Synopsis

Background

Public health importance of epilepsy

Epilepsy is a brain disorder defined by a persisting tendency for recurrent seizures. Epilepsy has a prevalence of 1%,1 is primarily a childhood-onset condition and costs the NHS £2B annually.2 Epilepsy covers ~30 types with different aetiology, manifestations and prognosis; and in this programme, we focus on school-aged children with common epilepsies who bear the greatest public health burden and for whom interventions would carry maximum national impact. The impact of epilepsy for children extends far beyond health, with higher and more persistent levels of emotional, behavioural, social and academic difficulties than children with other chronic health conditions or children in general.3–5

Evaluative research tools in childhood epilepsy

Despite its broad impact, there is no core outcome measure set (COMS) for childhood epilepsy research or established recommendations for epilepsy-specific patient-reported outcome measures (PROMs) for children that evaluate interventions for patient-centred research in the NHS or beyond. This is because there are few high-quality academic trials,6 and most industry-sponsored intervention trials are regulatory in nature and have focused on per cent seizure reduction and non-inferiority with benchmark drugs.7 There are, in fact, many outcome measurement tools,8 but some may lack evidence of validity, reliability and responsiveness to change – hence careful selection is necessary. These measures also need checking for their scope, applicability and acceptability with parents and children.

Changing agendas in childhood epilepsy

The evidence base for achieving optimal outcomes relevant to patients and parents in childhood epilepsy is underdeveloped,9 and it is crucial to understand that epilepsy-specific quality of life is not determined by seizures, neither from the child’s nor the parent’s perspective but rather to outcomes related to the child’s learning, mental health and social functioning.10–12 This insight invites alternatives to the current medical model that focuses on seizures, both as a primary target of intervention with antiseizure medications (ASMs) and as the primary index of treatment success. Through this programme, we aspire to shift this focus towards child and parent-relevant outcomes.

Interaction of epilepsy and sleep

In our stakeholder meetings, parents identified sleep as a problem area for children and for themselves. The interactions between epilepsy, ASMs, sleep and learning13 are complex. Sleep disturbance itself can trigger seizures; some seizures mostly occur in sleep, while epilepsy is also associated with up to 12 times increased prevalence (30–60%) of child-reported and parent-reported sleep disorders even without nocturnal seizures.14,15 Sleep disturbance leads to increased daytime sleepiness and worse behaviour, cognitive functioning and quality of life.15–17 The impact extends to the whole family: cosleeping or room-sharing parents wake seven times more often than controls,18,19 and spend an average of 4 hours asleep, with inverse effects on marital satisfaction and maternal health.20 New evidence shows that consolidation of memory during sleep (i.e. new learning) is disrupted by seizures and by nocturnal interictal discharges.21,22 ASMs have differential effects on sleep architecture in epilepsy: (newer) levetiracetam increases sleep efficiency, but (older) carbamazepine does not.23 Sleep disturbance therefore is a pivotal target of an intervention that could potentially improve child sleep and also learning, mood, behaviour, seizures and parental quality of life. Parent-based behavioural sleep interventions (BSIs) are shown to be effective in randomised controlled trials (RCTs) for typically developing, attention deficit hyperactivity disorder (ADHD) and autism populations,24–26 and such interventions could easily be modified for children with epilepsy (CWE).

Many of the same sleep problems seen in typically developing children are frequently experienced by CWE. BSIs are commonly and successfully used to treat these sleep problems in typically developing children and in some neurodevelopmental disorder populations (e.g. autism and ADHD) that overlap with epilepsy. Therefore, BSIs should theoretically be effective in CWE. However, there are special seizure-related considerations for CWE and their parents that may be important in designing an online sleep intervention.

Antiseizure medications: more harm than good in self-limiting childhood epilepsy?

Patients identify cognitive side effects of ASMs as their most important health issue.27 In our stakeholder meetings, parents of schoolchildren with self-limiting epilepsy were ambivalent about medicating their children and expressed difficulty in weighing up the potential benefits and harms of using ASMs. The available evidence in adults suggests that all ASMs adversely affect learning, older ASMs more than newer.28–35 So far, there have been no randomised trials in children that either assess learning or compare the outcomes of treatment versus no treatment in terms relevant to parents and children. We therefore proposed a pragmatic, RCT among CWE using a factorial combination of drug treatment options, with an online behavioural intervention or standard care (SC), evaluating medical, learning and patient-chosen outcomes. This innovative design allowed us to investigate the separate influences of drug and behavioural interventions on various aspects of quality of life and make ambitious advances to the evidence base for optimal outcomes in childhood epilepsy.

Original aims and objectives of Changing Agendas on Sleep, Treatment and Learning in Epilepsy

The programme aimed to improve outcomes for CWE by asking what impact different treatment policies have on their health and quality of life (Table 1). The specific aims were:

1. developing a core outcome set (COS) for childhood epilepsy research that is meaningful to children and their families and identifying and appraising existing epilepsy-specific PROMs of children’s health-related quality of life (HRQoL)
2. understanding and describing how concerns around drug treatment, learning and sleep problems influence their decisions about management.
3. tailoring and evaluating an intervention to ameliorate sleep problems in childhood epilepsy
4. evaluating the effectiveness and cost-effectiveness of drug-based seizure management and/or BSIs with regard to medical, learning and patient-centred outcomes.

Summary of changes to original aims

The trial design was modified following an internal pilot and was adapted in light of the coronavirus disease discovered in 2019 (COVID-19) pandemic. The revised aim was to evaluate the effectiveness of the Changing Agendas on Sleep, Treatment and Learning in Epilepsy (CASTLE) Online Sleep Intervention (COSI) BSI versus SC in a 1 : 1 parallel arm design (Figure 1). Comparison of ASMs was dropped owing to parental treatment preference,47 although UK paediatricians had originally expressed willingness to randomise.49 The new design was supported by the advisory panel (AP) and by participating clinicians. To adjust to the COVID-19 pandemic, all procedures from consent to outcome measurement were digitalised. The measurement of primary outcome – total Child Sleep Habits Questionnaire (CSHQ-22) score – was brought forward from 6 to 3 months. The principal secondary outcome measures were time to first seizure and 6-month seizure freedom; the child Strength and Difficulties Questionnaire; parental knowledge about sleep and parental self-efficacy (see Appendix 1, Table 5). The total sample size was revised to 110 to reflect the new primary outcome measure and to offset the closure of 16 recruitment sites in the aftermath of the COVID-19 pandemic. Inclusion criteria were broadened to any diagnosis of childhood epilepsy to increase recruitment eligibility. The revised aims are detailed in the work package sections below.

FIGURE 1 Revised programme research pathway diagram. PPI, patient and public involvement. (Figure not reproduced here; see the original article.)

Work package 1

Development of core outcome set for childhood epilepsy

This section contains material reproduced with permission from Wiggs et al.43, Crudgington et al.37 and Crudgington et al.38 These are Open Access articles distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) licence, 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/4.0/. The text below includes minor additions and formatting changes to the original text.

Prior to this work, there was no COS for childhood epilepsies, a major deficit for evaluative research. The original aims of work package 1 were to (1) develop a COS for childhood epilepsy research guided by the Core Outcome Measures in Effectiveness Trials initiative guidance; (2) complete a systematic review of epilepsy-specific PROMs of children’s HRQoL and appraise them guided by the COnsensus-based Standards for the selection of health Measurement Instruments initiative and (3) map the COS to existing PROMs.

We identified candidate outcome domains using systematic review methods and consultation with family advisory group. We used a Delphi survey to seek agreement on the most important outcomes with the participation of parents, young people and professionals. In the online Delphi survey, participants rated the importance of 48 candidate outcomes in two iterative rounds. In a subsequent face-to-face meeting, we sought final consensus on the critical outcomes and ratified a final COS of 38 outcomes across 10 domains. Using Delphi methodology avoids potential overinfluence of one type of stakeholder and captures different perspectives. Hence, our COS represents the view shared by young people with epilepsy, parents and various health professionals working with CWE. Further work could consider reducing the number of outcomes by ranking which outcomes are of most importance, relevant to each other.

Our COS captured commonly reported items such as ‘seizure frequency’ consistent with existing guidelines.50–53 However, by contrast, our COS highlights non-seizure-related outcomes such as ‘school attendance’ (attending school and engaging in school curriculum) and ‘feelings about epilepsy’ (emotions or reactions to having epilepsy such as embarrassment or stigma) agreed as critical for inclusion in the COS. We recommend that future evaluative research in epilepsy considers utilising the Core Health Outcomes in Childhood Epilepsy (CHOICE) COS as a framework for selecting outcomes for evaluative research. The COS was completed and published in 2019.37

We published a systematic review of epilepsy HRQoL PROMs in 2020.38 Our review identified 11 epilepsy-specific PROMs of children’s HRQoL. In particular, the Quality Of Life in Childhood Epilepsy (QoLCE-55) questionnaire has good evidence of structural validity, construct validity and internal consistency, and this was replicated in other studies. Health-Related Quality of Life in Children with Epilepsy (CHEQoL) is also a leading candidate, with good evidence of content validity, structural validity and construct validity. Choosing between these two might be decided by QoLCE being only by parent-reported questionnaire, whereas CHEQoL can capture both child- and/or parent-reported HRQoL. For clinicians and researchers to make an informed decision about which PROM to choose for a specific objective, there are other important properties to consider such as what constructs of HRQoL they measure, their importance to CWE and families, and how the questions and response options are understood by respondents. It is crucial to consider other important factors such as the content of a PROM and its face validity and acceptability to respondents.

Finally, we mapped the items of the PROMs to our COS and established the acceptability of the two leading candidate PROMs (CHEQoL and QoLCE-55).39 All 11 PROMs (17 questionnaire versions) cover a range of our COS outcomes, but none cover them all. Most PROMs cover aspects of Social Functioning and Mental Health, while others neglect items on Seizures or Sleep. Members of our AP found the two leading PROMs from our prior systematic review, QoLCE-55 (parent report) and CHEQoL (parent and child report), to be acceptable for research use in the context of the UK NHS.

When deciding which PROM to use for a specific purpose, it is essential to consider (1) matching the implicit conceptual framework to the outcomes of research interest; (2) robust evidence of measurement properties and (3) the appropriateness and acceptability of the questionnaire and individual questions to the respondents in the same context that the research will be conducted. We recommend researchers to consult with families in the context of public and patient involvement and engagement (PPIE) to ensure that measures are acceptable for their settings and their research questions. In our case, both QoLCE-55 (parent report) and CHEQoL (parent and child report) were considered acceptable and mapped well to our COS. This exercise was an important first step in applying our recently developed COS as a framework for selecting outcomes for evaluative research in childhood epilepsy in the UK.

Lay summaries were produced with the programme’s patient and public involvement (PPI) team. The COS was incorporated into the outcome measures (https://castlestudy.org.uk/castle-projects/choice-core-health-outcomes-in-childhood-epilepsy/) for the CASTLE clinical trial.

Work package 2

Qualitative research

This section contains material reproduced with permission from Gringras et al.48 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) licence, 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/4.0/. The text below includes minor additions and formatting changes to the original text.

Original aims

1. Explore and capture qualitative information relating to children’s and their parents’ experiences of the decisions and choices they make regarding the treatment and management of their epilepsy.
2. Explore the reasons why trial decliners made the decision to decline participation in the trial.

Planned design

Pragmatic qualitative study using semistructured, audio-recorded, face-to-face telephone/Skype^™ (Microsoft Corporation, Redmond, WA, USA) interviews with trial participants (up to 50 parent–child dyads) and trial decliners (up to 10 parent–child dyads), planned to take place at three time points (TPs): TP1: 2–4 weeks post randomisation; TP2 2 : 6 months post interview 1; and TP3: 12 months post interview 1.

Findings

Two families participated in the qualitative study (this represented 100% of total trial recruits eligible to participate in the qualitative substudy). One family consisted of both parents (child receiving intervention) and one family consisted of the mother (child receiving SC). Neither child chose to participate in the substudy. No trial decliners participated in the study. One family was interviewed at TP1 and TP2 and one family at TP1. No participants were interviewed at TP3 as the pilot had been stopped. The numbers were below target but are explained by the low recruitment rate in the trial. The findings presented are therefore based on a small evidence base and are thus tentative. They are presented within three main themes: Reasons for participating in the trial; Treatment preferences; and Living with my child with epilepsy:

Reasons for participating in the trial: Improve their own understanding of the condition; help other families; have greater access to information to help their child.

Treatment preferences: Parents entering the trial already had a strong preference for either treatment or non-treatment before taking part in the CASTLE trial; parents being aware of the side effects of medication and its possible impact on learning.

Living with my child with epilepsy: Anxiety about seizures, concerns about adverse effects from medication and worry about stigma remained a concern following diagnosis even with professional support and reassurance their child would ‘grow out’ of epilepsy. Parents responded with heightened vigilance and were using behavioural strategies to improve their child’s sleep.

Revised aims of Changing Agendas on Sleep, Treatment and Learning in Epilepsy Sleep-E

The aims are to critically explore:

1. experiences of children and parents/caregivers in relation to living with epilepsy, sleep (importance, challenges, management strategies and changes in sleep/strategies during trial)
2. use of COSI
3. factors related to acceptability of trial participation (e.g. why participate, challenges to sustaining participation, use of actigraphs and using COSI).

Planned design

Pragmatic qualitative study using semistructured remote audio or audio–video interviews with trial participants (up to 60 parent–child dyads; 30 from each group) planned to take place at two TPs post randomisation and within 3 weeks of completion of other data collection: TP1 at 3 months and TP2 at 6 months. Reflexive thematic analysis54 was undertaken with comparison of qualitative data between TPs, across groups and with other trial-related data.

Findings

Twenty-two families (19% of eligible trial recruits) participated in the qualitative substudy.

Thirteen families were in the SC arm (7 boys, 6 girls; 4–13 years, mean 9.30 years), and 9 were in the COSI arm. Of these, six families were categorised as COSI engagers (COSI^E) (four boys, two girls; 8–12 years, mean 9.8 years) and three as COSI non-engagers (COSI^NE) (one boy, two girls; 10–13 years, mean 11.3 years). COSI^NE participants were slightly older than COSI children. Two of the three COSI^NE cited family crises as reasons for non-engagement.

Perceptions of COSI: Five COSI^E parents talked of being familiar with COSI content, noting that suggested routines and practices were already ‘well ingrained’ or had been ‘advised before’. Despite familiarity, benefits did accrue, such as success using a ‘star chart to try to encourage him to sleep in his own bed’ and introducing ‘wind down’ time or the potential for introducing ‘more relaxation’ practices. One boy thought COSI practices were ‘helping me a bit’ and believed they would help his sleep ‘improve over the long-term’. Perceived gaps in COSI information included advice about managing ‘restless … tossing and turning’ sleep.

Although COSI content was reported as of ‘high quality’, another father reported feeling ‘overloaded with information’, some of which conflicted with other sources. Time pressures and distractions affected engagement, particularly so for the mother of a neurodiverse child who even after accessing specific neurodiverse content thought COSI was more relevant for ‘neurotypical children’. Despite apparently coping with their child’s sleep issues, parental sleep disruption and tiredness were problematic for four COSI^E families. Reported sleep problems included falling asleep, staying asleep, waking up or restless sleep.

Reasons for participating and trial acceptability: Reasons for participating were either to help other families and/or despite apparently coping with their child’s sleep issues, they were interested in finding out more information. Children’s responses included personal curiosity, excitement about wearing an actigraph or encouragement from parents. Overall, trial participation trial was deemed to be OK, although questionnaires and sleep diaries were somewhat burdensome albeit straightforward.

Limitations

Owing to the change in primary outcome assessment at 3 instead of 6 months, only three interviews were undertaken at TP2 prior to the trial window shutting, thus limiting the ability to explore changes that occurred over the period of trial participation.

Work package 3

Online sleep intervention (Changing Agendas on Sleep, Treatment and Learning in Epilepsy Online Sleep Intervention)

This section contains material reproduced with permission from Gringras et al.48 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) licence, 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/4.0/. The text below includes minor additions and formatting changes to the original text.

The aims of this work package were to:

1. tailor the core content of two existing standardised, parent BSIs into an online resource for parents of children with for epilepsy (COSI), in line with findings from PPI consultation and coproduction interviews
2. embed this content and enable stepwise delivery within an e-learning platform
3. determine the feasibility and acceptability of COSI for parents.

The key content (informational and instructional content) of two existing BSIs used in RCTs of children with ADHD24 and autism25 as identified from the published reports and clarification with the authors, and this core content was drafted into text format. The content from these studies was selected for a number of reasons, including: (1) these previous studies are evidence-based level 1 studies with good effect sizes; (2) they utilise well-defined interventions that are easy to tailor to specific family needs and goals; (3) their relative brevity; (4) their established effectiveness in children with neurodevelopmental disorders and (5) their suitability to be adapted to an online intervention. Both existing BSIs comprise a number of reliable combinations of behavioural change techniques defined as ‘systematic procedures included as an active component of an intervention designed to change behaviour’.55

The development of COSI began with qualitative investigation of what parents wanted from an online behavioural intervention for sleep problems in CWE.42 We conducted semistructured interviews with nine mothers of CWE. Thematic analysis54 revealed 10 themes or preferences for an intervention. Parents wanted: (1) other parents’ views and real-life experiences to be included; (2) recognition of how changes over time may influence the appropriateness of using various sleep-management options; (3) to be presented with a range of sleep management options from which they could select; (4) personalised information and suggestions for behaviour change options; (5) help to address child anxiety around sleep; (6) for the advice and behaviour change options to be practical; (7) general educational information about sleep and the relationship between sleep and epilepsy; (8) for parental worries and concerns to be acknowledged; (9) to receive help, support and reassurance around children’s sleep; and (10) to include the child in the intervention.

Armed with these parental views, we adapted two existing theory-driven paediatric BSIs into a novel online BSI (COSI) with the input of CWE.43 Six parents then fed back on COSI until 100% parental approval for content and functionality was achieved. COSI was subsequently deployed online in the CASTLE Sleep-E RCT. Parental experience with COSI in the trial was studied qualitatively in WP2 and quantitatively using digital metrics.

To achieve this, a COSI Evaluation Questionnaire allowed for parental quantitative assessments of the content and functionality of COSI (see Appendix 2). There were four elements to this questionnaire: (1) the functionality of COSI, (2) the acceptability of the content, (3) the acceptability in terms of length and content of the embedded COSI parent feedback scale (to be used to assess parents’ experiences of COSI when it is used in the CASTLE Sleep-E trial) and (4) allowing parents opportunity to make any further comments (see Appendix 2, Figures 3–6).

The COSI was subsequently deployed online in the CASTLE Sleep-E RCT. Parental experience with COSI in the trial was studied qualitatively in WP2 and quantitatively using digital metrics.

The functionality of COSI

Parents were asked to rate (on a 5-point Likert scale of ‘Always’, ‘Often’, ‘Sometimes’, ‘Rarely’ and ‘Never’) the frequency with which they could log on, access the web pages’ content, see and hear the videos, access and respond to the quizzes, print off any materials they wanted and navigate easily around the intervention website. Using the same 5-point scale, parents were also asked if they would recommend COSI to other families of CWE.

The acceptability of content

For each of the possible suggested 20 behaviour change strategies, parents were asked to report how acceptable they thought it was to include the technique as a suggestion (even if they personally might not use the technique). Parents rated these on a 5-point Likert scale of ‘Very acceptable’, ‘Acceptable’, ‘Neither acceptable nor unacceptable’, ‘Unacceptable’ and ‘Very unacceptable’.

The acceptability of the COSI parent feedback scale

Parents were also asked to rate the acceptability of the embedded COSI feedback scale in terms of the burden of the time taken to complete (on a 5-point Likert scale of ‘Far too long’, ‘A little too long’, ‘About right’, ‘A little too short’ and ‘Far too short’) and the understandability of the language and phrasing used (‘Very difficult’, ‘Difficult’, ‘About right’, ‘Easy’ and ‘Very easy’).

The final version of COSI included 11 modules as shown in Table 2. The first three COSI modules (A, B and C) were compulsory as they provided important information relevant to subsequent sections. Parents were required to complete each of these sequentially and were unable to access later modules until each was completed. Having completed module C (screening questionnaire), all subsequent modules were made available to parents and could be accessed at any time, although, as explained, to personalise the intervention, a list of ‘recommended modules’ specific to their child’s sleep was derived based on their responses to the screening questionnaire.

Work package 4

Changing Agendas on Sleep, Treatment and Learning in Epilepsy Sleep-E Clinical Trial

Parental interviews have revealed unmet needs in dealing with sleep problems,44 and at the same time, a mismatch in the ability of healthcare professionals to meet these needs.45 The primary objective of this trial was therefore to determine whether COSI in addition to SC improved parent-reported sleep outcomes in CWE aged 4–12 years compared to SC alone over a 3-month intervention period (Figure 2). Secondary objectives included outcomes at 6 months; assessing the impact on seizure control; objective measures of child sleep and memory consolidation; change in parental sleep knowledge and self-efficacy and the cost-effectiveness of the intervention.

FIGURE 2 The CASTLE Sleep-E CONSORT flow chart. This figure contains material reproduced with permission from Gringras et al.48 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) licence, 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/4.0/. The figure includes minor additions and formatting changes to the original text. (Figure not reproduced here; see the original article.)

This was a multicentre, parallel-group, unblinded RCT with internal pilot50 conducted across 26 NHS paediatric outpatient clinics in the UK (see Appendix 1, Table 3 for participant characteristics). The study was prospectively registered on the International Standard Randomised Controlled Trial Number Registry (ISRCTN13202325). The study protocol has been published.46 Trial reporting followed guidance from the Consolidated Standards of Reporting Trials (CONSORT) 2010 and the Consolidated Health Economic Evaluation Reporting Standards56 for the indicative economic evaluation.

The results of the trial showed that COSI + SC was not superior to SC alone in improving parent report of child sleep problems in young people with epilepsy. However, there was a significant decrease in objective actigraphy measures of child sleep onset latency between baseline and follow-up in the COSI + SC group compared with SC group, a difference of 16.5 minutes. This is approximately half the decrease in sleep latency achieved with melatonin administration.57 Parental knowledge of child sleep also increased in the COSI group. The results of the economic evaluation indicated that COSI + SC had a low likelihood of being cost-effective when compared to SC alone (see Appendix 3, Table 10).

When behaviour change techniques are translated to digital format, there is often a trade-off between access and scalability against low user engagement and inconsistent adherence, which are common problems in remote self-guided programmes.58,59 In this trial, for example, only 53% of families in the COSI arm engaged with the behaviour change content. Despite this, there were objective indicators of child sleep benefit, although these were perhaps not of sufficient magnitude to register as shifts in parental perception of their child’s sleep problems.

The online intervention resources met with high parental approval scores recommending COSI to other parents of CWE – 86% of strategies were rated as helpful or very helpful (see Appendix 2, Tables 7–9). However, both quantitative and qualitative data offered important insights into parental motivation and limited capacity to engage with the digital format, owing to time pressures, domestic issues and social distractions. We are already aware of the deficit that healthcare professionals perceive in meeting the needs of CWE who have sleep problems.45 Taken together, these results suggest that an exclusive digital approach, despite objective efficacy and parental approval, has limited effectiveness and cost-effectiveness as a public health strategy to solve sleep problems in CWE in the NHS. A subsequent Policy Lab (https://doi.org/10.18742/pub01-227) reached similar conclusions and emphasised the need for online resources blended with trusted human contact to facilitate engagement with behavioural sleep management strategies.

Additional work was published surveying clinicians and parents on families needs44,45 as well as parental fears and concerns.41 Actigraphy and diary studies of cosleeping and dyadic parent–child sleep relationships are in preparation.

Patient and public involvement

The CASTLE research programme has been informed throughout by the views and experiences of an AP (http://go.microsoft.com/fwlink/p/?LinkId=255141) made up of four young people with epilepsy, nine parents of CWE and three adults with epilepsy. The AP members were trained in research approaches and codeveloped a handbook to guide our collaborative working. The AP members have met face to face and in person regularly throughout the programme, facilitated by a Family Engagement Officer, and have informed all aspects of all work packages of the programme grant. Some of the main contributions are listed below:

Work package 1: core outcome set

The AP informed the design of the materials used to recruit and inform families about the Delphi survey, informed the wording of the Delphi survey to ensure acceptability to participants, and coproduced the lay summaries. The AP was consulted about the candidate outcome domains.

Work package 2: qualitative research

The AP informed the interview topic guides, the use of augmented activities for children, the materials used for families to gain consent and assent, as well as insights into data analysis process.

Work package 3: Online Sleep Intervention

The AP tried COSI and helped develop the recruitment materials.

Work package 4: CASTLE sleep-E clinical trial

The AP and the lay principal investigators endorsed the focus on sleep and helped the trial team redesign the study when the original trial design that included antiseizure drugs struggled to recruit in the pilot phase. The AP developed all materials shared with families, codeveloping an instructional video for families; trialling the digitalised consent process; the instructions families received for the iPads and COSI; the use and content of postcards to stay in touch with families enrolled on the trial between follow-up appointments; the outcomes measures; and surveys completed. The AP developed a guide to help clinicians have conversations about the trial with families and an easy-to-understand flow chart of the study, including randomisation.

Additional work conducted by the AP

Clinician Survey:45 consultation for the survey questions, insights into the data analysis.

Systematic Review on parental concerns and fears:41 involvement in construction of search terms, insights into analysis.

Parents have contributed to and been acknowledged on four papers written from the work of the programme and have copresented at two events, we have cowritten blogs and the PPI section of the programme website (https://castlestudy.org.uk/).

The AP wrote and created a short YouTube (YouTube, LLC, San Bruno, CA, USA) video (https://youtu.be/fQbAJVMBj0I) to raise awareness of the challenges associated with living with epilepsy as well as working to codevelop a ‘Top Tips guide’ (http://castlestudy.org.uk/wp-content/uploads/2024/03/CASTLE-top-tips-for-clinic.pdf) and ‘prompt sheet’ (http://castlestudy.org.uk/wp-content/uploads/2024/03/CASTLE-clinic-prompt-sheet.pdf) to help parents’ prepare and get the most out of attending an epilepsy clinic for their child. Each year, the AP has led activities associated with Purple Day to raise awareness of epilepsy.

The work of the AP, the impact of their contributions on the programme, research team and AP members have been formally evaluated in a research project which was designed in collaboration with the AP members. Online surveys gathered the views of AP and research team members alongside an independently moderated online group discussion with AP members.

General discussion

We used mixed quantitative and qualitative approaches across work packages to better understand the needs and preferences of families of CWE experiencing sleep problems and to conduct a pilot and then a full-scale RCT with an integrated economic evaluation of parent-based online sleep behaviour intervention compared with SC alone.

We first created a novel COS for childhood epilepsy research, mapping existing PROMs to the COS. Research in epilepsy has largely concentrated on evaluating how antiseizure treatments influence seizure frequency. However, it is equally important to consider broader aspects potentially impacted by both seizures and their treatment, including child’s learning, sleep, mental health, school and social life.

We used the COS in the clinical trial. We completed a related work for epilepsy mental health PROMs in 2024.40 Following the pilot study, we revised the trial design to compare the BSI against SC, dropping the drug versus non-drug comparison, which was incompatible with parent preference. We collected nearly all PROMs online to adapt to pandemic limitations.

We found that while parents viewed sleep problems as very important, they did not generally raise these with their paediatric team. At the same time, healthcare professionals reported a lack of capacity to manage non-seizure-related issues. These two seemingly paradoxical findings partly reflect a lack of both public and professional knowledge about the treatability of sleep problems in children.

The online sleep intervention COSI proved no more effective than SC in reducing parental perception of their child’s sleep problems and was not cost-effective. However, COSI did objectively improve children’s sleep latency by 16.5 minutes and it did increase parental knowledge about sleep problems. Like many other digital health interventions, COSI faced a low uptake, which doubtlessly reduced its effectiveness.

Clinical equipoise versus parent preference in paediatric randomised controlled trials

This section contains material reproduced with permission from Wiggs et al.11 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) licence, 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/4.0/. The text below includes minor additions and formatting changes to the original text.

One aim of our original trial was to address a critical issue in the childhood epilepsy evidence base – the lack of robust evidence comparing drug treatment with no-drug treatment in a common self-limiting form of childhood epilepsy.6,60 As in other paediatric clinical trials, researchers undertaking paediatric epilepsy clinical trials face a range of challenges,61 including recruiting children to epilepsy medication effectiveness trials, the role of parent/guardian as ‘gatekeeper’ for trial participation62–64 and factors related to consent and outcome selection and assessment. Without a robust evidence base for the best treatment option for the child, uncertainty can exist for both parents and paediatrician. In clinical trials, parent or patient preference for treatment (or non-treatment) is a common cause for recruitment and participation challenges and for declining randomisation and this may compromise external validity.65,66

Our engagement with parents in the pre-trial stakeholder meetings meant that we were aware that some parents of CWE had ambivalent feelings towards medicating their children and they talked of the difficulties they experienced when trying to weigh up the potential benefits and harms of treating their child’s epilepsy with ASMs. While monitoring the internal pilot, we found that although eligibility and approach rates were as expected, consent rates fell considerably below expectations.

Consultation with clinicians highlighted that some parents expressed strong medication-related preferences; this meant that randomisation to a potentially non-preferred arm discouraged parents. Clinicians considered that the parent treatment preference issue was less likely to impact on recruitment to a trial that only involved randomisation to either sleep intervention or SC. Discussion with our AP indicated that a trial focused on sleep alone would be more acceptable than one involving drugs.

We had some prior indication of treatment ambivalence in the pre-trial planning stakeholder meetings, but we had failed to anticipate its translation to such strong parent treatment preference. The programme may have benefited from earlier, more comprehensive PPI consultation and preference elicitation to inform the design. Specific misunderstandings about drugs have the potential to influence parents’ decisions, but parents rarely voice concerns during discussions with practitioners.64 Other research has shown that parental reasons for strongly held preferences include concerns about adverse effects and negative attitudes towards ‘new’ or ‘experimental’ interventions.67 The original trial design compared two well-established ASMs.

Self-limiting epilepsies perfectly illustrate the uncertainty about the risk–benefit equation of ASMs in childhood. Hence, we believe it is still important to address the ASMs/no-ASMs question in a design more acceptable to parents and children. While changing the randomisation arms is one obvious option to make the trial design more acceptable to potential participant families, incorporating parent and patient preferences is another increasingly used adaptation.67 Patient Preference Trials (PPTs) allow patients to choose their preferred intervention arm but at the expense of increased sample size and trial duration; such designs do not necessarily improve informed consent though.67 Discrete choice experiments are another approach that could elicit and quantify trade-offs between parent preference against the perceived utility of an intervention.47 Although a PPT design was offered as an option following the pilot phase and was acceptable to parents and children on our AP, a simplified trial design evaluating the BSI versus SC, with no ASM arms, was preferred by the trial funder.

Evaluative research in childhood epilepsy

The coproduction of a COS for childhood epilepsy research fills an important methodological gap. The active participation of professionals, young people and parents means that outcomes included are of relevance to young people with epilepsy and their families. We found it difficult to narrow down the scale of the COS to fewer than 10 domains, and this may reflect both the breadth of issues that are important to measure as well as the heterogeneity in priorities between individual children and families. The impact of the COS can be maximised by mandating its use in research sponsored by United Kingdom Research & Innovation, major funding agencies and affiliated charities.

Implications for practice

Improving sleep in CWE is very important to young people and their parents because sleep problems can affect learning, behaviour, seizure susceptibility, drug response and family functioning as well as lead to higher costs for health care and society. The CASTLE programme has shown that clinicians generally lack capacity to manage non-seizure-related issues and families struggle to raise issues, including sleep problems with their clinicians. A parent-based sleep behaviour intervention delivered exclusively online did not reduce parent-reported sleep problems but did improve sleep latency in children.

The implications for clinical practice can be summarised as follows:

1. Parents of CWE may benefit from different forms of support that facilitate their engagement with digital health to manage their child’s sleep problems.
2. There are gaps in capacity of healthcare professionals’ sleep-related training and knowledge as well as service-related challenges to address sleep problems.
3. Greater integration of physical and behavioural health in childhood epilepsy secondary care.
4. A better understanding of the needs of families of CWE and their preferences for intervention delivery in the NHS will support service planning, evaluation and research.

Challenges and limitations

Contractual and ethical review delays

Twelve months were lost through a combination of contractual delays with the sponsoring NHS trust and protracted correspondence with the HRA. These factors were beyond the control of the programme executive and resulted in slippage to our timelines. Both of these administrative issues were disproportionately costly to the programme. We successfully applied for a variation to contract to make up for the 12-month slippage.

COVID-19 recruitment difficulties

During the COVID-19 pandemic and in its aftermath, substantial adaptations were needed to sustain the programme. In clinical research in general, but in paediatrics especially, there was a profound loss of in-person and outpatient interactions as well as in research staff capacity. Ten of our 41 recruitment sites closed as a result. At many sites, there was a long-term redesign of outpatient services with more use of remote consultations. Families were faced with many challenges, including providing education at home. We partially mitigated these adaptations by introducing remote digital consent and data collection.

Attrition between recruitment and randomisation

We experienced some attrition between recruitment and randomisation during the pilot phase. We introduced another measure to mitigate this – the Trial Manager and chief investigator hosted a weekly Teams drop-in session for site researchers. By sharing experience and best practice, we prevented further early attrition of participants. These sessions also led to better participant engagement and improved data collection and modifications to trial documentation as well as maintaining morale in the post-pandemic environment.

Choice of outcome measure

We also acknowledge that the outcome measure we used may have had limitations. All currently available subjective outcome measures in paediatric sleep medicine have recognised shortcomings, with only very weak correlations with objective laboratory assessments. At the same time, obtaining reliable objective data remains a challenge: even relatively simple measures such as actigraphy are difficult to implement successfully, with low levels of adherence and usable data.

Although we employed a shorter, unvalidated version of a widely used sleep problems questionnaire, our psychometric analyses showed that this was unlikely to have influenced the findings. Specifically, Cronbach’s alpha calculations and subanalyses of the items common to both the short and long forms (see Appendix 1) revealed no significant differences between treatment groups.

Severity of initial sleep problem

We agree that, with hindsight, focusing recruitment only on children with severe sleep disorders might have yielded different results. However, as discussed, our recruitment was already severely challenged, and we were also managing several unforeseen circumstances, including the impact of COVID-19. In this context, broadening the eligibility criteria was necessary to achieve recruitment targets.

Reflections on issues relating to equality, diversity and inclusion

Participant representation

There were two main sets of participants in our research: the young people included in the trial, and their parents who accessed the intervention.

We included CWE meeting the eligibility criteria from 26 NHS trusts geographically distributed across the UK. They ranged in ages from 4 to 12 years, and two-thirds of them were male as expected from the epidemiology of childhood epilepsy. Ethnic backgrounds were diverse but with slightly fewer non-White participants (10.6% vs. 14%) than in the national population. Since the intervention and several outcome measures, including an electronic memory assessment, were only available in the English language, fluency in English was included as one of the eligibility criteria. This may have limited the diversity of participants.

Research team

Four of our co-applicants were women, and four men. The chief investigator was from an ethnic minority group. Among the researchers appointed to the work, four out of five were women and one was from an ethnic minority group. There was a wide range of expertise and experience, including both clinical and non-clinical backgrounds, in the team. Junior research staff were encouraged and supported to develop their expertise through mentoring from senior team members and supported to attend training opportunities and conferences. One junior researcher was accepted onto a PhD programme and is now a postdoctoral associate at the University of Oxford. Another researcher was promoted to Research Fellow at her institution. Both have been highly successful in conference submissions and publications.

The PPIE representatives included during the research were mostly female, including nine parents and three adults with epilepsy plus four young people with epilepsy. The diversity of the panel in terms of ethnicity was limited. We attempted to find families using many different approaches at the beginning of the programme through social media posts, epilepsy charity events and speaking with clinicians who knew families who may be interested in being public advisors. While we did not have ethnic diversity, the AP included disabled and/or neurodivergent parents and young people and those from low socioeconomic status areas.

Recommendations for future research

There is a wide range of recommendations for future applied health research arising from this programme. These are summarised as follows:

1. producing and evaluating a parallel training manual and accredited course for Epilepsy Specialist Nurses to be used in conjunction with the parental COSI resource
2. examining the barriers and enablers for the upskilling of Epilepsy Specialist Nurses in each ICB
3. developing a quick assessment tool to stratify families’ needs to match a tiered intervention strategy, for example to account for duration of sleep problems and previously attempted strategies
4. mapping family needs for sleep management to available professional resources for each ICB
5. health economic evaluation of the impact of inclusion of COSI into paediatric epilepsy care
6. examining the potential synergies of integrating sleep management training into mental health assessment and management at secondary care level
7. conducting discrete choice experiments to inform the adaptive design of future drug trials in paediatric epilepsy.

Conclusions

Findings from this programme suggest that the sleep needs of young people with epilepsy and their families are not being adequately met as evidenced by the reported capacity of clinical practitioners and through qualitative interviews with both families and clinicians.

The shift in perspective towards broader outcomes in childhood epilepsy is in its nascent phase in the UK. Initiatives like the CHOICE COS complement the drive by charities, such as the James Lind Alliance and the Epilepsy Research Institute, to place the needs of patients high on the research agenda. However, both researchers and funding agencies need to push this agenda forward to achieve lasting impact.

The digital route to solve problems posed by limited professional resources in sleep behaviour management is logical to explore but is beset by real-life challenges that are commonly experienced in this emerging digital health field. Principal among these are poor app uptake and high attrition coupled with an uncertainty about functional engagement.

Access to the COSI intervention did not alter parental perceptions about their children’s sleep problems even though some aspects of their children’s sleep objectively improved. Aside from low uptake, other reasons may have included patient selection. Qualitative interviews suggest that complex psychosocial contexts reduce motivation and capacity to engage with an unsupported digital intervention. These insights suggest potential for a blended intervention facilitated by the family’s epilepsy specialist nurse.

In conclusion, parents did not engage with COSI as anticipated as a stand-alone digital intervention. However, since the underlying behaviour change techniques remain valid, rethinking the mode of delivery should be explored in future research.