A Randomised Controlled Feasibility Trial of Immersive Virtual Reality Treatment with Cognitive Behaviour Therapy for Specific Phobias in Young People with Autism Spectrum Disorder

This was a single blind RCT comparing a virtual reality treatment for specific phobias in children with ASD with usual care (immediate and control (delayed) treatment arms). After consent and baseline measures were taken, participants were randomised to one of these arms.

The immediate treatment group were followed up at 2 weeks and 6 months after treatment. The control arm completed outcome measures at these time points. The control group then received the treatment after completing the measures at 6 months (see Consort Diagram); after their treatment they were subsequently followed-up at 2 weeks and 6 months post treatment. The immediate treatment group had one additional follow-up at 12 months post treatment.

Thirty-two young people with ASD were recruited over a 12 month period from two UK mental health services—Northumberland, Tyne and Wear NHS Foundation Trust and Tees, Esk and Wear Valley NHS Foundation Trust.

Inclusion criteria: Age 8–14 years, diagnosis of ASD, verbally fluent and able to understand instructions to enable treatment participation and completion of outcome measures. Verbal fluency and comprehension were determined as part of usual clinical practice by the child’s clinical consultant, who judged the child’s suitability for the study. All participants had a specific phobia as identified during the Anxiety Diagnostic Interview Schedule (ADIS) interview (see below). Exclusion criteria were participants whose phobia could not be visually represented and addressed in the Blue Room (e.g. phobia of injections), children with severe and complex anxiety disorder, and/or children with a learning disability that the referring mental health clinician judged would affect their ability to participate.

The child’s usual mental health clinician initially discussed the study with the family and then completed an expression of interest form and the Children’s Global Assessment Scale score (CGAS—a brief measure of emotional and behavioural functioning, with range 1–100) (Shaffer et al. 1983). All participants had a confirmed ASD diagnosis according to DSM-IV or ICD-10 criteria from NHS multidisciplinary diagnostic teams.

On receiving the Expression of Interest form, a member of the research team contacted families to evaluate whether their specific phobia was suitable for treatment and for computer generated, graded scene design. Forty families were identified by clinicians; five of these did not have a phobia that could be treated in the Blue Room and three declined to take part after receiving further information. Thirty-two families received a preparatory home visit where baseline measures were completed and the family shown a video of the Blue Room VRE; written informed consent was obtained. Table 1 shows the characteristics of participants. The two groups were well matched for mean age and gender and on outcomes captured by the standardised measures. Table 2 shows the specific phobias that were identified by participants and their families as the target to address during the Blue Room treatment.

Maskey et al (2014) reported that Target Behaviour ratings are appropriate outcome measures of real life change in behaviours of concern. This is further supported in a review of the treatment of specific fears and phobias in children with ASD (Lydon et al. 2015), where target behaviours were found to be the primary outcome measure in 10 of the 16 studies reviewed. This measure was therefore included as the main outcome measure, alongside questionnaires, to investigate the utility of different outcome measures with this clinical population.

For the treatment and control groups the following outcome measures were collected:

Target Behaviour rating This measure recorded a rating of change over time in the specific phobia to be tackled through Blue Room treatment. The protocol was developed by the Research Units on Paediatric Psychopharmacology (RUPP) Autism Network (Arnold et al. 2003). Questions regarding the child’s specific phobia behaviours, and questions such as ‘how often?’ and ‘how distressed?’ were asked in a standard format to the parent, enabling a vignette to be written. Following an interview about phobia during the initial home visit, pre-treatment vignettes were written prior to randomisation, by the first author (for examples see Table 5). These baseline vignettes are not rated per se, but rather serve as a baseline from which Target Behaviour change is measured. Follow-up vignettes were written from telephone interviews with parents, undertaken by the blinded outcome assessor. All efforts were made to maintain the assessor as blinded. Each vignette pair (baseline vs post treatment vignettes) was evaluated for change over time on a 9 point scale (from ‘normalised’ to ‘disastrously worse’). An expert panel of raters received training with examples, before rating the pairs of vignettes. Raters were blind to group allocation and time point, and each vignette pair was rated by four different raters. Arnold and colleagues (Arnold et al. 2003) reported an Intraclass Correlation Coefficient (ICC) of 0.895 for a panel of 5 experts; in this study ICCs for the two time points were 0.869 (95% CI 0.775 to 0.930) and 0.935 (95% CI 0.887 to 0.965). The Target Behaviour rating is reported both dimensionally, and also by categorical cut-score using a mean of 3.0 or less, corresponding to a rating of ‘definitely improved’ or better, to define positive treatment response (‘responders’). In addition, a cut-score of 6.0 or more was used in this study to define those whose symptoms had worsened compared to baseline.

Spence Children’s Anxiety Scale-parent version (SCAS-P) and child version (SCAS-C) The SCAS (Spence 1998) was developed to assess anxiety symptoms in children in the general population and has 38 items with a 0 (never) to 3 (always) scale. The measure has been widely used in ASD studies (Sofronoff et al. 2005; Maskey et al. 2014). High internal consistency for the total scale score has been reported (Spence 1998), and both convergent and divergent validity (Nauta et al. 2004). In the current study, internal consistency at baseline was α = 0.900 for SCAS-P and α = 0.863 for SCAS-C.

Fear survey schedule for children—revised (FSSC-R) (Ollendick 1983) This is an 80 item parent-report questionnaire with an overall intensity and fearfulness score. The FSSC-R is the most commonly used tool for assessment of common fears and phobias, with good construct, convergent and divergent validity (Gullone et al. 2000) and strong test–retest reliability and internal consistency (Burnham and Gullone 1997). In the current study, internal consistency at baseline was α = 0.932.

Children’s Assessment of Participation and Enjoyment (CAPE). This was completed by the child at baseline and 9 months and intended to measure any increase in participation in community activities. CAPE is a 50 item child-report of activities, presented pictorially, to assess children’s participation in a range of solitary and group voluntary activities. Reliability and validity of the CAPE was established through study of 427 children with disabilities (King et al. 2007).

Attendance: whether children attended all sessions was recorded.

Confidence ratings: during treatment children rated their confidence at tackling their goal situation at the beginning of session one, end of session two, beginning of session three and end of session four. Parents rated their perception of their child’s confidence at parallel time points. Ratings were from 0 (not comfortable) to 6 (very comfortable); parent and child ratings were taken in separate rooms and not shared. Examples of a confidence scale used are given in a previous publication (Maskey et al. 2014).

When families were approached but chose not to participate, ethical permission was granted to pass the following anonymised data to the research team: CGAS score, age, gender, ASD diagnosis and type of diagnosis. This was to allow ‘refusers’ to be characterised and compared with trial participants.

Participants were randomly allocated to immediate treatment group (n = 16) or control group, for whom treatment was offered after the 6 months outcome measures were administered (n = 16). Allocation was by computer using a password-protected Newcastle University Clinical Trials Unit website. Randomisation was by mixed block design, using block sizes of two and four, stratified by site. Due to the nature of the treatment, participants, clinicians and the main researcher for the study were aware of group allocation. Participants were randomised and informed after the initial home visit as to whether or not they were in the immediate treatment group. Another researcher, blind to treatment allocation, conducted outcome measurements through telephone discussion of target behaviours with families and receiving postal questionnaires. Blinding was strictly maintained; this outcome assessor had no other trial role, no access to documents and did not attend trial meetings. At each telephone or postal contact, this outcome assessor reminded parents she was unaware of group allocation.

The Blue Room VRE is a patented immersive technology using interactive computer generated audio visual images projected onto the walls and ceilings of a 360 degree screened room (Fig. 1). The room was 4 m³ and the participant and therapist sit side by side. A therapist remains with the participant throughout the treatment sessions, delivering the CBT techniques (described below). Scenes are individualised, incorporating an exposure hierarchy related to the feared stimulus. For example, for dog phobia, adaptions include the dog’s size, whether on or off a lead, barking, and proximity to the participant. This gradation allows the participant to experience levels of mastery in managing their anxiety and to repeatedly practice this at one level of challenge before moving to the next (Maskey et al. 2014). The following link shows a session in progress: https://​www.​youtube.​com/​watch?​v=​9U-rRC8jc28.

Before VR sessions, each participant and parent attended a 45 minutes session with their allocated therapist. The therapist was a health professional (for example an assistant psychologist, or a specialist nurse) with experience in ASD and/or CBT, who had attended the training workshop (see below). Simplified CBT techniques were introduced, including: (1) identifying feelings (how different parts of the body feel; how thoughts, emotions and behaviours are connected); (2) the concept of a visual ‘feeling thermometer’ using the participant’s words to describe anxiety; (3) two relaxation exercises (muscle relaxation and deep breathing, with scripts for home practice); (4) identification of the participant’s positive coping statement, e.g. ‘I can do this’, ‘I’m going to be ok’, to use in the treatment sessions. These CBT elements were repeated and consolidated during VRE sessions. The goal for the end of treatment was agreed with the participant; this goal was used for the confidence rating charts for parent and participant.

Following scene creation, participants attended 20 min treatment sessions. Two Blue Room treatment sessions were completed at one visit, with a fifteen minute break between. The second two sessions were conducted around one week later. The therapist allocated to the participant was present during all sessions. Parents watched treatment via a video link, and the session content and purpose of activities was explained. For the first two sessions, a supervising qualified clinical psychologist attended to observe and give feedback to the therapist. This supervising therapist also answered any questions the family had during the session.

Materials for the treatment sessions (the treatment manual, customised visual scales and relaxation scripts) were provided to therapists. Each Blue Room session started with a relaxation scene, allowing the participants to become familiar with the environment, and to practice relaxation techniques and coping self-statements. The two available relaxation scenes were of swimming dolphins, and a field in the country; scenes had soft background music that could be turned off if requested. The duration the child spent looking at these scenes and practising relaxation exercises was at the discretion of the therapist, as the aim was to be responsive to the needs of individual participants; for most participants, one cycle of muscle relaxation and breathing exercises was sufficient at the beginning of each session. The relaxation scenes were returned to during a session if the therapist thought the participant’s anxiety was severe or if the participant was finding it difficult to manage their anxiety during a particular scene.

Following the relaxation scenes, the participant was introduced to the VRE scene designed for them. The initial scene was designed to be the lowest level of the exposure hierarchy for that participant (e.g. a quiet dog on a lead in the distance). The participant gradually moved through the hierarchy; they progressed to an increased level of challenge when they were consistently reporting low levels of anxiety (score of 2 or less on a six point scale) on the visual scale for a scene, and there was agreement between the participant and therapist to move on. At each level of the hierarchy, participants were supported by the therapist to practise techniques to reduce anxiety, including relaxation exercises, thought challenging and anxiety monitoring. If anxiety increased as the scene became more challenging, the therapist suggested relaxation and breathing exercises, moving to the relaxation scenes if needed. Progress through the scenes was determined by the progress participants made towards maintaining low anxiety at each level and was individualised to each participant. After completing the fourth session, the therapist spoke with participants and their family regarding graded real world exposure to the anxiety situation. The therapist explained the need to gradually introduce the participant to the feared situation in real life and discussed various steps in the hierarchy of exposure relevant to the particular phobia.

Before delivering treatment, local therapists read the manual and attended a 2 hours, group training workshop delivered by an experienced child clinical psychologist (author 10). The manual for the treatment is copyrighted and is available on request from the corresponding author. Training involved discussion of why children with ASD may develop anxiety, explanation of the steps in treatment including evolution over four sessions, review of video material from live sessions, and individual practice with the tablet computer in the VRE.

All Blue Room treatment sessions were video recorded. A sample of 30% of sessions in the immediate treatment group were rated for fidelity to delivery as per the manual. The sessions were chosen at random but always included at least one session conducted by each of the 11 therapists, and an even spread of VRE sessions 1 to 4. Fidelity was recorded on a checklist to assess (a) Delivery of CBT best practice and (b) the manual Content and Structure [checklist designed by author 2, drawing on sources including Roth & Pilling (Roth and Pilling 2008)]. Delivery ratings included Techniques used (9 or 10 elements e.g. collaborative approach, modelling reflection, using relaxation strategies, using praise), Generic Acceptable components (5 elements e.g. therapeutic alliance, managing emotional content, appropriate flexibility) and Undesirable components (6 elements e.g. didactic approach, allows off-topic deviation); Content and Structure included around 10 elements (e.g. setting agenda, summarising, scenes presented in increasing levels of difficulty). Rating definitions and number varied between sessions for Techniques and Content, as different elements were introduced or became irrelevant. Delivery ratings were: 0 (not at all), 1 (minimal evidence), 2 (several examples) with ratings reversed for Undesirable components. Content ratings were: 0 (not covered), 1 (covered insufficiently) and 2 (covered adequately). Senior co-authors (authors 2, 3 and 10) established mean inter-rater agreement at 83.6% for Techniques, 96.0% for Acceptable and 92.2% for Undesirable components. Agreement of mean 69.7% (range across raters and sessions 56–94%) for Content and Structure was lower so was not rated further. Content rating proved difficult for several reasons: the quality of audio in recordings; some aspects perhaps being covered outside the VR; expectation that CBT would be flexibly individualised.

Analysis was conducted according to a pre-specified statistical analysis plan. Post hoc testing of the main outcome measure found sensitive to change in our development study (Target Behaviour rating) was then conducted to explore potential efficacy.

Analysis was undertaken by author 4, blind to group status and supervised by author 2.

Group equivalence at baseline was investigated using Fisher’s exact test, Pearson’s chi square and t-tests. Exploratory group comparison over time was made for the Target Behaviour ratings using Mann Whitney U test and chi square. Descriptive statistics (mean, standard deviation) are presented for all questionnaire data, along with effect size (Cohen’s d), where values 0.2, 0.5 and 0.8 indicate a small, medium, and large effect size respectively.

Collection of 12-month post treatment data from the treatment group: To undertake a preliminary investigation of the medium term treatment effectiveness, children’s outcomes for the immediate treatment group were investigated 12 months post treatment. Target behaviour follow-up vignettes were written following telephone interviews with parents; these were undertaken by the same blinded outcome assessor and vignettes written (questionnaire data were not collected at this timepoint to reduce the burden on parents). Vignettes were rated as described above by the blinded expert rater group. Descriptive data are reported separately from the main trial outcomes.

Treatment outcomes for the control (delayed treatment) group: After the wait phase and completion of measures at 6 months, parents and children from the control group were offered VR treatment sessions. 15/16 parents accepted; one child was not treated as following discussion with the local treating clinician, it was agreed that due to family circumstances and generalised anxiety symptoms, phobia treatment was not the main priority. Fifteen children received treatment according to the treatment plan described above. Outcome measures were collected at 2 weeks and 6 months post treatment, and target behaviour interviews were undertaken with parents; vignettes were written by the same outcome assessor, who at this stage was not blind as only control group data was being collected at this point. Questionnaire data were not collected. Vignettes were rated as described above by the expert rater group, who remained blind to group status. Descriptive data are reported separately from the main trial outcomes.

A favourable ethical opinion was provided by the UK National Health Service Tyne and Wear South Ethics Committee (reference 14/NE/1177). The research sponsor was Northumberland, Tyne and Wear NHS Foundation Trust. The trial was conducted and reported in accordance with CONSORT guidelines, and was registered on the ISRCTN trial database (Trial ID: ISRCTN7886185). The trial team and co-applicants met regularly during the study. A parent advisory group met regularly to advise on trial procedures.