Method

Summary of trial design

This study is a UK three-site, two-group RCT of a parent/carer group programme to address functionally impactful RRB in autistic children. Families were randomised at each site to receive either the 8-week URB parent/carer group programme or the 8-week Learning About Autism (LAA) parent/carer group programme, equivalent to best current practice and operating as an ‘attentional control’ for time spent in a parent group and attention from autism professionals and other parents/carers. Assessments were administered on entry (baseline) to the trial, at the end of the eight parent/carer group sessions (10 weeks), at 24 weeks primary end point and at the 52 weeks follow-up. The study is a phase 3 superiority trial, with a parallel group design in which each family is randomised 1 : 1 to one of the two parent/carer group programmes.

Parent/carers were randomised 1 :  1 to receive either the URB programme or the LAA attentional control programme. The study was conducted through three research sites:

• Cumbria, Northumberland, Tyne and Wear NHS Foundation Trust (CNTW)
• NHS Lothian
• Tees, Esk and Wear Valleys NHS Trust (TEWV)

Understanding Repetitive Behaviour programme

Understanding Repetitive Behaviour is a manualised programme designed to help parents/carers of young autistic children to recognise, understand and learn how to support their child’s functionally impactful RRB. It is an 8-week programme designed to be delivered by community-based professionals with knowledge and experience of working with young autistic children and their families, who have been trained to deliver the URB programme. Each weekly session lasts for 2 hours (total duration of group programme is approximately 16 hours).2

Each parent/carer is provided with an URB folder with copies of the weekly session slides and materials and individual support to identify sensitive approaches to address functionally impactful RRB. ‘At home’ activities are set each week for parents/carers to complete with their children between sessions. Parents/carers are asked to keep a diary of these activities. The opportunity to identify and practise new ideas outside of the clinic setting is considered to be important for investigating how best to generalise the approach into different settings for each child and family.2

Understanding Repetitive Behaviour supports families to begin to recognise and understand their child’s functionally impactful RRB, differentiate these behaviours from those that do not impact function, consider how they respond and the impact that their responses may have on their child and the behaviour. Firstly, parents/carers recognise and consider what the function might be of the identified functionally impactful RRB for their child in a particular context. For example, if a child responds negatively (e.g. becomes distressed or physically aggressive) to the front door being opened or the vacuum cleaner being turned on, then the function of the child’s behaviour may be to reduce or remove the level of uncertainty or the aversive sensory input. Alternatively, some RRB may provide an enjoyable pastime for the child. Functional analysis principles help parents/carers to understand their child’s RRB and to decide whether, as well as how, to intervene. The URB programme supports parents/carers to develop alternative strategies and techniques to reduce any negative impact from their child’s functionally impactful RRB across a range of everyday contexts. Strategies might include providing advance warning about unpleasant stimuli (such as vacuum cleaning) and a place to hide away for the duration; providing designated spaces for collections to be laid out undisturbed, while allowing other family members to move about the home freely; gradually building up tolerance to wearing certain items such as a coat during colder days; providing alternative items for a child who likes to touch people’s hair to hold or touch such as a doll; limiting the purchasing of new additions to collections to a specific frequency or context; providing alternate foods not surreptitiously, but overtly and with no insistence to try them. Such strategies are designed to provide autistic children with ways to manage uncertainty, cultivate interests and engage with the world safely and without resorting to extreme restriction as a coping strategy.

Parents/carers have opportunities for both individual support with the professionals trained to deliver URB and group learning. In weeks 2 and 7, they meet the group leaders individually to select a target functionally impactful RRB to focus on during the group and to apply new knowledge and practise the new skills they are learning, thus ensuring that strategies are individually tailored for each child and are developmentally appropriate.2

There are also opportunities for mutual learning and sharing of ideas. As part of the URB programme, parents/carers are asked to video their chosen/target functionally impactful RRB at home. These videos are then shared in the sessions, allowing parents/carers opportunities to discuss how they are recognising and learning to support their child. This fosters opportunities for parents/carers to learn from and support each other as they share their experiences as parents/carers of young autistic children. These videos are never shared outside the group.2

Psychoeducation Learning About Autism parent group (attentional control)

The LAA group acts as an attentional control for time and attention. LAA comprises eight 2-hour parent/carer group sessions focusing on understanding autism. The groups offer parents and carers psychoeducation on understanding autism and what that means for their child, such as helping them to understand their child’s social communication difficulties and behaviour in different environments. It also provides advice and guidance on strategies and approaches for dealing with, for example, behaviours that parents/carers find functionally impactful. There are opportunities for mutual support and sharing of ideas with other parents/carers, thus increasing parents’ and carers’ understanding, confidence and responsiveness to their child’s patterns of communication and interaction. This is equivalent to current best practice. It does not include any specific information about the role and functions of RRB, functional analysis and tailored strategies to manage functionally impactful RRB.

Study settings and sites

Both the URB and LAA programmes have been developed to be delivered in local community settings by early years professionals experienced in working with young autistic children and their families. The parent/carer groups took place in community settings in different geographical locations across the three sites in CNTW, Lothian and TEWV [or remotely via online delivery as a consequence of coronavirus disease 2019 (COVID-19) restrictions post March 2020]. We have carefully considered the level of professional expertise necessary to deliver both the URB and LAA programmes effectively and safely. Both programmes were delivered by staff with experience in working with young autistic children and their families, with appropriate training and supervision.

Participants

Parents and carers aged 18 years and older were eligible for study entry if their child met the following criteria:

1. Aged 3 to 9 years 11 months at the time of consent with a clinical diagnosis of autism or ASD across a range of functioning levels and abilities (verbal and nonverbal).
2. Parents/carers had to have sufficient spoken and written English to provide written informed consent and complete the assessments, including being able to identify one or more functionally impactful RRB and participate in the group-based programme.
3. Parents/carers had to be willing to be randomised and attend all the group sessions for the allocated arm of the study and agree to maintain their child’s current medication regime up to 24 weeks (unless change is advised by the child’s clinician) and to agree not to participate in any other trials while involved in the trial up to 24 weeks.

Parents/carers were not eligible for study entry if their child met the following criteria:

1. no clinical diagnosis of autism or ASD;
2. no functionally impactful RRB that could be identified;
3. currently taking part in another parent/carer group-based programme trial; or
4. had a sibling already taking part in this study.

Furthermore, parents/carers with a severe learning disability or a significant mental health disorder that would interfere with their ability to take part in a group-based programme were not eligible to participate.

Recruitment

Potential participants were identified through the three research sites, as well as through Participant Identification Centres (PICs). Clinicians were provided with information on the URB study and asked to introduce the study to families. This might be in person at a clinic appointment, in a community setting/school or by letter. Clinicians gave potential families copies of the patient information sheet (PIS) and an expression of interest (EoI) form to be returned by the clinician or the family to the research team using the stamped addressed envelope provided. Families were also invited to take part through diagnostic clinic databases and research databases such as ASD database-UK (ASD-UK). The local Clinical Research Network (CRN) at each site assisted with recruitment.

Interested parents/carers were asked to complete and return to the research team an EoI form. The research team then contacted parents/carers who were interested in hearing more about the study to arrange a meeting or phone call to discuss the study, what participation involves, answer any questions they may have, and obtain written informed consent via an informed consent form (ICF).

Newsletters summarising the progress and findings of the study were designed by the research team and parent/carer advisers. These were sent to families and local professionals who had taken part in recruitment and supported the study during the trial, to support retention.

Screening

Eligibility procedure

• Where there was any doubt, after first appointment or after all assessments, eligibility was discussed by the research associate (RA), the principal investigator (PI) at each recruiting site and other members of the senior research team.
• If the participant met the eligibility criteria, the RA sent a letter/e-mail thanking the parents/carers for participating in all the assessments, and to confirm eligibility and that the family had been put forward for randomisation. A separate letter was sent to the GP/referrer to notify them of this.
• If the child did not meet the eligibility criteria, the RA or the PI contacted the family to discuss.
• The RA or the PI also informed the referrer if the child did not meet the eligibility criteria.

Consent

At each recruiting site, once the EoI was received by the research team, parents/carers were contacted directly and an initial appointment with the RA made at a mutually convenient time to discuss the details of the information sheet and the purpose of the study. Parents/carers were given the opportunity to discuss any questions or concerns they might have to ensure that they were fully informed about the study. They were then asked to give informed consent at this initial contact appointment. A minimum of 24 hours was afforded for consent.

Randomisation

Randomisation was done through the Sealed Envelope (www.sealedenvelope.com/) web-based randomisation service. Randomisation was at child level using equal allocation ratio. Each parent/carer was automatically considered in their child’s randomisation group. We opted for child-level randomisation instead of parent/carer-level randomisation because the primary outcome is at child level, and it is important to account for child-level characteristics that can affect the primary outcome.

Age (3–5 years vs. 6–7 years vs. 8 to 9 years 11 months), gender (male vs. female) and ethnicity (white vs. non-white) were accounted for in allocating children to either the URB group or the LAA comparison group. Given the prevalence of RRBs across the autism severity range, severity was not accounted for in allocating groups. Due to the nature of the study and the few factors (age, gender and ethnicity) that needed to be accounted for in the randomisation, a minimisation scheme instead of stratified randomisation was used to minimise sample fragmentation due to the number of strata and to avoid accidental imbalance between the URB group and the LAA comparison group. Unlike stratified randomisation, minimisation works towards minimising the total imbalance for all factors together instead of considering mutually exclusive subgroups.

The clinical leads at each site informed the families of the outcome of randomisation by telephone and the randomisation decision was also confirmed by letter. This was also recorded in the child participant’s NHS patient record.

Participant allocation

Two hundred and twenty-seven parents/carers of young children (aged 3 to 9 years 11 months) with a clinical diagnosis of ASD across a range of functioning levels and abilities (verbal and nonverbal) were recruited through child health and child and adolescent mental health teams across three sites (CNTW, Lothian and TEWV). Following recruitment, the first 25 children were randomly allocated to receive the URB parent/carer group programme or the LAA psychoeducation group. For the minimisation scheme, the first child was randomly allocated to either the URB group or the LAA group using simple randomisation with 50% equal probability. The remaining 202 children were allocated to either the URB group or the LAA group by minimising marginal imbalance between the two groups (based on age, gender and ethnicity) and a prespecified probability of 10%. For example, if the second child to be randomised was a male, 3–5 years and from white background, suppose a white male child, aged 3–5 years was already allocated to the URB group. Although allocating the second child to the LAA group would lead to less overall imbalance score, he/she might be assigned to the URB group because of the prespecified probability of 10% to conceal the allocation.

Blinding

The RA at each of the three recruiting sites was based in a separate location (usually university premises) to the group leaders and remained throughout the study ‘blind’ to the group status of all the recruited children and families in the trial. The RAs were trained to reliability in all baseline characterisation and outcome measures. For each recruited child and parent/carer(s), once informed consent and all the baseline measures were completed and scored and eligibility confirmed by either the site RA or PI (see Eligibility procedure above), the participants were randomised. The site clinical leads were informed (in writing via e-mail) of the outcome of the randomisation. The parent/carer participant then received a telephone call and a letter from the clinical lead at each site confirming the randomisation outcome. The RAs were not informed of the randomisation outcome. Then prior to and at each subsequent follow-up visit, parents/carers were reminded not to disclose to the RA the child/family randomisation status. This means that throughout the trial the RAs remained blind to the randomisation status of all study participants.

Unblinding

Participants, group leaders and clinical leads were not blind to participant group allocation. All other members of the research team at each site were blinded. As such, there was no anticipation for the need of unblinding while the study was in progress. However, in the unlikely event of a RA being inadvertently unblinded to the randomisation status of an individual child, this was recorded on the trial database and the site clinical and research leads were informed. Where possible, future assessments with this participant were carried out by an alternative blinded RA.

Patient and public involvement

The URB programme and the RCT study protocol have been developed jointly in partnership with parents/carers of young autistic children through two development parent/carer groups and a pilot feasibility and acceptability study.22 The pilot study funded by National Institute for Health and Care Research (NIHR) Research for Patient Benefit (RfPB) programme included end-of-treatment participant focus groups led by trained parent/carer(s). The trained parent/carer focus group leaders followed a semistructured topic guide with the aim of considering three key areas of interest: experiences of participating in a research study, opinions about the URB programme and the impact of the URB programme on the participants, their children and the family. Fourteen parent/carer participants attended the focus groups. Participants were positive about URB, and they particularly liked the group-based format. Most participants reported that they had little knowledge of RRB before attending the programme and that it had had a positive impact on them, their children and their family. The two parent/carer focus group leaders have successfully co-authored a peer-reviewed publication on the findings from the focus groups.26

We had two parent representatives [one from the National Autistic Society (NAS) and one from our development work] on the URB RfPB Steering Committee. The findings from the NIHR RfPB pilot study have been disseminated by several of the co-applicants (including one of the parents who took part in the study) at local and national parent/carer and practitioner meetings/conferences.

Furthermore, we consulted with Professor Helen McConachie, Newcastle University regarding the design of our programme following the NIHR Health Technology Assessment (HTA MeASURe systematic review of tools to measure outcomes for young children with ASD).27 Professor Ann LeCouteur, Professor Jacqui Rodgers and Deborah Garland were co-applicants on this review.

We have continued to involve the UK organisation NAS in the research process, from the design of the LAA psychoeducation comparison group, to the planning and management of the delivery of the RCT study to the dissemination of research findings. We have also continued to work collaboratively with parents/carers throughout the design and conduct of the current RCT informed by INVOLVE guidance on training, support and remuneration of expertise to ensure parents/carers are included in a meaningful way. Two parent advisers were Trial Steering Committee (TSC) members for the pilot study and have continued to advise on recruitment and retention strategies and implementation of study procedures. We have convened a new parent advisory group, including autistic parents, to provide additional consultation to the research team with respect to interpretation of findings and dissemination.

Training and fidelity

All URB group leaders and site clinical leads attended a one-day URB training course covering how to use the training manual and the materials for each of the eight 2-hour weekly sessions, planning for homework tasks and supporting parents/carer in using the technology. Each group leader learnt about group processes and gained skills to deliver the manualised programme in a participatory style, in combination with strategies to personalise the programme for each parent/carer. The senior trainer (LD) alongside the chief investigator (VG) delivered the introductory training course and also visited each site as each new group had started. Furthermore, to ensure that the programme was being delivered to the families as intended with high fidelity to the manual, the senior trainer, alongside the local clinical lead, also undertook weekly supervision of group leaders at each site and monitored the quality of parent/carer group sessions by video, providing timely feedback to group leaders.

Group leaders delivering the LAA parent/carer group attended a one-day training course specifically designed for this study to ensure that they were trained in the skills needed to deliver the programme. This course had been designed by the NAS and was delivered by DG.

All parent group sessions in both arms of the trial were recorded, to facilitate opportunities for supervision of the group leaders and to provide access to a random sample of recorded sessions across all sites for the independent evaluation of fidelity. The fidelity checklists for each programme (URB and LAA) measured (1) the fidelity to the delivery of the programme (i.e. session structure, techniques and therapeutic components) and (2) the fidelity to the content of the programme.

Four independent raters were trained to use the URB and LAA fidelity coding checklists. The independent raters were randomly allocated 10% of the recorded parent/carer group programme sessions.

Parent/carer attendance at group sessions was recorded and monitored. Parents/carers who missed a URB session were contacted by a group leader to provide a brief overview of the session and handouts for that session were also posted out. For both groups (URB and LAA), session attendance and receipt of other programmes outside of the protocol were monitored.

Avoidance of contamination

There were separate clinical and research leads at each site and separate training and supervision structures. Researchers were located separately from staff involved in delivery of the URB and LAA groups. Research interviews and assessments were conducted mostly at home visits (or via telephone post COVID-19 restrictions being implemented in March 2020). All parents/carers were reminded prior to every contact with research staff not to talk about group allocation. Further, we also considered additional issues around potential contamination of the active attentional control LAA group, by ensuring that the LAA families were not on the clinical NHS caseload of the URB programme group leaders. The clinical and research leads in the trial were experienced in studies of this type and therefore were able to take additional steps, if needed, to discern and avoid any potential risk of contamination bias.

Outcome measurements

The clinical effectiveness of this programme was assessed by measuring whether children show overall improvement using the Clinical Global Impression - Improvement scale (CGI-I; primary outcome) at 24 weeks (primary end point) after the URB programme compared to the children whose parents/carers had attended the LAA group. There were several secondary outcomes that were measured at the end of the programme (10 weeks) and at 24 weeks and 52 weeks follow-up to capture independent (blinded researcher), teacher and parent-reported changes. These included change in functionally impactful RRBs. Parent/carer questionnaires also provided information on child RRB, child participation and daily living skills, parent/carer stress, self-efficacy and impact on family life. Parents/carers were aware of group status, but teachers and the primary outcome assessors (RAs) were blind to group allocation.

Baseline characterisation and outcome measures

The following baseline measures were collected:

• Demographics: a bespoke demographics tool was designed for the study. Parents/carers were asked about their child’s age, gender, type of nursery/school provision, diagnosis and ethnicity, current medication and additional diagnoses. Information was also obtained on parent/carers’ level of education, employment status, family structure, and if they had attended any previous course or intervention programmes for autistic children.
• Autism Diagnostic Observation Schedule-2 (ADOS-2):28 this is a standardised observational assessment undertaken by a trained examiner. It is a semistructured set of play and social communication activities that involves both specific activities and spontaneous social interaction between the examiner and the child. It is designed to elicit a range of behaviours including social communication skills and difficulties. Children were assessed with the developmentally appropriate Module (Module 1, 2 or 3 according to language level and chronological age). During the ADOS-2, elements of the child’s behaviour are observed and scored in two domains: social affect and RRB. The scores for the domains are combined into a total score. The total score is then used to derive a comparison severity score ranging from 1 to 10. Comparison scores of 1–2, 3–4, 5–7, 8–10 indicating minimal to no evidence, low, moderate and high levels of autism spectrum related symptoms, respectively. From March 2020 onwards, ADOS-2 assessments were not able to be undertaken once national governments’ and NHS COVID-19 restrictions were implemented.
• Social Responsiveness Scale – Second Edition (SRS-2):29 The SRS-2 (preschool form or school form according to child’s chronological age) is a 65-item questionnaire measure of the social impairments that are characteristic of ASD. It is completed by the parent/caregiver. Interpretation is based on a single score (total score) reflecting the sum of responses to all questions, which serves as an index of severity across the autism spectrum. A total T-score of 76 or higher is considered severe and strongly associated with clinical diagnosis of autism. T-scores of 66 through to 75 are interpreted as indicating moderate difficulties in reciprocal social behaviour that are clinically significant and lead to substantial interference in everyday social interactions, whereas T-scores of 60–65 are in the mild range and indicate mild to moderate deficits in social interaction. T-scores of 59 and below are considered to be within typical limits and are generally not associated with a clinical diagnosis of ASD.

Primary outcome measure

Clinical Global Impression – Improvement scale:30 The CGI-I provides a standardised framework for clinicians to assess how much symptoms have improved or worsened relative to the child’s baseline state using a 7-point scale (1 – very much improved; 2 – much improved; 3 – minimally improved; 4 – no change; 5 – minimally worse; 6 – much worse; or 7 – very much worse). Researchers and clinicians, blind to group allocation, were trained to reliably and independently rate global improvement over the 24 weeks (from baseline to primary end point), using all available child information from baseline, and week 24 [SRS-2 (baseline only), Vineland Adaptive Behaviour Scales 3 (VABS-3), target behaviour vignettes (TBVs), parent/carer and teacher Repetitive Behaviour Questionnaire – 2 (RBQ-2)]. In line with other published studies, ratings of 1 (very much improved) and 2 (much improved) are regarded as clinically significant ‘improvement’ and were used to define the binary outcome of improvement or no improvement for each child for the comparison of the URB intervention group and the LAA attentional control group at the primary end point.

For 10% of participants, the CGI-I score was independently rated by a second trained researcher to assess inter-rater reliability.

Secondary child outcome measures

Measurement of RRB – Target Behaviour Vignette:31 As part of the baseline characterisation, parents/carers were asked to identify two functionally impactful RRBs. Parents/carers were asked questions about the duration, impact and possible triggers and functions of the functionally impactful RRB using a standardised semistructured interview. The protocol for measuring change in the parent/carer defined TBV was originally developed by The Research Units on Paediatric Psychopharmacology and Psychosocial Intervention Programmes (RUPP Autism Network). At each outcome assessment point, the parent/carer completed the follow-up version of the standardised semistructured interview. The parent/carer responses at each time point were audio recorded and contributed to a vignette written by the RA (blind to randomisation status). Audio files were stored as a password-protected secure file with anonymised file name (using study ID code). In keeping with the procedure developed by RUPP, after all data were collected, a panel of blinded autism experts independently rated change in each target behaviour and change in relation to the impact on the family. Three pairs of vignettes [comparing each time point (10; 24; 52 weeks) to baseline] were rated for each child on a 9-point scale of improvement/deterioration (1 – very much improved; 2 – markedly improved; 3 – definitely improved; 4 – equivocally improved; 5 – no change; 6 – equivocally worse; 7 – definitely worse; 8 – markedly worse; 9 – disastrously worse). A positive response (also described as a ‘responder’) was defined as a rating of 3 or less.

Measurement of RRB – Repetitive Behaviour Questionnaire – 2 (RBQ-2):32 The RBQ-2 is a 20-item questionnaire completed by parents/carers that measures the frequency and intensity/severity of RRB known to occur in both autism and typical development. The RBQ-2 was developed using items from the original RBQ33 and the diagnostic interview for social and communication disorders (DISCO). The RBQ-2 has been reported to be a valid measure of RRB in a sample of children with ASD aged 2–17 years, showing good internal consistency.34 Scores range between 1 and 3 and a higher score is indicative of more RRB.

The Teacher Repetitive Behaviour Questionnaire – 2 (Teacher RBQ-2):35 This is the corresponding version of the parent/carer RBQ-2 for completion by teachers/teaching assistants. It measures the frequency, intensity and severity of RRB in a classroom setting. Scores range between 1 and 3 and a higher score is indicative of more RRB.

Vineland Adaptive Behaviour Scales 3 (VABS-3):36 The VABS-3 measures aspects of the child’s level of adaptive functioning. The parent/carer rating form was used. This focuses on four domains of everyday functioning: communication, daily living skills, socialisation and motor skills (0 for never, 1 for sometimes and 2 for usually or often). The assessment was undertaken with parents/carers by a trained researcher. The domain composite scores provided an adaptive behaviour composite (ABC). A higher score indicates a greater level of ability.

Secondary parent outcome measures

Secondary parent/carer measures were completed by the parent/carer who attended group sessions in both arms or, if both parents/carers planned to attend sessions, the nominated main parent/carer was asked to complete all parent/carer report measures.

Parent self-efficacy:37 This 15-item questionnaire completed by parents/carers measures behaviours typically exhibited by autistic children. Parents/carers indicate ‘yes’ or ‘no’ to whether the child displayed each of the behaviours in the previous month and then rate their confidence in managing the behaviours on a 6-point scale ranging from 0 (no confidence) to 5 (complete confidence). A mean self-efficacy score is calculated by dividing the total confidence score by the number of behaviours reported as displayed. A higher score indicates more parental self-efficacy.

Autism Parenting Stress Index (APSI):38 This is a measure of parenting stress specific to core and comorbid symptoms of autism. It was designed to be used to identify areas where parents/carers need support with parenting skills, and to assess the effect of the interventions on parenting stress. Exploratory factor analysis suggested three factors impacting parenting stress relating to core deficits, to comorbid behavioural symptoms, and to comorbid physical symptoms. Psychometric properties are good (e.g. Cronbach’s alpha 0.83). A higher score is related to more self-reported parenting stress.

Warwick–Edinburgh Mental Well-being Scale (WEMWBS)39 is a psychometrically robust parent/carer rated 14-item well-being questionnaire with good internal consistency (Cronbach’s alpha 0.89) and test–retest reliability (ICC 0.83). It is recommended by the Department of Health and Social Care as the preferred measure of mental well-being and that it is important to incorporate in parent-mediated studies where parental/carer well-being may impact on child outcomes. A higher score is an indicator of a greater level of parental well-being.

Secondary family outcome measures

Autism Family Experience Questionnaire (AFEQ):39 This questionnaire was developed to measure the broader impact of an intervention on young autistic children and on their families in terms of participation in everyday activities. It was commissioned by the Medical Research Council as part of the Preschool Autism Communication Trial40 and based on focus groups and piloting with parents/carers of young children with ASD to reflect what changes in their lives would ‘make a difference’. A lower score indicates a more favourable outcome.

Data collection

Baseline assessment and follow-up measures were collected by RAs blinded to the outcome of randomisation. RAs were trained to high levels of reliability in all baseline characterisation and outcome measures (Table 1). All families were allocated a unique number that was used to identify them on all paper assessment forms throughout the trial. All data collected on paper were inputted into a data management system for eventual statistical analysis and all identifying data were stored securely separately. The Clinical Data Management System (MACRO) used for this trial is fully compliant with all regulatory frameworks for research of this nature. Participants cannot be identified from electronic case report forms (eCRFs). The CI or delegated person monitored completeness and quality of data recording in eCRFs and corresponded regularly with site RAs, research leads and PIs (or their delegated team member) with the aim of capturing any missing data where possible and ensuring continuous high quality of data. All study data were treated in accordance with the latest directive on good clinical practice (GCP).41

Data management

Data were entered by sites on to the MACRO (Elsevier B.V., Amsterdam, The Netherlands) database and were checked throughout the recruitment period to ensure that the eCRFs were as complete and accurate as possible. There were two types of validation to ensure data integrity: manual and electronic. The following types of checks were performed: range checks, consistency checks, protocol checks and accuracy checks. All issues arising from the checks were queried with site staff. All changes to the data were documented in the audit trail including details of who made the change, when the change was made and why the change was made, to prove data integrity. Essential data will be retained for a period of at least 5 years following close of the trial in line with sponsor policy and the latest European directive on GCP.41 Data were handled, digitalised and stored in accordance with the Data Protection Act 1998 and the Data Protection Act 2018. In line with General Data Protection Regulations (GDPR), the sponsor was the data controller for this study and Newcastle Clinical Trials Unit (NCTU) was the data processor.

Trial management

The Data Monitoring Committee (DMC) was independently chaired and comprised of a panel of independent members including a principal clinical psychologist and a post-doctoral research associate as well as non-independent members that were part of the Trial Management Group (TMG). Serious adverse events (SAEs), and actions taken, were logged by the senior trial manager (STM) or trial manager (TM) at NCTU and a report presented to DMC. The DMC met once a year to receive reports on recruitment and SAEs. The DMC (which is independently chaired and includes an independent statistician) members evaluated the findings of the internal pilot in August 2019 and submitted a report to the NIHR.

The project also had a TSC with an independent chair and a panel of independent members including a statistician, health economist and two parent representatives as well as non-independent members, from the TMG. The TSC met approximately annually throughout the trial moving to bi-annually during the COVID-19 pandemic (Table 2).

The DMC and TSC members, appointed by the funder, were independent of the sponsor and funder and declared no competing interests. The TMG consisted of the CI, co-applicants, and members of NCTU including the TM, STM, clinical trial administrator and data manager, trial statisticians, health economists, members of each site, sponsor and CRN staff.

Throughout the duration of the trial, the RAs discussed any challenges with data collection with research leads at each site and the research team at each site. The RAs were responsible for ongoing review of data completeness and any concerns were discussed with the research team, NCTU and the sponsor, as appropriate. Throughout the latter stages of the trial, the TMG was split into a section specific to site and a core TMG where specific trial items were discussed.

The NCTU were responsible for oversight of the day-to-day running of the trial on behalf of the sponsor. The TM conducted several on-site monitoring visits throughout the course of the trial to ensure that all standards were met according to GCP and any issues or concerns were identified and rectified. Source data verification and review of the medical notes to ensure safety information was captured appropriately was also conducted. Throughout the pandemic, monitoring visits were conducted with one conducted remotely.

Statistical methodology

Sample size

We planned to approach approximately 325 families and expected to randomise 250 families (125 randomised to each arm). Assuming 5% type I error, 90% power, 10% intra-group correlation and equal allocation ratio, 224 families (an average of 8 families per parent/carer group) were required to detect 20% improvement rate between the URB programme and LAA group at 24 weeks. Allowing for an attrition rate of 12%, 250 families were expected to be randomised. The 10% intracluster correlation was based on review of group programmes in education trials.28 Sample size was calculated in R using n4props in Sample Size Estimation Functions for Cluster Randomised Trials (CRTSize) package.42

Statistical analysis plan

The analysis of the primary outcome at 24 weeks used generalised estimating equations (GEEs) with binomial distribution and logit link. Exchangeable working correlations were used to account for the clustering of children by parent/carer groups. The continuous secondary outcomes were first analysed at 24 weeks using a difference-in-difference model based on linear mixed-effect model accounting for paired data (at baseline and at 24 weeks) per child and clustering of children by parent/carer groups. The same model was applied to the data at week 52, which was analysed as longitudinal data incorporating data at baseline and 24 weeks. All binary or categorical secondary outcomes were analysed using GEE. We also performed safety analysis and sensitivity analysis for missing data and assessed the impact of the COVID-19 pandemic on primary and secondary outcomes.

All continuous variables were summarised using n (non-missing sample size), mean, standard deviation (SD), minimum and maximum. The frequency and percentages (based on the non-missing sample size) of observed levels were reported for all categorical data. All summary tables were structured with a column for each trial arm, and they are annotated with the total population size relevant to that table/treatment, including any missing observations. Primary analysis according to CGI-I at 24 weeks was performed using GEE with binomial distribution and logit link. We accounted for clustering of children by parents’ groups using exchangeable working correlation assumption and the analysis used the full analysis population and intention-to-treat (ITT) principle. We also adjusted for the minimisation variables [age (3–5 years, 6–7 years and 8 to 9 years 11 months), gender (male vs. female) and ethnicity (white vs. non-white)], as this has been shown to improve efficiency of statistical inference.

As part of sensitivity analysis for the primary outcome, we evaluated the impact of missing data using the so-called weighted GEE [weighted generalised estimating equations (WGEE)].43 This method works under the assumption of missing at random (MAR) missing data process, which is often the basis for carrying out multiple imputation of missing data. The main idea of WGEE is to supplement the outcome with a predictive model for the missing rates conditional on the observed ones. Due to insufficient observations, the subgroup analyses for age categories were not conducted. For the TBVs we compared change scores between the trial arms at 10, 24 and 52 weeks. The prespecified analysis is GEE, in a similar version to the primary analysis. Due to model non-convergence, we reduced the level of the models and analysed the data using logistic regression, with effects of interest captured using odds ratios (ORs). All the models were adjusted for the minimisation variables as prespecified. For the remaining secondary outcomes, we prespecified an alternative model to analysis of covariance (ANCOVA) using the linear mixed-effects models for mean difference as the change from baseline in the questionnaire measures with continuous scores. The latter has been shown to produce robust standard errors. The model is specified such that baseline values are part of the outcome vector, constraining the baseline across trial arms. In this formulation, the inference is not based on the point estimates of trial arms but the interaction between the trial arms and follow-up time points. All other secondary outcomes were analysed using appropriate models using the ITT principle. We also noted that the per-protocol population was essentially the same as full analysis population because in both all participants met the inclusion and exclusion criteria in the protocol, received the treatment group, completed 24 weeks follow-up, and did not deviate from the protocol as agreed with the TSC. Since there was no clear cut-off point to define adherence to the programme (dosage of URB), complier average causal effect (CACE) analysis was not performed, instead we provided descriptive statistics on attendance rates in parent/carer groups in the Results chapter.

The impacts of the COVID-19 pandemic that started during the course of this trial were assessed on the primary outcome and TBV using difference-in-difference models between the participants who had their 24-week (primary end point) visit pre and post COVID-19 lockdown starting on 23 March 2020 (see Appendix 4). This was specifically done by creating an indicator variable which takes values of 1 ('post lockdown’) for data collected after 23 March 2020, and values of zero (‘pre lockdown’) for the data collected before the lockdown. Although this study was not powered to detect interactions, a significant interaction between lockdown indicator and URB programme status would indicate that the outcomes differ between the pre- and during lockdown period.

Recording and reporting serious adverse events

For the purposes of this trial, serious adverse events [adverse events which met the criteria for seriousness (SAEs)] were captured for the parent/carer and child participants. SAEs were captured from the start date of participation in either LAA or URB until the follow-up assessment at week 24.

As soon as a site suspected that a SAE might be related to the trial intervention, they contacted the TM immediately. The reporting time frame started at day 0 when NCTU was in receipt of a minimum set of information:

• sponsor trial reference and trial name (sponsor reference)
• patient trial number and date of birth
• parent or child participant
• name of intervention
• date of notification of the event
• medical description of the event
• date and time of the onset of the event (including event end date if applicable)
• causality assessment
• seriousness of the event, particularly if life-threatening or fatal
• an identifiable reporter (e.g. PI).

This information had to be provided on the trial-specific SAE form. The site was expected to fully cooperate with NCTU in order that a full and detailed report could be submitted to the NHS Research Ethics Committee (REC) within the required timelines.

Events of special interest

As well as collecting and ensuring SAEs were reported, events of special interest (ESI) were also collected. An ESI was defined as any event relating to child well-being and family/life difficulties which was not expected and not anticipated in ‘normal day-to-day life’ but was not a physical medical event. ESIs were recorded for both the parent/carer and child participants from the start date of either URB or LAA until the follow-up assessment at week 24.

Compliance and withdrawal

Participants had the right to withdraw from the trial at any time without having to give a reason. Investigator sites tried to ascertain the reason for withdrawal and document this reason within the CRF and the child participant’s medical notes. For participants who withdrew, data captured up until the point of withdrawal were retained, unless participants withdrew consent.

The clinical lead at each site could discontinue a participant from the trial at any time if they considered it necessary for any reason including but not exhaustive to the examples below. This decision was initially made at each site but discussed with the wider research team if necessary. The CI and NCTU were then informed:

1. symptomatic deterioration
2. parent/carer withdrawal of consent
3. significant protocol deviation or non-compliance
4. investigator’s discretion that it is in the best interest of the child and/or parent/carer to withdraw
5. an adverse event that requires discontinuation of the trial programme or renders the child or parent/carer unable to continue in the trial
6. termination of the clinical trial by the sponsor.

Participants were classed as enrolled in the trial once randomised. Participants who withdrew from the trial after randomisation were not replaced.

Changes to study design

Several amendments were made throughout the course of the trial. Table 3 depicts the substantial amendments and their summary.

Sources of bias

There was no source of bias within this trial in relation to study design or recruitment of participants.

End of trial

The definition of end of trial was at the completion of the participant’s last follow-up visit at 52 weeks.