The aim of this study is to assess the feasibility and inform the design of a full-scale RCT of U-VBF for children with cleft speech characteristics. The primary aim of U-VBF is to enable learning of new articulatory gestures (new speech sounds), with secondary aims of improving intelligibility and health-related quality of life.
Design and Setting
Mixed methods will be employed. The study is a single blind pilot randomised controlled trial, with control offered U-VBF therapy at the end of the study. A qualitative study (focus group) of the acceptability of both interventions and the study design will also be undertaken. All intervention will take place at the speech and language therapy hub in the Royal Hospital for Children in Glasgow. Eligibility screens and pre- and post-intervention assessments will take place either in person in a university clinic room or via telehealth (ZoomTM or Microsoft TEAMSTM). The roles and responsibilities are outlined in Appendix 2.
No definitive comparisons of the interventions will be undertaken. The feasibility of a full scale RCT will be determined by evaluating a number of objectives against set success criteria (bulleted below) taken from a similar pilot RCT of children with speech disorders :
To determine recruitment and attrition rates
• 75% of children and their families identified agree to participate
• 75% of children allocated in each group are retained for the duration of the study
To measure pre-post and follow-up outcome measure completion
• 75% of outcome measures are completed
To measure within-session outcome measure completion
• Data is reported from 75% of intervention sessions
To determine acceptability of randomisation to children and their families
• 75% of children and their families rate randomisation as acceptable in a questionnaire
To determine the acceptability of ultrasound visual biofeedback as an assessment tool (both groups) and intervention tool (U-VBF group)
• 75% of children and their families rate ultrasound as an acceptable technique in a questionnaire.
• Focus group analysis contains more positive than negative themes regarding acceptability
To measure adherence to the treatment protocol
• 75% of sessions reach the minimum dosage of 100 trials in both treatment arms
A single-centre two-arm parallel group pilot randomised controlled trial with blinded assessors will be carried out. Cases will be stratified by three age groups (4;6 – 7;11; 8;0 – 11;11 and 12;0 – 16;0). Due to the nature of the U-VBF, therapists and patients will not be blinded to treatment allocation but the limitations of this will be mitigated by the use of an assessor, blinded to group, and evaluators, blinded to both group and treatment time point. Unblinding to the assessor or the evaluator will not be permissible and is considered unnecessary given the clinicians delivering the intervention are not blind to treatment allocation.
Children managed by the Scottish Cleft Lip and Palate service aged 4;6 to 16 will be identified and recruited when attending SLT clinics at the Royal Hospital for Children in Glasgow. Invitation letters containing the complete study information sheets will be sent to families. Inclusion criteria regarding age and cleft-type are broad to reflect current clinical practice, however, inclusion criteria for type of cleft speech characteristic are narrower to ensure children are likely to benefit from either U-VBF or the control intervention. Children are eligible if they have at least one speech error that would normally be a candidate for articulation intervention. We aim to recruit 20 children to each arm of the trial. Recruitment will stop once the target number of participants is recruited and within two years from the beginning of the study. No interim analyses will be performed before all data is collected. Children with an uncorrected bilateral hearing loss of greater than 30dB (from previous reports), planned surgery within the next three months, or severe language deficit (from previous SLT reports and a receptive vocabulary standard score<70 on the BPVS-3,  ) will be excluded. Children will be allowed to continue all other medical and speech and language interventions during the trial. We will collect information on any other speech and language interventions received during the trial.
Following baseline, the children will be randomised by the Glasgow Clinical Trials Unit in a 1:1 ratio, stratified for age. Children randomised to the control arm will be offered U-VBF at the end of the trial if they still present with CSCs which are candidates for U-VBF as previous studies have shown this improves the acceptability of a randomised trial to families .
We will screen potential participants from case-notes and invite them to attend an initial screening and baseline assessment. The person carrying it out will obtain consent, ensuring the participants and their families have had time to read, understand and discuss the information about the study, prior to any assessment or procedures. This assessment will be either in person or via video conferencing, with in-person preferred. Screening assessment will comprise the British Picture Vocabulary Test 3  to screen for adequate receptive vocabulary; and a speech assessment protocol to determine whether patients present with at least one cleft speech characteristic which would be amenable to both U-VBF and the control intervention. This assessment protocol comprises the Diagnostic Evaluation of Articulation and Phonology  articulation and phonology subtests and an ultrasound tongue imaging protocol designed in a previous project  to identify covert speech errors from consonants at all places of articulation and sentences from the GoS.SP.aSS.’98/ CAPS-A  (Appendix 1). Families who opt for the assessment over video-conferencing will complete the same assessments, but the ultrasound tongue imaging protocol will be replaced with a audio-perceptual assessment of the same materials.
Speech Target Selection
Children with CLP may present with multiple CSCs affecting intelligibility. We intend to select as intervention targets speech sounds which are 1. amenable to treatment with both interventions and 2. likely to have the biggest functional impact on intelligibility. Following the screening assessments, we will select wordlists targeting each child’s specific lingual errors from a battery. In English, lingual speech sounds (imageable with ultrasound and amenable to treatment with both interventions) are /t,d,n,r,l,s,z,∫,ʒ,tʃ,dʒ,j,k,ɡ,ŋ/ and all vowels. Children with CLP are more likely to have difficulty with anterior consonants /t,d,s/. Wordlists containing these speech sounds will also form a key outcome measure (see below). Two wordlists per error type will be selected, firstly an “untreated probe” (i.e., the words will not be used in the course of therapy, this allows us to check for generalisation of targets). Secondly a “treated probe”, containing high frequency and functionally useful words will be used to train speech targets in the course of therapy. The wordlists contain lingual targets in increasingly complex contexts from single syllable words/pseudowords through to multi-syllabic words and sentences. Where the child has more than one error, multiple wordlists will be used, however only one treatment target (the speech sound with the most errors) will be selected and the other errors will serve as “control segments” (i.e., speech sounds that should not improve during the course of therapy unless maturation is a factor). Wordlists will be analysed for percentage target consonants correct (PTC). Children must score <30PTC at baseline to be eligible for the study.
Both interventions will be delivered by the cleft palate specialist SLTs in the Royal Hospital for Children in Glasgow. Therapy in both treatment arms will be once per week for six sessions with each session lasting up to 45mins. The number of sessions is pragmatic in nature, reflecting current practice, and is designed to highlight initial response to both interventions. It is likely that some children in both arms of the trial may require further speech intervention in the future (after the follow-up measures are taken) and this will be provided in line with standard practice, which includes further U-VBF. Both interventions will focus on acquisition of new speech sounds. Previous studies of children with non-cleft speech disorders show that a new speech sound can be acquired within one to two sessions  of U-VBF for most children but that some children take four to six sessions. Target articulations will be decided individually. Both interventions begin with a pre-practice phase where the aim is to teach the child to approximate the target articulation before they can begin the practice phase where at least 100 repetitions are required for learning and generalisation. This dosage will be measured in both interventions. In this pilot we will focus on both the pre-practice phase and the first stage of practice: acquisition of a new sound in simple contexts such as “ta, tea, toe” building to short words such as “tap, team, tore” as this is feasible within six intervention sessions. If children are super-responders (i.e., they quickly retain the new speech sound and are able to produce it in complex contexts) then the protocol will also allow us to measure this. Participants will be discontinued from the intervention if they show any adverse effects to the ultrasound treatment- in rare cases the ultrasound gel can cause contact dermatitis 
Articulation Intervention (AI)
This intervention involves working on a single speech sound at a time. The SLT uses modelling, demonstration, verbal description, and feedback in the pre-practice phase to teach the child the new sound at first in limited contexts and then in words and finally in conversation in the practice phase. To increase parity with U-VBF, and in line with newer theories of motor learning , we will standardise the in-session dosage during the practice phase to at least 100 trials, i.e. each child will be given 100 attempts to articulate their target articulation in each session.
Ultrasound Visual Biofeedback (U-VBF)
This intervention is grounded in the principles of motor learning. The patient sees a real-time image of their tongue moving and guided by the SLT uses this biofeedback to learn a new articulation, building productions to increasingly complex contexts, as in AI. Again, a minimum of 100 trials are required in the intervention. The intervention is set out in an open access manual  and involves using the software SonospeechTM. The software has functionality to be used as an assessment and intervention tool, allowing the SLT and patient to record and playback ultrasound video with synchronised audio or to view it live. The clinicians delivering the intervention have completed training and currently use ultrasound in their clinical practice.
A blinded assessor will collect measures at baseline (-t1: pre-treatment), 2 months post-randomization (t7, to allow for any delays in referral to therapy), and 3 months post-randomization (t8) to see if any benefit is maintained. Assessment will take place in a university clinic or via video-conferencing and will be carried out by a research SLT blinded to group for primary outcome measures and by the treating SLTs for within-session measures. See table 1.
Within-treatment session outcomes
Previous research suggests that one of the main benefits of U-VBF may be efficiency rather than overall efficacy of treatment . We will therefore measure treatment response during each session, rather than just after the course of treatment. The treating SLTs will therefore audio record short treated, and untreated word lists at every treatment session (t1-t6). From this we will determine how quickly children achieve a new articulation as a measure of response to treatment. These will be rated at the end of the project by SLTs blinded to group.
Candidate Primary Outcome Measures
The key primary outcome linear measure for change in speech will be percentage target consonants correct (PTC). We will measure this at single sound level (stimulability and in /aCa/ contexts); single word level; and sentence level in treated and untreated wordlists (see above section on speech target selection). All direct speech measures will be recorded with audio and where possible ultrasound tongue imaging in both groups, allowing us to perform ultrasound analysis of data from both groups.
Patient Reported Outcome Measures
We will use the Intelligibility in Context Scale  as a carer reported outcome. This short scale asks parents/carers to rate how easy to understand their child is to a variety of listeners ranging from family members to strangers. It has been shown to have a high internal reliability and construct validity . Quality of life will be measured using the CLEFT-Q speech function and quality of life scales for children aged 8 and over . This instrument has good content and construct validity, and good reliability, as established using a large and diverse international sample of participants . We will also use the Experience of Service Questionnaire  to measure patient and carer satisfaction with both interventions at the end of the project. It was shown to have good construct validity and precision in measuring satisfaction with care . The bespoke data collection forms, such as the participant information sheets and the consent form templates, can be found in a publicly accessible folder on the University of Strathclyde open access Knowledge Base “PURE”.
Intervention Acceptability Measures and Qualitative Evaluation
Families will complete a questionnaire about the acceptability of both interventions at the end of the study. These questionnaires will also be offered to families of participants who drop out of the study. Participant retention will be promoted by dedicating time to discuss any questions and problems that may have arisen for the participant and their family at the start of each session and by offering further ultrasound or articulation intervention after the final follow-up if indicated. Parents/carers and children over 12 will be invited to join focus groups to discuss their experiences of taking part in the trial and to contribute to planning a larger trial. Each focus group will include up to 10 participants. Responses to the focus groups will be analysed using thematic analysis.
Table 1 shows the timeline for the project, including the timing of each assessment.
Analyses and Statistical Power
Definitive comparisons of the interventions will not be undertaken due to the feasibility nature of the study. Details of patient screening, recruitment, retention, withdrawal, follow-up will be summarised (see research questions above). Adherence to U-VBF will be measured according to the number of patients who complete the intervention in accordance with the treatment manual. Adherence to treatment dosage will be recorded using an intervention pro-forma in each session where the SLT records a tick mark for each trial (i.e., each time the patient attempts to produce the speech sound in treatment, this should be around 100 for both interventions). All sessions will be audio-recorded (consent permitting) and 20% of sessions will be fidelity checked for dosage and adherence to protocol.
All of the speech measures at baseline, and follow-up will be recorded with simultaneous ultrasound in both groups by a research SLT blinded to group where possible. Our previous work showed that the addition of ultrasound to transcription increases inter-rater reliability and allows identification of covert (imperceptible to the ear alone) errors . This will allow us both to calculate PTC (the primary outcome measure) with increased reliability and to perform an error analysis. 25% of the data will also be rated by two specialist cleft SLTs (not involved in the project) trained in ultrasound-aided transcription. These SLTs will also rate the audio recordings from the within-treatment sessions (t1 to t6 in Table 1), blinded to group.
All data will be stored in a dedicated secure shared drive managed by the University of Strathclyde where it will be automatically backed up. Hard copies of data will be kept at the Hospital and the University of Strathclyde. Participants will be pseudo-anonymised during the project, with a key linking codes to actual names held securely at the University. The key will be deleted at the end of the project and data will become anonymous. Participants and their families will be informed that recordings of voice can be identifiable.
Based on previous studies using ultrasound intervention for speech sound disorders, no harm to the participants is anticipated. In rare cases, some people may have an allergic reaction (contact dermatitis) to the gel . If this is observed the gel can be substituted with water. If water does not achieve a sufficiently good ultrasound image, then the participant may be discontinued from the trial. If any adverse consequences are observed, they will be recorded in the participant’s clinical notes and reported to the NHS ethics committee using standard procedures. The ethics committee and the University ethics committee can audit the trial at any time.
The results from this study will be published in academic journals, such as The Cleft Palate Craniofacial Journal, Clinical Linguistics and Phonetics. Additionally, results will be presented at academic and clinical conferences such as the International Congress of Cleft, Lip and Palate and Related Craniofacial Anomalies, and the conference of the Royal College of Speech and Language Therapists, where it will reach both an international and a local audience of clinicians.
The protocol, statistical code and the anonymized final numerical dataset will be made available publicly on the University of Strathclyde’s open access repository “PURE”. Protocol changes will be reported to the trial registry and the ethics committee.
 The assessment sessions are not part of core clinical care and therefore we will offer these over video-conferencing to comply with any current local Covid restrictions under which travelling for research assessments may not be an “essential journey”.