Study design
The study is a randomized controlled trial comparing speech and language treatment as usual with a music assisted programme for learning language. The online version of the free open-source Python programme, MinimPy [19], will be used to assign the participants randomly to either the music-assisted (MAP) intervention or treatment as usual (TAU) group using minimisation randomisation. This method prevents selection bias and minimises the imbalance in prognostic factors of participants between treatment groups [20]. Gender, developmental stage (IQ >= 70, or 50-69), and echolalia (present or absent, as reported by the parents) will be used to balance participants between groups.
Participants
Thirty children with ASD will be recruited. Participants will be screened for autism spectrum disorder using the Social Responsiveness Scale-2 (SRS-2) [21]. They will be between 24 and 60 months old, with little or no spoken language, defined operationally as fewer than 20 functional words [22]. Adapting Dawson et al. (2010: e18) [23], the following will be exclusion criteria:
a neurodevelopmental disorder of known etiology (e.g., fragile X syndrome);
significant hearing or motor impairment;
major physical problems such as a chronic serious health condition;
seizures at time of entry;
history of a serious head injury and/or neurologic disease:
IQ below 50 as measured by the Vineland Adaptive Behavior Scales – Third Edition [24];
not meeting the cut-off score for ASD on the SRS-2.
The participants will be identified from special education preschool provision, National Health Service (NHS - state funded), voluntary groups, social media and privately funded clinics in the United Kingdom. The parents of potential participants will be approached by letter or email initially and asked to complete an online questionnaire assessing the child’s eligibility and asking for consent to take part in the trial. All data will be held on a secure password protected database accessible only to the research team.
Ethics
Ethical approval was granted by the European Research Council (ERC) ethics review panel (Reference no. MAP, 838787), the NHS (National Health Service) HRA (Health Research Authority) and HCRW (Health and Care Research Wales) Approval service (Reference no. 262697) and the University Research Ethics Committee of the University of Reading (Reference no. UREC 19/07). Any changes to the protocol will be submitted to the ERC, NHS HRA, HRCW and the University of Reading Research Ethics Committee.
Since 24-60-month-old nonverbal and minimally verbal children are unable to give informed assent, consent will be sought from their parents, legal guardians, or caregivers. They will be sent via email a copy of the information sheet to read, and then be verbally briefed via video-conferencing through Microsoft Teams by the research speech and language therapist. If they agree for their child/dependent to participate, they will be given a consent form to sign. Children's assent will be monitored by their behaviour and responses to their parent/carer during the sessions. Sessions will cease immediately if children become distressed.
Interventions
To ensure social distancing as required by UK Covid-19 regulations, assessments and interventions will be carried out by the parents under online guidance from the research speech and language therapist using Microsoft Teams. The interventions will consist of 36 training sessions of 45 minutes duration delivered at the rate of 2 sessions a week for 18 weeks, during which the parent will be coached by the therapist carrying out a range of activities with the child. In addition, parents will be encouraged to carry out a minimum of 5 ten-minute practice sessions with their children on per week. Each week, one such session will be recorded and jointly reviewed by the therapist and the parent to enable better tailoring of the treatment.
The MAP intervention will be as follows (Protocol version MAP-005): 1. Naturalistic strategies will be used, such as incidental learning, high density repetition, time-delay and mand-modelling. 2. For each of the 36 target words, we will create a set of songs providing the contexts where it occurs. During each session, the songs will be played using a computer or a phone, and sung with a range of home-made music instruments such as shakers. 3. The children will be taught to sing the songs, in which the target words will be occurring repetitively, together with other engaging and interactive activities such as dancing, vocalising, improvising, and playing musical games. 4. Parents will be taught strategies including intensive interaction and communication temptations to help them engage with their children. Parents will also be provided with an app (software programme for Android phones) written to provide the target words songs. The app enables tracking of the amount of practice sessions (i.e., without the therapist present) undertaken by each parent-child dyad.
For the treatment as usual (TAU) group, the speech and language therapy will be modelled on that provided in normal clinical practice for young children with little or no spoken language in the University of Reading Speech and Language Therapy clinic but delivered via online methods (Protocol version MAP-005). It will be based on Social Communication Intervention for Preschoolers (SCIP) [25] modified to teach the 36 target words identified. SCIP is a naturalistic developmental intervention for the core features of autism and uses a parent-mediated approach as recommended by the National Institute for Health and Care Excellence [2] guideline on psychosocial interventions for children with autism. The approach teaches parents child-centred, play-based strategies to improve their children’s joint attention, engagement, reciprocal social interaction, communication and language. Video feedback and coaching is used to improve a parent’s synchrony with their child’s interaction, to teach them how to develop reciprocal play routines, and how to provide general language stimulation.
Feasibility measures
The study is designed to estimate recruitment and retention rates from multiple referral sources. One aspect of these estimations is the feasibility of recruiting a narrowly defined group of preschool children with ASD. The initial screening is carried out by a therapist working jointly with parents to determine the history and skill levels of the child. We will gather information about the proportion of referrals that meet the criteria for inclusion and the effect of intake assessment on retention. Once the caregivers have completed the screening stage, they will then be randomized to the two interventions. Since the MAP intervention is novel, the trial will document changes to the methods and assess acceptability of the intervention to the caregivers and children. In part this will be through careful documenting of reasons for withdrawal.
One outcome measure (words learnt from a restricted list, Table 1) may need to be revised in light of evidence about numbers of words already known by the participants and which words are observed to attract most attention and learnt most rapidly during the intervention. If words on the target list are already known by the participant, we will provide alternatives from the list in Table 2.
Finally, we plan to assess the burden of the follow up measures through a qualitative study of a subsample of the participants. The qualitative study will use semi-structured interviews to examine the experiences of participants and their caregivers about the duration, intensity and practicalities of treatment and assessment via video link with the person’s home.
Baseline and outcome measures
Outcome measures will be collected at four time points: pre-, mid-, post-intervention, and 3-month follow-up (see Figure 1 and the SPIRIT checklist [26] in Supplementary Material). The measures will be administered by the research therapist using telecommunication methods including video conferencing technology to direct the caregiver to manipulate materials appropriately. The measures can be successfully administered with non-verbal and minimally verbal children and can ultimately be easily integrated in clinical and research settings. They will be scored by the research therapist and checked by a research assistant who will be blind to allocation. However, the parent administering the measures will also have been using the intervention methods with their children and thus there is very limited blinding possible in this trial. Demographic details and a medical history will be obtained at intake to the trial.
- One key outcome is the expressive (picture naming) and receptive (pointing to correct picture) tasks to assess the production and comprehension of the 36 target words at baseline, post-intervention, and 3-month follow-up. A preliminary list of 36 target words has been drawn up (Table 1), but the protocol includes 35 backup words (Table 2) and allows substitutions of up to 10 words if required by the prior skills of the child;
- Social Responsiveness Scale - 2 (SRS-2) - Participants’ ASD symptomatology will be assessed using the Social Responsiveness Scale – 2 [21]. The SRS-2 is a questionnaire that uses a scale of 1–4 (1 = not true, 4 = almost always true) to rate items, with higher scores indicating greater impairment [27,28]. A designated raw score cutoff value of 70 is considered to have a sensitivity of 0.78 and specificity of 0.94 for ASD [29]. Raw total scores are converted to gender-normed T scores, with a T-score of 75 indicating severe impairment. Item-level and full scale (both raw and T scores) scores were included in analyses.
- Receptive and expressive language skills will be measured using the Expressive One-Word Picture Vocabulary Test–4 (EOWPVT-4) [30] and the Receptive One-Word Picture Vocabulary Test-4 (ROWPVT-4) [31], and by MacArthur Bates Communicative Development Inventories (CDI): Words and Gestures Forms [32]. The EOWPVT-4 and ROWPVT-4 tests give raw score which can be used to determine standard score, age-equivalent score and percentile rank. The tests are appropriate for use with typical and atypical populations, including those with learning difficulties. The CDI has high parent-teacher agreement for children with ASD (Intra Class Correlation 95% CI estimates 0.77 to 0.93) [33].
- The child’s social communication will be evaluated during a 10-minute video recorded free play session between the child and their parent. Social communication will be measured using the Parent-Child Dyadic Interaction Measure which has good intrarater (κ = .81) and good interrater reliability (κ = .80) [34].
- Functional language will be also evaluated during the 10-minute video recorded free play session between the child and their parent. Casenhiser et al.’s (2015) [35] approach to coding language functions will be used which shows good internal consistency (Cronbach’s α range 0.830-0.836) as is the validity compared with other measures of communication level. A week-long language diary to be filled out by parents at pre-, mid-, post-intervention, and 3-month follow-up will also help determine the child’s progress on language development.
- Communication, daily living skills, socialisation, and maladaptive behaviour will be measured using the comprehensive parent/carer form of Vineland Adaptive Behavior Scales – Third Edition (VABS-3) [24], to be filled out by parents/caregivers. The VABS-3 has high internal consistency (Cronbach’s α range 0.86-0.97), good test-retest reliability (corrected r values range 0.62-0.92), and good inter-rater reliability (range 0.61 to 0.87). Data show that it has criterion validity for identifying those with intellectual disability.
Data handling and analysis plan
All electronic data will be stored in a secure password-protected folder to which only the trial team will have access. Video recordings and questionnaire-based outcome measures will be coded and scored by research assistants who are blinded to the study purpose, treatment allocation, and testing order.
We will record and report the participant flow according to CONSORT guideline and produce a CONSORT flowchart [36,37]. As a feasibility trial, we will report recruitment and retention figures for both MAP and TAU together with reasons for loss of participants. The trial management will consider any adverse effects and use that information to continue or halt the trial. In addition, we will discontinue the trial at the request of the caregivers, ongoing distress on the part of either the child or the caregiver, and if the caregiver takes up an additional language intervention during the trial. We expect to analyse recruitment and retention data using descriptive statistics involving both intention to treat and actual completed participant data. Recruitment and retention will be encouraged by payment of £202.50 ($283.00) to participants families for completing the research. Imputation methods will be used to estimate secondary variables missing data [38,39]. An important part of the feasibility is the extent to which it is possible to conduct two training sessions per week and still maximize the homework practice. We will therefore review the uptake/usage pattern of all MAP parent-child dyads to establish the feasibility of delivering 36 sessions over 4 months. Descriptive statistics focusing on mean, variance, confidence interval and standard deviation will be performed on the outcome measure data. The results from these analyses will provide data for effect size estimation. Together with the trial parameter data (i.e., recruitment, retention, follow-up, and completion rate), these data will be used to determine the size of sample necessary to carry out a fully powered randomized controlled trial comparing MAP with SCIP. A preliminary categorical analysis of words learnt may be used to inform the selection of future vocabulary items for a subsequent trial. The numerical data will be analysed blind to allocation. The blinding will be carried out once the data is collected by generating a new series of participant codes, removing data that could lead to identification (TW) and then be analysed by the PI (FL). Anonymised behavioural data and statistical analysis codes will be made available following open access guidelines.
The qualitative interview data will be transcribed and analysed using the thematic analysis framework [40] and Nvivo 12 [41]. Research assistants involved in transcription and analysis will be blind to the allocation of participants.
If during the course of the trial, we make observations which raise concerns about the children’s wellbeing and safety, we will follow the procedures and policies as outlined by the Keeping Children Safe Standards in the EU [42], the Safeguarding Vulnerable Groups Act 2006 in the UK [43], and the University of Reading DBS policy [44] and safeguarding policy [45] to respond to any safeguarding concerns. There will be no data monitoring committee because the number of participants is small and the primary data offer few opportunities for analysis methods. The trial therapists (MJ and MK) are supervised by TL and SF-M as part of the clinic quality control procedures.
Dissemination policies
The aim of dissemination will be to inform clinicians and parents of the potential usefulness or otherwise of the music assisted programmes and the app that we are able to develop. This will be achieved through presentations to parent and voluntary groups. Other outputs will include conference presentations for organisations involved in the care and education of preschool children with ASD, and scientific conferences about interventions for ASD. A paper will be written for a peer reviewed publication which we will make available cost free online.