Types of studies

We will include the following type of trials:

1. RCTs or quasi‐RCTs

2. Double‐blinded, single‐blinded or unblinded

3. Parallel, cross‐over and cluster design

Types of participants

We will consider studies of behavioural intervention or self‐management eligible for this review if they include anyone aged younger than 18 years with any clinical diagnosis of new or recurrent epilepsy. If the study is a mixture of adults, young people and children, we will consider the study for inclusion if: 1. results are reported separately for those younger than 18 years, 2. at least 50% or greater subset of population are younger than 18 years or 3. if no data report on percent younger than 18 years, we will consider studies reporting mean age younger than 18 years. In this latter scenario, we will consult with clinical experts to agree that the interventions are applicable to children.

For service delivery, we will include models delivered specifically for children. We will exclude models targeted for adults. Models for a range of ages groups will be carefully considered for inclusion based on 50% being participants younger than 18 years or mean age younger than 18 years. Models that do not specify a focus on children but do not provide information of the range of ages the service is delivered for will be carefully consider for inclusion, consulting clinical experts. If included, we will label studies as 'not specified' in this respect.

We will also consider studies that include epilepsy with other medical conditions if the results are reported separately for each condition.

Types of interventions

We will include any type of behavioural intervention or self‐management used to promote seizure control and improve quality of life in children with epilepsy.

The behavioural interventions for epilepsy can be broadly grouped, but not limited to the following.

1. Psychobehavioural interventions, such as motivational interviewing, psychoeducation interventions and cognitive behavioural therapy

2. Mind–body approaches such as mindfulness, yoga, meditation, exercise programmes, progressive muscle relaxation and neurofeedback

3. Creative arts therapy such as music and art therapy

The self‐management interventions for epilepsy can be broadly grouped, but not limited to the following.

1. Educational or counselling programmes with provider‐led self‐management education components and aims to develop knowledge, understanding, decision‐making skills, or a combination of these

2. Self‐guided self‐management materials and health condition tracking, including digital platforms such as Apps with automated interactivity and programmed user outputs

3. Social support from peers; in person or via digital media

We will aim to group behavioural interventions and, separately, self‐management by likely mechanism(s) of action and compare different groups of behavioural interventions or self‐management with usual care for children with epilepsy. Where possible, we will compare one group of behavioural interventions with another group or compare different behavioural interventions within the same group.

There will be no restriction on the mode, setting or duration of delivery of intervention.

We will include models of service delivery intending to improve the quality of care and outcomes for children with epilepsy. To be eligible for inclusion, the focus of study will be on the efficiency of service delivery model. For example, the same type of intervention delivered by a specialist nurse versus non‐specialist nurse or if the same elements of a service are compared in different settings, such as a clinical‐based service versus an online service.

For studies where the service delivery model encompasses behavioural intervention or self‐management, to be included in the review, a model of service must include information about at least one of the following.

1. Provider – the person(s) delivering the service, that is a comparison between a professional such as use of epilepsy specialist nurses versus the usual model of care.

2. Service model, such a named model of care or recognised service model versus usual model of care.

3. Specific contextual factors, that is comparing different settings for service delivery such as services delivered in a specialist centre, hospital, community urban or rural setting; remotely using information technologies such as secure video conferencing or at different times (timing) along the person's epilepsy pathway or frequency of care contacts.

We will list all treatment arms of each study in the 'Characteristics of included studies' table.

Types of outcome measures

Electronic searches

We will search the following databases.

1. Cochrane Register of Studies (CRS Web) using the search strategy shown in Appendix 1. CRS Web includes randomised or quasi‐randomised, controlled trials from specialised Registers of Cochrane Review Groups including Epilepsy, Cochrane Central Register of Controlled Trials (CENTRAL), PubMed, Embase, ClinicalTrials.gov, and the World Health Organization International Clinical Trials Registry Platform (ICTRP).

2. MEDLINE (Ovid, 1946 to current) using the search strategy shown in Appendix 2.

3. CINAHL Plus (EBSCOhost, 1937 to current) using the search strategy shown in Appendix 3.

4. PsycINFO (EBSCOhost, 1887 to current) using the search strategy shown in Appendix 4.

There will be no language restriction in our searches. The search strategies do not specify any age range, to allow us to capture studies with mixed adult and children populations.

Searching other resources

We will check the reference lists of retrieved studies for additional reports of relevant studies.

Selection of studies

We will merge and deduplicate the search results and use Rayyan to manage the results of our searches (Ouzzani 2016; www.rayyan.ai/). We will group multiple reports and papers related to a single study under a single reference ID from the first publication of the study. We will screen papers in two stages.

1. At stage one, two review authors will independently screen all titles and abstracts of papers identified by the search of relevance. We will exclude papers that are clearly irrelevant at this stage.

2. At stage two, two review authors will independently screen the full‐text papers and assess for eligibility of studies for inclusion.

We will resolve any discrepancies with discussion, if unable, we will involve a third review author. The selection process will be detailed in a PRISMA flow diagram, and we will list all records excluded at stage two in the 'Characteristics of excluded studies' table.

Data extraction and management

Two review authors will independently perform data extraction. Another review author will check data, and we will resolve any potential discrepancies via discussion. We will extract the following information from all studies, using a prepiloted data extraction form. We will initially pilot the data extraction form on a selection of included studies from the original Cochrane Review's 'Characteristics of included studies' table (Fleeman 2018).

1. Methods

   1. Study design

   2. Date of study

   3. Method of randomisation

   4. Allocation concealment

   5. Blinding

   6. Duration of participation

2. Participants

   1. Population characteristics

   2. Setting

   3. Method of recruitment

   4. Total number recruited, randomised and analysed

   5. Baseline seizure type, seizure frequency and disease duration

3. Interventions

   1. Detail nature of intervention(s)

      1. Type

      2. Duration

      3. Timing

      4. Delivery

      5. Provider

      6. Was the delivery to individuals or groups?

   2. Detailed nature of comparison(s)

4. Outcomes

   1. Definitions, unit of measurements and person measuring

   2. Time point measured

   3. Missing data

5. Notes: any conflict of interest of trial authors

We will identify which studies have data available for numerical meta‐analysis, data that need converting for meta‐analysis and data that are suitable for a narrative synthesis. We will transfer data to Review Manager Web (RevMan Web 2022).

Two review authors will extract data. We will resolve any discrepancies through discussion and, if necessary, with an additional review author. We will contact study authors if there is inadequate information within available reports.

We will synthesise the characteristics of all studies that will contribute to each comparison and present these in the 'Characteristics of included studies' table in our full review.

Assessment of risk of bias in included studies

As the review considers diverse inventions with complex pathways between delivery and outcomes and considerable heterogeneity in people and settings studied, interventions, outcomes and outcome assessment, these features will be responsible for underlying uncertainties across the review. Therefore, we will use the original RoB 1 tool to assess studies. We will refer to the Cochrane Handbook for Systematic reviews of Interventions for assessment of risk of bias in studies (Higgins 2011).

Two review authors will independently assess each study comparing their assessments, with any differences resolved through discussion or recourse to a third review author. The RoB 1 tool examines: random sequence allocation, allocation concealment, blinding of participants and personnel, blinding of outcome assessors, incomplete outcome data, selective outcome reporting and other biases. We will assess bias for each of the parameters ('low', 'high' or 'unclear') according to the Cochrane Handbook for Systematic reviews of Interventions (Higgins 2011).

We are interested in quantifying the effect of assignment to the interventions at baseline using intention‐to‐treat analyses. Risk of bias assessment will inform GRADE and the summary of findings table(s).

Measures of treatment effect

For outcome measures reported as continuous data, we will use mean differences with 95% confidence interval (CI) if studies use the same scales, or standardised mean differences with 95% CIs if studies use different scales. For dichotomous/binary data, we will use risk ratios with 95% CIs. If this is not possible, we will present the measures of treatment effect or summary statistics narratively as reported in the published papers.

Unit of analysis issues

We will attempt to manage the following unit of analysis issues according to the Cochrane Handbook for Systematic reviews of Interventions (Higgins 2022).

1. Cluster and cross‐over trials: we will use adjusted effect sizes as reported in included studies. If studies do not report adjusted effect sizes, we report these studies narratively. In the case of any cross‐over trials, we will extract data from the first treatment period only.

2. Multiple arms: we will analyse separate comparisons. We will not attempt to pool treatment arms or split control groups. If studies include different 'doses' or 'intensities' of the same intervention versus a control, then we will attempt to pool intervention arms after consulting with topic experts as to the relevance of this approach to practice.

3. Multiple measurements: we will not attempt to pool multiple measurements over time in meta‐analyses. We will present short‐term and long‐term outcome data separately.

Dealing with missing data

We will contact study authors directly to obtain missing relevant outcome data, and calculate missing data using data from other trials where possible as advised in the Cochrane Handbook for Systematic reviews of Interventions (Higgins 2022).

Assessment of heterogeneity

We will assess clinical heterogeneity between studies by reviewing the differences across trials, including population factors, interventions and study design between studies. We will evaluate statistical heterogeneity using the I² statistic. We will interpret the I² statistic as:

1. 0% to 40%: might not be important;

2. 30% to 60%: may represent moderate heterogeneity;

3. 50% to 90%: may represent substantial heterogeneity;

4. 75% to 100%: considerable heterogeneity.

If there is considerable clinical heterogeneity between trials (i.e. I² greater than 75%), we will not conduct any meta‐analyses.

Assessment of reporting biases

For all studies, we will check whether outcomes intended to be measured (reported in the methods section) were reported in the results section of papers. Where data were missing, we will attempt to contact study authors. If more than 10 studies are included in an analysis, we will assess funnel plot asymmetry.

Data synthesis

We will present results of data extraction and quality assessment for each study in structured tables and as a narrative summary. If a meta‐analysis is feasible for subsections of the outcome measures (see Types of outcome measures), we will use a random‐effects model. There is a broad range of self‐management interventions, and these studies are likely to use a variety of different outcome measures; therefore, the results are likely to follow a distribution across studies rather than a single true effect. Where meta‐analysis is not feasible, we will present results for each outcome in a narrative synthesis. We will follow the guidance in Chapter 12 of the Cochrane Handbook for Systematic Reviews of Interventions on synthesis without meta‐analysis (McKenzie 2022).

Subgroup analysis and investigation of heterogeneity

We have not planned any subgroup analysis a priori. If we decide to combine any studies in a meta‐analysis and find evidence of statistical heterogeneity, we will consider conducting post‐hoc subgroup analyses where appropriate and where the data allow.

We will note data reported by characteristics relevant to potential inequalities (such as gender, ethnicity or socioeconomic status) and discuss any evidence of differential effectiveness.

Sensitivity analysis

We have not planned any sensitivity analysis a priori. If we decide to combine any studies in a meta‐analysis and find evidence of statistical heterogeneity, we will consider conducting post‐hoc sensitivity analyses where appropriate and where the data allow.

Summary of findings and assessment of the certainty of the evidence

For the summary of findings table(s), we will include the intervention and comparison, the main outcomes, details of scale, number of studies and participants with each outcome, relative and absolute effect (where reported) and GRADE consideration for each study. We will present a separate summary of findings table for each comparison if applicable. We will use the GRADE approach and consider overall risk of bias, consistency of effect, imprecision, indirectness and publication bias in GRADE consideration (Schünemann 2013; Schünemann 2022). Two review authors will independently make judgements, with any differences resolved through discussion or involving a third review author. We will justify all decisions to downgrade the certainty of the evidence using footnotes and make comments to aid reader's understanding of the review where necessary. We will use GRADEpro GDT to create summary of findings table(s) for the following outcomes (GRADEpro GDT).

1. Seizure frequency

2. Participant report of health and quality of life using validated scales