explained in detail and participants were given the opportunity to ask questions. Once the consent form was signed (see Appendix 10), baseline data were collected.

Primary outcome measure

The primary outcome measure used in the SMILE (UK) study is the QOLIE-31-P score at 12 months post randomisation. It is a short, easy-to-use measure based on the original QOLIE-31. 80 The QOLIE-31-P varies slightly from the original QOLIE-31 with an addition of one item at the end of each subscale, asking the participant to rate the degree of ‘distress’ caused by that particular topic. The measure was completed at baseline and at 6- and 12-month follow-up (Table 4). The questionnaire provides a total QoL score and contains seven individual subscales: (1) energy level (four items), (2) emotional wellbeing (five items), (3) social functioning (five items), (4) cognitive function (six items), (5) medication effects (two items), (6) seizure worry (five items) and (7) overall QoL (two items). 80 The scale asks participants to reflect on how they felt over the past 4 weeks. Scores range from 0 to 100, for which higher rank indicates better overall QoL. Internal consistency reliability (Cronbach’s alpha) has previously been reported as α = 0.79–0.85. 80

Secondary outcome measures

A number of measures were used to assess secondary outcomes (see Table 4).

Facilitators

Demographic data regarding course facilitators were collected and recorded in a separate electronic database on MACRO. Data included initials, gender, current occupational role, years in current role, highest level of education reached, specialist qualifications, size of current caseload, prior experience in facilitating patient groups, and pay scale grade point. Each facilitator was given a randomised therapist identification number to ensure anonymity and this number was recorded on treatment attendance logs for each participant.

Primary outcome analysis

Outcome data are presented, when appropriate, as means with SD, medians with 25–75% IQR and minimum to maximum ranges. All analyses of outcome measures followed the ITT principle as we aimed to evaluate the effectiveness of the SMILE (UK) intervention. In other words, participants were analysed in the groups to which they were allocated, irrespective of intervention receipt. The primary outcome was total QOLIE-31-P score at 12 months post randomisation.

Prior to the statistician becoming unblinded, a binary variable was created to represent participants who received full intervention [in the SMILE (UK) arm only] and a second variable represented whether primary outcome data were present or missing. An independent statistician tested whether or not intervention receipt was predictive of missing primary outcome data using a chi-squared test. In addition, the relationship between baseline variables and missing primary outcome data was assessed using logistic regressions, which looked at the predictive effects of each baseline variable separately. A number of baseline variables were found to be predictive of missingness (see next paragraph) and treatment receipt also predicted missingness in the intervention arm; therefore, multiple imputation was used to produce inferences that are valid under such a missing at random data-generating process. The number (proportion) of missing QOLIE-31-P data was 4 (1.0%) at baseline, 91 (22.5%) at the 6-month follow-up and 73 (18.1%) at the 12-month follow-up.

Multiple imputation consists of two steps: an imputation step and an analysis step. As a guiding principle, all variables included in the analysis model need to be included in the imputation step; further variables can be included in the imputation step only. The first step was to impute missing values using multivariate imputation by chained equations (MICE). All predictors of dropout (gender, education, comorbidities and treatment receipt within active arm) were included to allow them to drive missingness. All variables that formed part of the analysis model were included: trial arm [SMILE (UK) course or TAU], educational group within SMILE (UK), the randomisation stratified by trial centre (the site from which participants were recruited) and baseline QOLIE-31-P. The MICE model was used to impute QOLIE-31-P at baseline, 6 and 12 months simultaneously (i.e. QoL at different time points was allowed to predict QoL missingness and can provide extra precision if predictive of QoL values). A total of 100 imputations were run. Once the multiple imputation was complete, the Stata^® (StataCorp LP, College Station, TX, USA) command ‘mi estimate’ was used as a prefix so that the imputed data sets were used for the analysis (i.e. Rubin’s rules were used to generate inferences based on multiple imputed data sets). The analysis model was a linear mixed-effects model: total QOLIE-31-P score at 12 months post randomisation was the dependent variable. Baseline QOLIE-31-P score, trial centre and treatment arm were included as fixed effects and educational group was included as a random effect within the intervention arm only. The random effects were added to account for potential clustering as a result of participants attending the same educational group in the SMILE (UK) arm.

Secondary outcome analyses

There were seven secondary outcomes that were measured as continuous variables: (1) HADS-anxiety (HADS-A), (2) HADS-depression (HADS-D), (3) self-mastery and control, (4) impact of epilepsy, (5) medication adherence, (6) medication AEs, and (7) stigma of epilepsy. All of these were analysed in the same way as QOLIE-31-P (i.e. using MICE followed by linear mixed model for the respective secondary outcome variable). The only terms that were amended between each analysis were the relevant outcome measures.

There were two secondary outcomes that were not continuous variables and they had to be analysed slightly differently to reflect their distributions: (1) seizure recency and (2) seizure frequency.

Seizure recency was constructed from the date of last seizure and counts the ‘number of days since last recalled seizure’. In addition to this variable allowing for positive values only, it has a maximum. As the last recalled seizure can, by definition, involve recall bias (this was assumed at 6 months), and as the eligibility criteria for the RCT stated that participants had to have at least two seizures in the last year, seizure recency maximums were defined as follows: 548 days at baseline (1.5 years); 730 days at 6-month follow-up (2 years); and 913 days at 12-month follow-up (2.5 years). Therefore, in order to analyse seizure recency, the variable was transformed onto a continuous scale using the logit transformation. This transformation converted days into a proportion and then mapped the odds onto the continuous scale via applying the logarithm. The logit-transformed recency variable was then analysed in the same way as the other continuous variables.

Seizure frequency was collected on two different scales: Baker et al. 99 and Thapar et al. 35 The seizure frequency variable as measured by the Baker scale was analysed as a binary outcome: less than one seizure per month versus one or more seizure per month. A similar analysis approach was used as above, except MICE imputed missing outcome values by assuming a logistic regression and a logistic mixed-effects model was used as the analysis model. Similarly, seizure frequency as measured by the Thapar scale was analysed as an ordered categorical outcome: 0–3 seizures, 4–6 seizures, 7–9 seizures or ≥ 10 seizures. In addition, the MICE and mixed-effects models used ordinal logistic regression.

Intervention receipt

The primary analysis was modelled as an ITT analysis. Owing to non-receipt of the intervention [attending all four SMILE (UK) sessions] being fairly high (38.5%), a complier average causal effect (CACE) analysis was performed to assess the efficacy of the intervention in the presence of non-attendance. For this purpose, binary intervention receipt of SMILE (UK) intervention was defined as attending all sessions (i.e. receiving full treatment).

An instrumental-variables approach was used to estimate CACE for the primary outcome. Specifically, a two-stage least squares estimate of an instrumental-variables regression was used (‘ivregress 2sls’ in Stata). The estimator obtained by this approach is equivalent to that obtained by, first, regressing the treatment received variable on the treatment allocated variable and baseline covariates and, second, controlling for the saved residuals, regressing QOLIE-31-P on the treatment received variable and baseline covariates.

Statistical analysis was performed using Stata version 14.

Recruitment

Participants were approached to take part in the process evaluation. They were largely recruited from the TAU arm but also included some from the intervention arm. Participants were purposefully chosen to ensure a variety in age, gender, ethnicity and severity of epilepsy was represented.

Interview procedure

Semistructured interviews were held with topics selected after completing the external pilot study, and in collaboration with research staff that held these interviews. All participants were interviewed face to face. The interviews were undertaken by two researchers. Interview times were variable, ranging from 25 to 40 minutes each.

Analysis

All interviews were recorded and then transcribed verbatim by an external third party. Transcripts were checked for accuracy by the researchers who conducted the interviews. Sample transcripts were checked by two collaborators. Researchers then undertook a line-by-line coding approach with initial codes noted in the margins of the transcripts and then grouped into broader themes. This process involved regular discussion between the researchers and study supervisors.

Full methods are presented in Chapter 9.

Recruitment

The process of generating patient lists and sending out study information occurred from September 2013 until May 2015. At the start of trial recruitment, lists were created with the names of patients seen at eight different epilepsy clinics within the past year (≥ 4000 patients in total; Figure 2). Letters were sent to these patients explaining the study and giving them the chance of a first opt-out of further contact. During this first stage of opt-outs, 262 people sent in opt-out slips or were unable to be reached by post.

FIGURE 2.

Recruitment pathway for SMILE (UK). Reproduced with permission from Ridsdale et al. 114 2018. This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by-nc/4.0/.

Reproduced with permission from Ridsdale et al. 114 2018. This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by-nc/4.0/.

Recruitment sites were opened progressively through the study and recruitment overlapped between sites. For this reason, medical notes were screened by instalments for those not opting out, resulting in 2854 patient notes screened. When medical notes had been screened by clinical staff, 569 patients were identified as not eligible. A second letter was then sent to the remaining patients (n = 2285) with another option to opt out of the study. During this second stage of opt-outs, 299 patients either returned their slips or could not be reached by post.

In summary, the two opt-out stages resulted in 529 opt-out slips being received (reasons provided in Table 5). Those who had not already opted out were contacted by telephone in order to assess their eligibility and interest in the study.

Recruitment progressed in sequence at each of the centres involved with the study (Table 6). Owing to overlapping 12-month follow-ups with continuous recruitment, we were able to adjust recruitment targets in light of a higher than anticipated retention rate. Recruitment ultimately finished on 6 August 2015, with an enrolment rate of 37% (n = 407; eligible patients, n = 1088).

Enrolment

Of those contacted by research workers (n = 1458), 46.7% declined to participate in the study, 27.9% consented and 25.4% were ineligible. The process of enrolment (i.e. participant consent and baseline assessment) began in December 2013 at KCH and concluded at the beginning of August 2015 at St George’s Hospital.

Sample size

The recruitment target was 426 participants with a 12-month retention rate of 75% (i.e. effective sample size at 12 months of 320 participants). Owing to higher follow-up rates than expected (82% at the time), we could stop recruitment after 407 participants were consented into the study and 404 randomised. This decision was made with the approval of statisticians involved with the trial (SL, EJR) and agreed on by study collaborators and the TSC.

Randomisation into trial arms was carried out by the King’s Clinical Trials Unit as explained in Chapter 3. Early in the recruitment phase, the trial manager noticed an imbalance in the group allocation. This was attributable to a software error and was rectified by the Clinical Trials Unit. For this reason, the number of PWE was slightly unbalanced between the two groups (Figure 3).

FIGURE 3.

Retention through SMILE (UK) for intervention and control groups. Reproduced with permission from Ridsdale et al. 114 2018. This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by-nc/4.0/.

Reproduced with permission from Ridsdale et al. 114 2018. This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by-nc/4.0/.

Retention and follow-up

We enrolled 404 participants into the study and 331 (82%) completed follow-up assessments at 12 months (see Figure 3). Withdrawal was defined as withdrawing from any further data collection. The most common reason for withdrawal from the trial was being unable to contact the participant. Some participants chose to withdraw as a result of time constraints, not feeling well enough to complete questionnaires or a change in personal circumstances. One participant no longer had a diagnosis of epilepsy at the 12-month follow-up and was withdrawn.

Follow-up retention rates were 77% for 6 months and 82% for 12 months. Final 6-month follow-ups were completed in February 2016. A non-response at 6 months did not lead to an automatic withdrawal from the study. Some participants who did not provide follow-up data at 6 months post randomisation subsequently provided 12-month follow-up data. Final 12-month follow-ups were completed in July 2016.

Of note, the withdrawals at 6 months were initiated by participants only. If participants could not be contacted at the 6-month follow-up, then they remained in the study and some could be reached at the 12-month follow-up. This explains why the completion rate at 12 months is higher than at 6 months.

Adverse events

The AEs were recorded at 6 months over the telephone after postal questionnaires were received and at the 12-month follow-ups. A total of 41 participants reported AEs over the course of the study. Between them, 63 AEs were reported (Table 7). Fourteen were considered mild, 21 were of moderate intensity, 26 were severe and two new psychological diagnoses were of unknown intensity. In the intervention group, 33 AEs were reported by 25 participants, while 16 participants in the control group reported 30 AEs. Twenty-two AEs were considered SAEs, but no AE or SAE was considered to be related to the intervention.

Contamination

Three participants in the control group attended the intervention early, rather than after the 12-month follow-up. Two were invited in error and one came to the course without having been invited by the research team. These participants were analysed in their original TAU group in order to adhere to the ITT principle. They are not included in the course attendance numbers below.

Researcher unblinding

As participants knew their treatment group, there was a high risk of research workers becoming unblinded during follow-up assessments. Following every 12-month follow-up or withdrawal, researchers completed the ‘Research Worker Treatment Guess’ form. For the 331 participants completing the 12-month follow-up, research workers reported being unblinded for 56 assessments (16.9%). When the blinding remained, most guesses were that the participant was in the TAU group (73.7%) and this was mostly a random guess (random guess 73.7% vs. educated guess 9.4%).

Course attendance

Eighteen SMILE (UK) intervention courses were held between December 2013 and August 2015 (Table 8). The majority were held at KCL (15 courses), which is a university-based research centre next to a teaching hospital (KCH) in south London. The remainder were held at the NHNN (two courses) and the University Hospital Lewisham (one course).

Course attendance was recorded to determine receipt of intervention. To do this, we divided the course into four sessions, and a session was defined as a half-day. Of the 205 participants who were randomised to the intervention arm, 74% (n = 151) attended at least one session and the majority [62% (n = 126)] completed all four sessions of the course (Table 9). Participants were invited up to three times to attend a course. Overall, 80 reasons for missing sessions were given by 79 participants who did not attend all four sessions. Reasons for non-attendance were typically as a result of feeling ill or prior work commitments (Table 10). We hypothesised that often lengthy travel to the venue and course length may also have been factors in course attendance rates. Seizures occurring during the course were recorded. There were six observed seizures throughout the study period lasting between 2 and 20 minutes. Of these, three were instances of non-epileptic seizures occurring in two participants, which were observed and documented by the ENS facilitating the course. One participant had been aware of having these seizures in the past. It was possible to document these instances of non-epileptic seizures as a result of the extensive experience of the ENS facilitators (> 15 years each).

Effect of intervention receipt on retention

We assessed whether or not attending the SMILE (UK) course was associated with retention in the intervention group (Table 11). Full receipt of the intervention was defined as having attended both days of the SMILE (UK) course/four sessions (morning and afternoon of both days). Not having received the full intervention was defined as having attended 0–3 sessions. Extent of receipt of the intervention was associated with the likelihood of remaining in the study: 120 out of the 126 participants (95.2%) who received the full intervention completed the 12-month assessment, compared with 43 of the 79 participants (54.4%) who did not receive the full intervention. This informed the need to use multiple imputation in the inferential analysis of outcome measures.

Study sample characteristics

Baseline demographics of the study group are presented in Table 12. Overall, the group had an average age of 41.7 years, ranging from 16 to 85 years. There was a slightly higher proportion of females taking part (54.2%) and the majority of participants were of white ethnicity (75.2%). The IMD score, measuring levels of deprivation according to participants’ postcodes,111 indicated that the majority of our group (60.7%) lived in areas of high deprivation. The group was highly educated, with approximately 50% having post-secondary level qualifications [beyond General Certificate of Secondary Education (GCSE) level]. Despite the high level of education, only 21.8% of the group were employed full time, 13.1% were employed part time and 6.9% were self-employed. To better evaluate the proportion of participants not in work, we grouped the employment category into a binary factor looking at participants aged < 65 years. This showed that half of the group of working age were not employed. Roughly three-quarters of the group lived with others and about 40% were with a partner.

Clinical details at baseline (see Table 12) showed a median 18 years since epilepsy diagnosis. Almost half of the sample reported experiencing a comorbid health condition (45.9%), including 13.2% with a current mental health diagnosis (e.g. depression, anxiety).

Evaluation of Quality Of Life In Epilepsy 31-P and subdomains

We found strong pairwise correlations (using Pearson’s r) between QOLIE-31-P subscales and the total score (0.63–0.71), suggesting that each subscale has a reasonable, but not overly strong, association with the total score. Higher scores on the QOLIE-31-P reflect better QoL. Correlations between subdomains were lower, indicating that each scale is measuring a sufficiently different topic in this sample (Table 13).

We then looked specifically at the relationship between QOLIE-31-P subscales and the HADS measure (Table 14). As shown in Table 14, mood and seizure anxiety were most strongly negatively associated with the anxiety subscale (HADS-A). Mood, energy and daily activity scores were most strongly negatively associated with the depression subscale (HADS-D). This finding suggests that greater psychological distress is associated with diminished QoL status.

Primary outcome: quality of life

The primary outcome measure was QOLIE-31-P measured at 12 months after randomisation to the study. This measure was also reported at 6 months post randomisation. The observed QoL outcomes are summarised in Table 16. This table shows that the two domains with consistently the lowest scores across all time points were energy and cognition. The highest scored domains were mood, daily activity and medication effects. As many published studies115–117 use QOLIE-31 without the patient-specific weighting, we calculated the total score according to the QOLIE-31 specifications for comparative purposes. These scores are presented in Table 16. Similarly to QOLIE-31-P, the observed total scores were slightly higher after 1 year in both groups. The total QOLIE-31 score improved in the intervention group by 2.7 and in the control group by 2.5 at 12 months.

Importantly, the post-randomisation QoL scores in all domains were similar in the two trial arms (see Table 16). We found no statistically significant differences between the trial arms at 6 months (p = 0.195) or 12 months (p = 0.564) (Table 17).

In total, 129 (32%) participants enrolled in the study received full treatment by attending all four SMILE (UK) course sessions. This consisted of 126 randomised participants who were allocated to SMILE (UK) and three participants who were allocated to TAU alone. The CACE efficacy estimate was –1.28 (p = 0.528), which was in the same direction as the ITT effectiveness estimate of –0.72 (p = 0.564). Neither result was statistically significant.

Secondary outcomes

Instrument development

The process of developing the implementation fidelity tool was based on past research. 118,122 As MOSES [and, thus, SMILE (UK)] was not developed according to a behaviour change model, we began by determining which modules of the course were likely to influence behaviour change. Eight out of nine modules were identified by study co-investigators; these were further reduced after reviewing suggestions from participants in the external pilot study. 82 Participants in the external pilot identified modules 3 (basic knowledge), 4 (diagnosis) and 6 (self-control) as those from which they had learned the most. Module 8 was also identified as helpful; however, this module is subject to participant input and, thus, is highly variable between course sessions. For this reason, module 8 was excluded from the fidelity analysis. The three modules ultimately selected for fidelity assessment were content heavy. Discussion points included knowledge about epilepsy, the science behind clinical diagnosis, treatment options and seizure auras or triggers.

Adherence

In order to assess facilitator adherence, we created a checklist based on facilitator workbook content. Each checklist item could be awarded a score from 0 to 2 (0, content undelivered; 1, partial delivery; and 2, full content delivery). In total, there were five checklist items for module 6 and six checklist items for modules 3 and 4 (maximum score of 2/2 per item).

Competence

We selected four criteria for evaluating facilitator competence: ‘group interaction’, ‘overall impression’, ‘didacticism’ and ‘trainer techniques’. Scores for ‘group interaction’ were allocated based on the number of participants who interacted with the facilitator (0, one participant was speaking; 1, two or three participants had spoken; or 2, four or more had participated in a discussion during the group session). As with past research,118 ‘overall impression’ was assessed to understand how well the modules were delivered. Scores ranged from 1 to 4 (1, poor performance – the session consisted of didactic teaching strategies with few instances of interaction; 2, average performance – there was some interaction and participant input, but not a great deal of cohesiveness 3, good performance with some didactic-style teaching and considerable group participation and cohesiveness; and 4, indicates excellent performance – minor instances of didactic teaching and major instances of group participation).

The total amount of speaking time was used as a proxy for ‘didacticism’ during recorded sessions. An annotation programme called ELAN (Max Planck Institute for Psycholinguistics, The Language Archive, Nijmegen, the Netherlands)123,124 was used to record instances of facilitator speech in number of seconds. Filler words used by facilitators such as ‘yeah’ and ‘oh’ were not scored. Total speech time was divided by the length of the module and was expressed as a percentage.

Finally, throughout each of the module recordings, raters also scored the number of the times that a facilitator used teaching resources such as a flip chart or PowerPoint slide.

Instrument testing

We conducted a test of the fidelity instrument to ensure its feasibility, with two members of the research team assessing a course delivered to the control group. For the fidelity assessment of the five courses, two independent raters were recruited to reduce the possibility of bias. The raters were trained to use the instrument by evaluating a control group offering of the course. A scoring guide was also developed in order to minimise the possibility of scoring error or inconsistency. This included a list of specific topics that the facilitator would need to cover during each module to receive a maximum score. For instance, in module 3 a score of 2 out of 2 would be awarded for the section on ‘seizure types’ if the following categories were described: generalised tonic–clonic, absence, complex partial, simple partial and myoclonic.

Statistical analysis

Several approaches were employed to measure inter-rater reliability. A weighted kappa statistic was used with ordinal scores derived from checklist items. Intraclass coefficients were used with continuous scores that were derived from didactic speech time measurements. The per cent agreement was calculated for items on the instrument checklists in order to assess the frequency of raters giving the same scores. Simple regression analysis was used to examine the associations between scores for different measures.

Fidelity measure evaluation

In total, 15 recorded sessions were analysed (i.e. five courses each with three modules) and 85 items were scored. The results were charted in a table then assessed for inter-rater reliability. With a weighted kappa score of 0.67, there was substantial agreement with allocated ratings. 125 Percent agreement was also high, at 81.2%.

A similar approach was taken in assessing the competence measure. There was also substantial agreement between raters with a weighted kappa statistic of 0.65 and percentage of agreement at 60.0%. The intraclass coefficient for didacticism was high at 0.97 (p < 0.0001), which indicates a high degree of reproducibility. Most sessions received the maximum score of 2 for ‘group interaction’, and thus this item was not assessed for inter-rater agreement.

SMILE (UK) course evaluation

Limitations of the fidelity assessment

Only three out of the nine SMILE (UK) modules were evaluated. The three selected modules involve more teaching than the others and, thus, some level of didacticism is expected. However, despite this, the facilitators still maintained a high level of group interaction. The modules not included in the fidelity assessment were participant led and would not have been suitable for a structured evaluation. Yet, they may be important in behaviour change by increasing self-confidence. 130,131 Because some modules were not assessed, it is possible that some overlapping content was touched on in other parts of the course. Thus, an item could receive a score of 0 in our evaluation but have been delivered in another session. Our group interaction score was based on a specific number of participants interacting, but this could also have been represented by a percentage of the group, which would better account for varying group sizes. Owing to constraints, we evaluated approximately 25% of the courses delivered. We were unable to assess the same facilitator across multiple courses and thus we are unable to monitor how consistent they were throughout the study.

Perspective

The primary perspective of the economic evaluation was the NHS/Personal Social Services perspective, which, for decision purposes, is generally preferred by NICE. Other resources relevant to a wider societal perspective, such as informal care and productivity loss (because of time off work) were included in the secondary analyses (societal perspective).

Data collection: service use and costs

An adapted version of the CSRI104 was administered retrospectively to collect self-reported resource utilisation for the 12 months preceding data collection at baseline and 12-month follow-up interviews. This measure is used to collect the range of services and support accessed by study participants. The CSRI has been successfully used in a variety of adult mental health populations since its design in 1985. Patients in this study have different health needs and are likely to require varying health and social services, hence the need to adapt the CSRI to ensure we capture services relevant to PWE. The CSRI was used to record epilepsy-specific hospital services and community-based health and social care services, medication, productivity losses as a result of illness, and help with usual activities provided by family and friends. Data on the number and duration of SMILE (UK) and TAU sessions were centrally recorded as part of the RCT.

Costs are reported in Great British pounds at 2014–15 prices and, given the 1-year time horizon, discounting was deemed unnecessary. Service use collected using the CSRI was combined with nationally applicable unit costs to estimate total costs for each participant. 134,135 Productivity losses as a result of epilepsy were derived based on the human capital approach, which combines days off work with average UK wage rate. Family and friends are not generally reimbursed for their support to patients but there is still an opportunity cost to this time. The cost of this unpaid care was also estimated based on the average UK wage rate for adults.

Intervention costs were estimated based on staff time (facilitator) needed to deliver SMILE (UK) including other non-staff costs, such as training and manuals. These were applied to group-session attendance data to work out the cost per participant.

Data collection: quality of life

Health-related QoL was estimated using the EQ-5D-5L,136 a widely used standardised method for assessing health-related quality. The EQ-5D-5L is a brief self-reported preference-based measure of health that considers five dimensions (mobility, self-care, usual activities, pain/discomfort and anxiety/depression), each consisting of five levels of functioning (e.g. no pain, slight pain, moderate pain, severe pain and extreme pain). This measure produces a possible 3125 distinct health states ranging from 11111 (full health) to 55555 (worst). Utility scores for each participant, between 0 (worse health) and 1 (full health)136,137 were produced by combining UK population value sets with these health states. QALYs were derived from the transformed EQ-5D-5L utility scores using the area under the curve method,138 which is the preferred measure for HRQoL in UK Health Technology Assessment research. The second part of the measure [EuroQoL-5 Dimensions (EQ-5D) VAS] takes the form of a VAS, in which the participant is required to rate their health from 0 (worst imaginable health state) to 100 (best imaginable state). It shows the respondent’s self-perceived HRQoL on a scale with endpoints labelled ‘worst’ at 0 and ‘best’ at 100.

Cost-effectiveness analysis

Stata (version 14) was used to analyse data. Data were assessed two ways: complete cases (i.e. only including participants with complete service use and QoL data) and on an ITT basis (i.e. according to the group to which they were randomised regardless of intervention receipt). Missing costs and outcome data were imputed using a single imputation method based on linear extrapolation, adjusted for baseline costs, EQ-5D-5L utility scores and QOLIE-31-P score. Other epilepsy-related variables, such as seizure frequency, were obtained by non-parametric bootstrap regressions to account for the non-normal distribution commonly found in cost data. 139

Costs and outcomes were compared between the two arms at baseline and 12-month follow-up and are presented as mean values with SDs and 95% confidence intervals (CIs). Mean differences and 95% CIs were obtained by non-parametric bootstrap regressions (1000 replications) to account for non-normal distribution often found in cost data. 139

Cost-effectiveness and cost-utility

Cost-effectiveness was assessed based on the health and social care (NHS) perspective, as well as the societal perspective, by combining the costs with data on the primary outcome (QOLIE-31-P) measure at 12 months.

Cost–utility was explored by combining total costs with QALYs, derived from EQ-5D-5L data. Assuming the intervention produced better outcomes and lower costs, it would be ‘dominant’. However, if it resulted in better outcomes but higher costs, then incremental cost-effectiveness ratios (ICERs) would be estimated to show the extra cost incurred for a 1-unit improvement on the QOLIE-31-P or one extra QALY (both at 12 months). There is no need to calculate ICERs for any combination in which one group has lower costs and better outcomes as it is then considered to ‘dominate’ the other group.

To evaluate the uncertainty around point estimates, cost-effectiveness planes (CEPs) and cost-effectiveness acceptability curves (CEACs) were created. These curves are an alternative to CIs around ICERs. Non-parametric bootstrapping was used to create a joint distribution of incremental costs and outcomes to explore the probability of SMILE (UK) or TAU being cost-effective, subject to varying willingness-to-pay (WTP) values attached to a unit improvement in the QOLIE-31-P or one extra QALY gained. One thousand cost–outcome combinations were plotted on a CEP, where the y-axis depicts the additional costs of the intervention compared with TAU, and the x-axis represents a difference in the primary outcome between the intervention and control groups.

Results were interpreted using CEACs140 to show the probability of the intervention being the cost-effective option for a range of different values placed on an improvement in outcome. The range of values for QALYs was £0 to £80,000; this includes the threshold recommended by the UK’s NIHR and NICE when judging the cost-effectiveness of a health technology. The range for improvements on the QOLIE-31-P was chosen such that values at which the intervention or TAU has a 50% and 70% and 90% likelihood of being cost-effective were identified.

A series of net benefits [calculated using the formula net monetary benefit = (ΔE) × λ – (ΔC)] were calculated for each individual for a variety of values (£0–80,000 for WTP for an additional QALY). Additional gain in outcome (ΔE) was multiplied by the ceiling ratio (λ) defined by the decision-maker’s WTP for an additional unit of health outcome, and the difference in costs (ΔC) subtracted. After calculating net benefits for each participant for each value of WTP, coefficients of differences in net benefits between the trial arms were obtained through a series of bootstrapped linear regressions (1000 repetitions), which included the same covariates used for comparisons of outcomes in the primary economic analyses. The resulting coefficients were then examined to calculate the proportion of times that the intervention group had a greater net benefit than the control group for each value of WTP. These proportions were then plotted to generate CEACs for all cost–outcome combinations.

Response rates

Resource use and QoL data were available for all participants at baseline and at the 12-month follow-up. At the end of the trial, 18% of participants did not complete the final assessment (Table 21).

Service use

Participants from both groups reported contact with a variety of health professionals, including use of health and social care services and the use of AEDs (Table 22). The service use between the two trial arms was similar. As patients were recruited through their neurologist, this explains the high proportion of participants in both groups who reported contact with a neurology consultant. At baseline, resource utilisation was similar between the two trial arms with GPs being the most frequently reported contact, at an average of three visits in the previous 12 months. Contact with ENS and mental-health services (counsellors and psychologists) was also reported by several participants, with a mean contact (number of visits) of 1 and 2, respectively, for both groups. Contact with community health professionals remained relatively similar at follow-up, with notable variation in the ‘other’ category. Although a small proportion of patients (in both groups) reported contact with home workers, activity centres with private nursing and other community services, the number and duration of contacts reported was quite high.

The use of hospital services was also similar between the two groups at baseline, except for the use of outpatient care, which was reported as higher in the SMILE (UK) group (but the number of contacts is reported as almost identical between the two groups). At the 12-month follow-up, the proportion of patients reporting use of hospital services had reduced for both groups. The percentage of participants who reported informal care was low at baseline, but those who did received substantial help from family and friends, especially for the ‘on call’ category. The number of hours spent on call per week by friends and family had increased by a large amount at follow-up for both groups. Productivity loss was reported by almost one-quarter of participants in both groups. Although remaining low at 12 months, the mean number of days taken off work reported by the SMILE (UK) group was approximately five times higher than in the control group (16.3 days vs. 3.19 days).

Costs

Quality Of Life In Epilepsy 31-P

Cost-effectiveness results based on the QOLIE-31-P are similar to those emanating from QALYs. Results from the complete-case analysis show the majority (36%) of scatter points in the north-west quadrant, indicating that the SMILE (UK) group costs more and has lower QoL than the TAU group (Figure 13). For the ITT analysis, findings are slightly different from the complete-case analysis, with most scatter points located in the south-west quadrant (i.e. lower costs and lower QoL) (Figure 14).

FIGURE 13.

Cost-effectiveness plane of complete cases (societal perspective) using QOLIE-31-P. Scatterplot of bootstrapped resampling of joint incremental costs and QOLIE-31-P score comparing SMILE (UK) with TAU, adjusted for baseline costs and QOLIE-31-P.

FIGURE 14.

Cost-effectiveness plane of ITT (NHS perspective) using QOLIE-31-P. Scatterplot of bootstrapped resampling of joint incremental costs and QOLIE-31-P score comparing SMILE (UK) with TAU, adjusted for baseline costs and QOLIE-31-P.

Methods

Interview procedure

Semistructured interviews were held with topics selected after completing the external pilot study, and in collaboration with research staff that held these interviews. Although the pilot study topics focused on experiences of participating in the course, the main study topics were more tailored to the patient’s experience of epilepsy, as well as their views on the SMILE (UK) course. An abbreviated topic guide can be found in Box 2 (see Appendix 7 for full guide). Topics included (1) participants’ experience of living with epilepsy, (2) negative and positive aspects of the course, (3) social contact after the course and (4) if or how they used what they had learned in the course. Interviewers probed and clarified responses as required. All participants were interviewed face to face. The interviews were undertaken by two researchers and interview times were variable, ranging from 25 to 40 minutes each.

Analysis

All interviews were recorded and then transcribed verbatim by an external third party. Transcripts were checked for accuracy by the researchers who conducted the interviews, and sample scripts were then checked by two collaborators (MM, LR). Researchers then undertook a line-by-line coding approach with initial codes noted in the margins of the transcripts and then grouped into broader themes. This process involved regular discussion between the researchers and study supervisors (MM, LR).

Methods

The methods of the external pilot are described in Chapter 2 and differed from the process evaluation in three main ways.

• Data were based on 10 participants.

• Participants were volunteers recruited through Epilepsy Action.

• Data gathered by semistructured interviews (seven participants) and a focus group (three participants).

Despite these differences, the findings regarding patients’ experiences and responses to the SMILE (UK) intervention identified similar themes that provide support for their validity and conceptual generalisation.

Main themes

Four themes were strongly represented in both studies (Figure 15), namely the importance of the intervention for reducing isolation, the opportunity to learn from each other, finding reassurance in the experiences of others and changing behaviours and practices after learning from others.

FIGURE 15.

Common themes reported by participants in the external pilot and process evaluation.

Changes noticed after the course

Participants described a number of changes after attending the SMILE (UK) course. In the process evaluation, 12 out of 20 participants spoke of changing their self-management practices after the course. Behaviour changes after the course included reducing alcohol intake, record keeping of seizures and improved medication adherence:

The programme has helped me to understand more about epilepsy and medication. I take my treatment more seriously now.

M, 39, process evaluation

Some spoke of sharing their newfound knowledge with others. Half of the participants in the process evaluation felt they were now talking more with others about their condition.

Fourteen participants across both studies (4/10 in the pilot study and approximately 10/20 in the process evaluation) described improved self-confidence after the course. This empowered many to have an open dialogue with their neurologist, GP or ENS either to suggest or to resist changes in medication. As one participant explained:

Last time I saw my neurologist he wanted to increase the medication. After having the course, I had the confidence to say that it isn’t worthwhile increasing the medication. I know that I’d only had these seizures because I hadn’t taken the medication.

F, 22, process evaluation

Similarly, participants described feeling more informed when discussing treatment options with their clinician, empowering them to take an active role in their health care:

It’s empowering you when you got to see the doctor to be more two-way about the discussion.

F, 60, pilot study

The participant handbook was also described as a new and useful tool in improving knowledge about epilepsy. Nine out of 20 participants from the process evaluation read the workbook during or after the course. Four participants had lent it to family members or friends. The workbook was used to access what they had learned during the course and to teach others about epilepsy. As one participant remarked:

I can’t stop carrying it around. Before I used to carry around my iPad all the time and bring that out all the time. But this is what I have to read all the time.

F, 32, pilot study

Limitations of SMILE (UK)

A group course offers participants the opportunity to share their own experiences and to learn from others. In cases when language or learning abilities vary, some participants may not understand the exercises or what is being discussed. PWE who had severe learning disabilities or were not able to understand English were excluded from the RCT. However, we recruited PWE from a diverse metropolitan population with varying levels of English mastery and learning skills. For three participants in the process evaluation, this was an issue. Four participants reported memory issues, which are frequent in PWE. This later had an impact on what they retained from the course and whether or not they could remember to do things in ‘real life’. And as mentioned above, two did not remember the course at all and thus were not interviewed. Together, nine PWE had language, learning or memory problems that affected what they retained from SMILE (UK).

External pilot

An external pilot, undertaken immediately after two pre-trial SMILE (UK) courses, demonstrated the feasibility and acceptability of providing group courses among volunteers with epilepsy in the UK. 82 Having two facilitators and locating the courses near a hospital centre can be an advantage when PWE report frequent seizures.

Main outcomes

We described the characteristics of 404 participants at baseline and the relationship between clinical, psychological and social factors and an epilepsy-specific QoL measure (QOLIE-31-P). 84 Participants’ mean age was 42 years, 54% were female and 75% were white. Median time since diagnosis was long at 18 years and a large proportion (69%) had experienced ≥ 10 seizures in the prior year. Nearly half (46%) reported additional medical or psychiatric conditions, 64% reported current anxiety and 28% current depression symptoms at borderline or case level, with 63% reporting felt stigma.

Baseline QoL was measured by the mean QOLIE-31-P (range 0–100). The mean score was 66, with a wide range (25–99). Psychosocial impairment at baseline appeared to be most closely linked to impaired QoL. In order of descending magnitude of association: current depression, anxiety, a history of medical and psychiatric comorbidity, felt stigma, greater seizure frequency, low self-mastery and low self-reported medication adherence were associated with lower QoL scores.

There were no significant differences in the primary outcome measure (QOLIE-31-P at 12 months) between the treatment groups [intervention mean 67.4 (SD 13.5) vs. control mean 69.5 (SD 14.8). Nor was there a significant difference at 12 months in any of the secondary outcome measures, including seizure control, impact of epilepsy, medication adherence, medication adverse effects, psychological distress, epilepsy stigma and self-mastery. 114

A process evaluation undertaken within 6 months of the intervention found that participants valued the opportunity to meet ‘people like them’. The structured learning methods encouraged them to share and compare feelings and experiences. Specific benefits included overcoming the sense of ‘being alone’ and improving self-acceptance through meeting people with similar experience. Over half reported that this and the group-learning exercises, eliciting comparison of attitudes and experience, helped them to improve their confidence to talk openly, as well as to make changes in their health behaviours.

However, almost half of those participants approached for interviews reported they experienced language or memory problems, which limited their ability to understand and/or learn on a group course, and sometimes their ability to manage their activities or behaviour changes in practice. Memory problems are well described by many PWE, particularly those with recurrent seizures. 143 Mood disturbance (like anxiety and depression) is also associated with self-reported memory impairment in epilepsy. This makes it possible, even likely, that without psychological interventions and reinforcement of a 2-day course for people with persistent seizures, it will be difficult to demonstrate sustained improvements after a 1-year period. Even those with average memory may struggle to recall the content of a 2-day course 1 year later. For this reason, Taylor et al. 59 emphasised the importance of integrating and reinforcing self-management education during routine care.

Recruitment

Recruiting volunteers via advertisements or through user groups requires an active response from patients. Those responding are often already more engaged in their health care and can result in a group less representative of the patient group as a whole. 144 Our study recruited from routine NHS epilepsy clinics. We used an opt-out process in creating patient lists from clinic lists, reducing to an extent how engaged patients needed to be in order to remain in the recruitment process. In a patient population who frequently report memory problems, such as those with epilepsy, people may forget to respond to invitations. It is also known that men are less likely to participate in research, including self-management trials. 145 Nevertheless our recruitment process resulted in a group fairly balanced between males and females.

Owing to the UK data protection regulations, which do not allow researchers to screen notes without consent, a large number of letters were sent out to patients attending neurology clinics prior to screening for eligibility for the study. This makes it difficult to assess the rate of enrolment as the first stage of recruitment included people without a confirmed diagnosis of epilepsy, people who had moved away and some with no recent seizures. Thus, we report the rate of enrolment from the final patient list, which contained patients who had not opted out of the recruitment process and who were deemed to meet the eligibility criteria. This list contained 1986 people, of whom 1458 could be contacted and 1088 were found to be eligible. Therefore, our recruitment rate was 37%. PWE may be reluctant to commit to a 2-day course because of the unpredictable nature of epilepsy, especially if additionally experiencing anxiety, depression and/or perceived stigma. Despite this, our recruitment rate was equal to or higher than that we have achieved in a prior individual self-management intervention for PWE recruited from EDs (27%). 12

Retention

At 82%, our retention rate in the study was high. There was no difference in follow-up rates between the two intervention arms, although those in the intervention arm who attended the course were more likely to complete the study (see Table 11). From the total group, 73 participants did not complete the study. Most (53 out of 73) could not be reached for the 12-month follow-up and, therefore, were withdrawn from the study. This can be an issue in large cities and in some of our catchment areas, where there is a 17% resident turnover per year. 146 Factors associated with loss to follow-up were being female, fewer years in education and more comorbidity.

Generalisability

We have already commented on recruitment and retention, which affect generalisability.

Our complex intervention trial made considerable demands on adults with poorly controlled epilepsy in terms of completing long questionnaires, attending a 2-day course and, for some, in-depth interviews. Recruitment might have been greater if the trial had been undertaken with participants who included volunteers, who were predominantly white and who were of higher socioeconomic and educational status, if the workload required of them had been less and, in our past experience, if they were recruited by GPs in primary care. 115,145,147

The pattern of predictors of high/low recruitment to complex interventions is consistent across research and also predicts uptake of educational interventions in routine NHS practice. For example, the offer by the NHS of a course on diabetes mellitus in south London met with a 30% uptake. 146 The diabetes mellitus group found that characteristics that reduced uptake of the offer of a course included lower socioeconomic status and black and ethnic minority status. 146

We offered an epilepsy course untested in the UK, which also required the completion of a questionnaire three times over 1 year and achieved a similar response rate as the offer of a fully evaluated and commissioned course for people with diabetes mellitus in practice. 146 Our recruitment rate was similar and the response rate at 1 year compared well with a more recent study of a UK cohort of PWE in primary care. 145 That research group found that, among all LTCs they studied, PWE had the lowest response rate. We therefore believe that our findings are as representative as it is possible to be of people with long-term poorly controlled epilepsy living in cities. We believe that they are generalisable to people internationally who are in receipt of specialist care in the context of a universal health-care system.

Quantitative findings

Comparison with MOSES

The RCT methods of evaluating MOSES and SMILE (UK), and the participant groups, differed in many ways. 1

1. The participants recruited in MOSES from epilepsy centres were informed about the trial via advertisements. Participants needed to actively engage prior to enrolment.

2. There was no minimum seizure requirement to participate in the MOSES study; 42% of the group did not have a seizure in the previous 6 months.

3. A total of 35% had comorbid conditions in the MOSES evaluation compared with approximately 50% in SMILE (UK).

4. The MOSES evaluation control group had a statistically significant longer median duration of epilepsy (18.2 years) than the intervention group (13.5 years). Thus, the duration of epilepsy in the MOSES control group was similar to the SMILE (UK) group, whereas the group receiving the MOSES intervention had a more recent diagnosis.

5. Seizure frequency was measured over a period of 6 months (6 months prior to intervention and 6 months post MOSES intervention). SMILE (UK) reported seizure frequency at 6 months and 12 months post intervention. SMILE (UK) also evaluated self-reported seizure frequency over the previous 12 months with a top category of ≥ 10 seizures.

6. The analysis of MOSES outcomes was based on a per-protocol assessment, which took into account participants having completed questionnaires at both follow-ups only. From 383 recruited participants, 242 remained in the analysis (63%).

7. The MOSES study included a ‘knowledge-of-epilepsy’ questionnaire, designed specifically for the study.

8. In the MOSES study, the longest follow-up duration was 6 months after the course.

The MOSES study resulted in an improved score with regard to knowledge measured by the questionnaire designed for the study, and also with regard to coping with epilepsy scores and in seizure frequency. No improvement was found on their generic measure of QoL. Statistical analysis using a per-protocol design was used to evaluate the efficacy of an intervention, or the effect of treatment receipt. The SMILE (UK) RCT was designed to evaluate the effectiveness of SMILE (UK) or, in other words, the effect of an offer of the intervention on the group. Efficacy results tend to be greater as they only include participants who have received the intervention. Participants not completing the assessments or the course are thus excluded. This can lead to bias in interpretation of the results as randomisation is no longer in effect. In our secondary analysis, we included a CACE study for the primary outcome to account for participants not receiving the intervention. We found no difference in QOLIE-31-P total scores at the 12-month follow-up between the participants in the SMILE (UK) group who had received the full intervention and the control group.

The only measure in common with SMILE (UK) that showed an improvement with the MOSES intervention was seizure frequency (although it was assessed differently). A greater proportion of participants in the MOSES group improved in terms of seizure frequency (i.e. they improved by two seizure frequency categories). As the actual numbers of participants in each category were not published, comparisons with SMILE (UK) are not possible. As the last follow-up was short, it is unknown whether or not the changes observed at 6 months would have persisted at 1-year follow-up.

Measuring knowledge and behaviour change

Evidence from prior trials supports the hypothesis that education can affect knowledge of epilepsy scores in people with well controlled and uncontrolled epilepsy. 1,2,95 However, the follow-up periods in these trials were 4–6 months, much shorter than the 12-month follow-up used here. This may be important in light of the memory problems described particularly by people with poorly controlled epilepsy.

In an earlier trial for people whose epilepsy was long term, we found a non-significant trend towards higher knowledge of epilepsy scores following a nurse intervention. 167 In a subsequent trial of nurse advice for people with new epilepsy, we found that for those groups of participants in lower quartiles of knowledge-of-epilepsy scores at baseline, there was improved knowledge-of-epilepsy scores after the nurse intervention. 95 These trials had a short follow-up period of 6 months.

In view of the above, we did consider adding a knowledge-of-epilepsy questionnaire in this trial, but decided not to do so. The reason for this was that the number of questionnaires needed to provide the information specified by our funding agency already demanded much time and energy from participants. More questionnaires might have reduced their participation and response rate at follow-up. These were also important concerns for our funders, ourselves and our ethics committees which reviewed the research proposal.

Studies of other LTCs with large numbers of participants have examined associations between knowledge increase and adoption of taught behaviours, and between adoption of taught behaviours and health outcomes;58 however, the associations were weak. A sequential concept of the health education triggering change in health outcomes is probably too simple. Lorig et al. 168 point out that patient education can bring about changes in behaviour and in health status, but the mechanisms involved are still not clear. 168 Research in many LTCs confirms that linking education to behaviour change and other outcomes is a continuing challenge. 59,169 Given the high prevalence of LTCs and the impact that they have on the individuals and service cost, there is a need for further research on this.

We believe that such research, especially in conditions such as epilepsy, for which few complex interventions have been evaluated, should be exploratory at first. The conclusions from this study suggest researchers need to be given flexibility in determining potential outcomes, undertake cohort studies to identify predictors of change and continue to use mixed-methods research, with process evaluations to describe ‘how’ and ‘why’ complex interventions work, or fail to do so.

Process evaluation: key themes

National Hospital for Neurology and Neurosurgery (Queen Square)

Matthew Walker (Principal Investigator), Fergus Rugg-Gun, Dominic Heaney and John Duncan.

Guy’s and St Thomas’ Hospital

Michalis Koutroumanidis (Principal Investigator).

Princess Royal Hospital

Jennifer Quirk (Principal Investigator).

Darent Valley Hospital

Robert S Delamont (Principal Investigator).

King’s College Hospital

Robert Elwes and Lina Nashef.

Lewisham Hospital

Ray Chaudhuri (Principal Investigator) and Asra Siddiqui.

Croydon University Hospital

Bridget MacDonald (Principal Investigator).

St George’s Hospital

Dora Lozsadi (Principal Investigator) and Hannah Cock.

Peer-reviewed publications

Kralj-Hans I, Goldstein LH, Noble AJ, Landau S, Magill N, McCrone P, et al. Self-Management education for adults with poorly controlled epILEpsy (SMILE (UK)): a randomised controlled trial protocol. BMC Neurol 2014;14:69.

Magill N, Ridsdale L, Goldstein LH, McCrone P, Morgan M, Noble AJ, et al. Self-management education for adults with poorly controlled epilepsy (SMILE (UK)): statistical, economic and qualitative analysis plan for a randomised controlled trial. Trials 2015;16:269.

Laybourne AH, Morgan M, Watkins SH, Lawton R, Ridsdale L, Goldstein LH. Self-management for people with poorly controlled epilepsy: participants’ views of the UK Self-Management in epILEpsy (SMILE) program. Epilepsy Behav 2015;52:159–64.

Wijnen BFM, Mosweu I, Majoie MHJM, Ridsdale L, de Kinderen RJA, Evers SMAA, McCrone P. Mapping of the QOLIE-31P to EQ-5D-5L utilities and comparison of the responsiveness of both instruments in epilepsy [published online ahead of print September 4 2017]. Eur J Health 2017.

Smith A, McKinlay A, Wojewodka G, Ridsdale L. A systematic review and narrative synthesis of group self-management interventions for adults with epilepsy. BMC Neurol 2017;17:114.

Wojewodka G, Hurley S, Taylor SJC, Noble AJ, Ridsdale L, Goldstein LH. Implementation fidelity of a self-management course for epilepsy: method and assessment. BMC Med Res Methodol 2017;17:100.

Ridsdale L, Wojewodka G, Robinson E, Landau S, Noble A, Taylor S, et al. Characteristics associated with quality of life among people with drug-resistant epilepsy. J Neurol 2017;264:1174–84.

Ridsdale L, Philpott SJ, Krooupa AM, Morgan M. People with epilepsy obtain added value from education in groups: results of a qualitative study. Eur J Neurol 2017;24:609–16.

Ridsdale L, Wojewodka G, Robinson EJ, Noble AJ, Morgan M, Taylor SJC, et al. The effectiveness of a group self-management education course for adults with poorly controlled epilepsy, SMILE (UK): a randomized controlled trial. Epilepsia 2018;00:1–14.

Conference presentations

Ridsdale L. Self-Management Education for Adults with Poorly Controlled Epilepsy: The SMILE Trial. Poster presentation. Joint Congress of European Neurology, Istanbul, Turkey, 31 May–3 June 2014.

Ridsdale L, Kralj-Hans I, Noble A, Landau S, McCrone P, Morgan M, et al. Self-Management Education for Adults with Poorly Controlled Epilepsy (SMILE): A Randomised Controlled Trial Protocol. Poster presentation. First Congress of the European Academy of Neurology, Berlin, Germany, 20–23 June 2015.

Ridsdale L, Philpott S, Krooupa AM, Morgan M. I Am Not Alone: Views of People with Epilepsy of Self-Management Education in Groups. Poster presentation. Second Congress of the European Academy of Neurology, Copenhagen, Denmark, 28–31 March 2016.

Ridsdale L, Wojewodka G, Noble A, Landau S, Robinson E, Mosweu I, et al. An RCT of Self-Management Education in Epilepsy: Participants at Baseline. Poster presentation. Association of British Neurologists’ Annual Meeting: ‘The Seven Ages of Man’, Brighton, UK, 17–19 May 2016.

Ridsdale L, Wojewodka G, Robinson EJ, Mosweu I, Landau S, McCrone P, et al. An RCT of Self-Management Education in Epilepsy: Characteristics of Participants at Baseline. Oral presentation. European Congress for Epileptology, Prague, Czech Republic, 11–15 September 2016.

Smith A, McKinlay A, Wojewodka G, Ridsdale L. A Review of Group Self-Management Courses for Adults with Epilepsy. Poster presentation. British Neuroscience Association and Association of British Neurologists’ joint symposium ‘Meeting of the Minds’, Cardiff, UK, 28 October 2016.

Ridsdale L, Wojewodka G, Robinson EJ, Landau S, Noble AJ, Taylor SJ, et al. Results of an RCT of Self-Management Education in Epilepsy: SMILE-UK. Poster presentation. Association of British Neurologists’ Conference, Liverpool, UK, 3–5 May 2017.

Wojewodka G, Hurley S, Taylor SJ, Noble AJ, Ridsdale L, Goldstein LH. Fidelity of a Self-Management Course for People with EpILEpsy (SMILE (UK). Oral presentation. Third Congress of the European Academy of Neurology, Amsterdam, the Netherlands, 24–27 June 2017.

Ridsdale L, Wojewodka G, Robinson E, Landau S, Noble AJ, Morgan M, et al. Self-Management education for adults with poorly controlled epILEpsy (SMILE-UK). Oral presentation. 32nd International Epilepsy Congress, Barcelona, Spain, 2–6 September 2017.

Ridsdale L, Wojewodka G, Robinson E, Landau S, Noble AJ, Morgan M, et al. Self-Management education for adults with poorly controlled epILEpsy (SMILE-UK). Poster presentation. British Chapter of the International League Against Epilepsy Annual Meeting, Leeds, UK, 4–6 October 2017.

Patient’s epilepsy

Can you tell me a bit about your epilepsy?

When did you find out you had epilepsy?

How did you feel when you were first diagnosed? How do you feel about it now?

How do you feel you cope with your epilepsy day-to-day?

Prompts: employment, relationships, anxiety

Help-seeking behaviour

In the past, have you tried to find out more about your epilepsy?

How have you done this?

Prompts: Nurses? Clinicians? Support groups? Internet? Alternative therapies?

SMILE course

When you first heard about SMILE, how did you feel about coming on the course?

Prompt: some people may feel worried or anxious before coming on a course like this, is this something that you experienced?

Prompts: any worries about going on the course? Looking forward to it?

So thinking about the SMILE course, what was the most interesting part of the course for you?

How did you feel about being part of a group?

How did you feel about hearing other people’s stories?

How did you feel about discussing your emotions in the group?

During the course you were asked to place stickers on a line to describe how you felt about certain topics, how did you feel about doing this?

I believe there was some discussion about medication for epilepsy, was this useful?

Did you find out any things that were helpful?

Did you learn anything new about your epilepsy?

Prompts: types of seizures, triggers diagnostic techniques, other people’s experiences

Since completing the course, do you think differently about your own epilepsy?

Prompts: triggers, medication management

Life after the course

Do you think the SMILE course has helped you in managing your epilepsy?

Prompts: Medication? Triggers? Warning signs? involvement in treatment?

Have you used any of the techniques you learned on the course?

Although the workbook is not an essential part of the course, have you found it useful?

Social network

Have you stayed in touch with anyone from the course?

Prompts: How many people? Have you found this useful?

Course improvements

Would you recommend other PWE to go on the SMILE course?

Is there anything about the course that you would like to change?

Is there anything else you would like to say about the SMILE course that you have not said so far?