Study setting

We have three participating study sites in Washington (WA), California (CA), and Ohio (OH), all with active cleft and craniofacial teams in tertiary-care hospital settings. We seek to enroll a sample that is representative of the demographic characteristics of our patient populations. For example, 37% to 60% of the patients seen in our centers receive Medicaid funding for their healthcare. Approximately 18% of our patients are Latino/Hispanic, and 27% are non-white. After English, Spanish is the most common language spoken by families. Therefore, we translated study materials and ensured that our team included bilingual/Spanish-speaking staff to facilitate the inclusion of Latino/Hispanic families. Translation procedures included initial translation by an outside service and independent back-translation by a separate translator who had not seen the original English versions. Bilingual/Spanish-speaking investigators from our team and content experts then reviewed all translated materials. We made final modifications to ensure the original meaning was retained and that translations were appropriate for a broad Spanish-speaking population. In some instances, we revised the wording of English materials to ensure comprehension and consistency across languages.

Eligibility criteria

Patients with CP±L ages 24- to 32-months will be eligible if their parent consents to participate and will be available for the duration of the study. We will include participants who received palate surgery at least six months before enrollment, primarily speak English, Spanish, or both languages in the home, and have access to a smartphone or similar device (e.g., tablet) capable of recording and sharing video. Exclusions are diagnosed conditions that affect neurodevelopment (e.g., 22q11.2 deletion syndrome), profound sensory impairment (e.g., blindness, deafness), delivery < 32 weeks’ gestation, placement in foster care or state custody, or a diagnosis of autism spectrum disorder (ASD). Questions related to eligibility will be reviewed by one of the co-investigators (EG) and principal investigator (BC), and eligibility decisions will be documented to ensure consistency.

Staff from each site will complete an initial screening with parents to verify eligibility and obtain informed consent. Once parents consent to participate, baseline measures will be completed, including a demographic and clinical interview and parent-report measures (child expressive language, shared reading frequency). Parents will use their smartphones to record a video of themselves reading with their child and complete a speech task in which the child names a series of pictures. To ensure that families are able and willing to complete study procedures, we will only randomize participants if they complete at least one shared reading survey and upload at least one parent-child reading video at baseline. Parents will use a free app or online website to assess the need for broadband or cellular connectivity support to determine upload and download speeds. As needed, we will assist families with connectivity resources (e.g., a "hotspot") to ensure access to the intervention.

All participants will receive an end-of-study assessment. This will include any participants found to be ineligible after enrollment (e.g., those with new diagnoses) and participants who are not adherent to the intervention protocol or assessments.

Intervention

All participants will receive five age-appropriate children’s books that they will be encouraged to read with their children. We will ask participants in both groups to video record themselves reading these books with their children and complete surveys of shared reading frequency. Families will use a study app to securely upload book-sharing videos with the research team. This will help to control for possible effects related to being observed. Children will continue to receive SOC treatment, including speech-language therapy and developmental interventions at enrollment or initiated during the study, and follow up with their craniofacial providers.

Participants randomized to the BOOST intervention group will receive support and encouragement to promote shared reading and dialogic book-sharing skills. During the study protocol development phase, we pilot-tested the BOOST program with 40 families recruited from participating centers and made minor modifications based on parent and interventionist feedback. The program is delivered in three 50-minute live, face-to-face telemedicine (Zoom) visits scheduled every 2–3 weeks. The BOOST program encourages parents to increase the frequency of shared reading, introduces skills to promote child engagement and involvement in reading, and offers strategies to promote language development. Depending on parents’ preference, we will provide English or Spanish teaching materials for each of the key skills taught. Interventionists will review the book-sharing videos parents upload between sessions to identify parent strengths, application of target skills, and opportunities for improvement.

Study assessments

After enrollment, we will complete a semi-structured caregiver interview to assess family demographics, including race and ethnicity, languages spoken in the home, parent education, parent occupations, maternal age at delivery, parent marital status, family income, and number of children and adults in the home. We will also obtain parents’ reports of their child’s history of developmental interventions, including speech and language therapy, physical therapy, occupational therapy, hearing services, and participation in birth-to-three intervention services. We will obtain parents’ permission to abstract information from children’s medical records, including gestational age at birth, birth complications, cleft diagnosis, age at cleft diagnosis, results from hearing evaluations, age at cleft lip surgery (when applicable), age at cleft palate surgery, other surgical procedures (e.g., tympanostomy tube placement), findings from speech assessments, and frequency of craniofacial clinic visits.

End-of-study outcomes will be assessed at least eight weeks after the intervention period (i.e., after the final parent-child reading video is uploaded). Fig 1 provides a schematic of study procedures, and Table 1 provides a sample participant schedule. All participants will have the opportunity to receive an end-of-study assessment, including any found to be ineligible after enrollment (e.g., those with new diagnoses) and participants who are not adherent to the intervention protocol or assessments.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Flow diagram with BOOST study procedures and timeline.

https://doi.org/10.1371/journal.pone.0304630.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Participant timeline.

https://doi.org/10.1371/journal.pone.0304630.t001

Our primary outcomes are children’s expressive and receptive language. Based on studies of children without orofacial clefts, DBS interventions result in moderate improvement in both domains, usually measured using single-word expressive and receptive vocabulary measures [13]. We have incorporated multiple assessment methods for these outcomes, including standardized clinician-administered measures, parent report measures, and observational coding of naturalistic language samples. Staff completing all outcomes assessments will be unaware of children’s randomization. Project staff completing standardized measures will be asked to practice and record an assessment with a non-study child in the target age range. These administrations will be reviewed by one of the investigators (AJ) and scored for reliability before certifying examiners to test RCT participants. During the study, we will video record every tenth administration for reliability scoring. Similarly, staff completing observational coding will practice on sample videos with scores compared to a "gold standard" (LS) to ensure reliability before coding videos for RCT participants. During the study, at least 20% of observations will be coded by more than one observer to assess inter-rater reliability.

Secondary outcome measure

Intermediate and exploratory outcomes

Throughout the study, we will use observational coding of parents’ book-sharing videos to assess the quality and frequency of book-sharing interactions in the BOOST and SOC groups. These are proximal targets for the intervention, hypothesized to mediate changes in children’s language outcomes. We also plan to code children’s language during book-sharing interactions, exploring whether intervention changes children’s vocalizations (e.g., number of unique words used, mean length of utterance) during reading.

Sample size considerations and participant recruitment and retention

We determined the sample size (N = 320) based on the number of participants needed to ensure 80% power to test the hypothesis that children in the BOOST intervention group will have better receptive and expressive language skills than those in the SOC comparison group. Power analyses were based on two-sample unequal-variance t-tests, an alpha = 0.05/2 to adjust for multiple comparisons, and up to 20% attrition (i.e., final N~256). Under these assumptions, we will have 80% power to detect treatment group differences of >0.4 SD for our primary outcomes. Treatment effects this large have been observed in DBS studies with other non-cleft populations and are considered plausible and clinically meaningful [13].

Our assessment of longitudinal changes in child language during book-sharing interactions (Hypothesis 1b) will use a generalized estimating equations (GEE) approach to linear regression with robust standard error estimates. The outcome for these analyses will be observed child language, calculated as an average number of utterances, unique words used, and mean length of utterance. Analyses will test the effects of the intervention on children’s language during book-sharing interactions over the five observations. We will have > 80% power to detect changes with an effect size > 0.4 between any two time points and changes with effect sizes > 0.3 overall. Meta-analyses of similar interventions with non-cleft participants suggest that effect sizes this large or larger are observed and considered clinically meaningful. Similarly, we will have > 80% power to assess longitudinal changes in children’s language during book-sharing interactions, the quality and frequency of book-sharing interactions, and changes in parents’ book-sharing behaviors as a mediator of differences in children’s language outcomes.

Recruitment and retention

Collectively, participating centers see approximately 200 new patients with isolated clefts each year. We anticipate a 50% consent rate, allowing us to enroll our target sample with up to 4-years of recruitment. Sites will complete ongoing reviews of medical records to identify potentially eligible participants in our target age range. Study flyers will be made available in the clinic, and information will be shared among cleft team providers, such as nurses, speech-language pathologists, psychologists, social workers, pediatricians, geneticists, and surgeons, so they can also inform families about possible participation. We will approach potentially eligible participants in the clinic or by phone when children are 24 to 32 months old. Parents who express interest will complete a brief interview, in person or by telephone, to confirm eligibility. In addition to receiving children’s books in their preferred language, families will receive incentives for completing study activities for sending shared reading videos ($40 per video shared) and completing the final evaluation ($150). We will also provide the results of children’s end-of-study language assessments in a brief report and encourage parents to share the results with their child’s pediatrician and other providers.

Study visits will be scheduled at times convenient for participating families to facilitate retention. We will use multiple strategies to communicate with families, including text and e-mail reminders for study visits. Participants will be considered "lost to follow-up" if they cannot be contacted, do not respond to outreach attempts, or cannot schedule their child’s outcome assessment within six months of the end of the intervention or active participation (controls). To minimize attrition, we will ask families for the name and contact information of an alternate contact who could help us locate them. We will also use formal and informal search strategies (e.g., online white page searches, Omnitrace) to find families as needed. We will make at least four attempts to contact families using several modes of contact, such as phone, letter, text, and e-mail. Our sample size of N = 320 provides adequate power to address the study aims with 20% attrition.

Randomization

Half the participants will be randomly assigned to the BOOST Intervention Group (n = 160). The other half of the participants (n = 160) will be randomized to a SOC comparison group that receives the same children’s books, uploads shared reading videos, and responds to surveys but does not receive individualized BOOST intervention sessions. Randomization will be stratified by study site, families’ preferred language (English or Spanish), and Medicaid status. Randomization will be blocked in blocks of 4 to 6 participants to ensure reasonable balance in small strata.

A data manager will prepare randomization lists for all sites using REDCap under the direction of the study biostatistician. These lists will be concealed from all other study staff. Once a patient has been determined eligible for participation and randomization, the site coordinator will enter the participant into the study database and receive the treatment assignment via REDCap.

Staff completing outcomes assessments (e.g., observational coding of parent-child interactions, child language assessments) will be unaware of group assignments. Masking will not be feasible for some procedures that, by definition, reveal participants’ group assignment (e.g., observational measures of treatment fidelity). We will use restrictions on user privileges in REDCap to limit access to this information in project databases, and we will use password/code protections to restrict access to calendars, data archives used to store videos, and any other information that might reveal group randomization. Masked staff will not participate in meetings where group assignments may be revealed. If an assessor or observational coder becomes aware of a participant’s group assignment, we will identify another staff member to complete the assessment/coding.

Data management

We plan to use direct entry into centralized REDCap databases to reduce the use of paper record forms and data entry steps. Before enrolling participants, staff involved in data entry will pilot test databases with sample record forms to ensure accuracy and assess any database issues that need to be addressed. Regular study coordinator teleconferences will address questions throughout the study and ensure continuity across centers. All data entry forms and databases will be password protected, with access restricted depending on staff roles/responsibilities. The study team and data entry staff will ensure that data entry is accurate and complete. Seattle Children’s Research Institute staff will run monthly reports to identify missing data with logic checks to assess potential data entry errors. The investigator or designee will review unanticipated problems and adverse events.

Statistical methods

In analyses for our primary outcomes, we will use an intention-to-treat (ITT) approach to test the hypothesis that children in the BOOST group exhibit better language skills than children in the SCO comparison group. We will calculate 2-sample unequal-variance t-tests to estimate treatment versus control group differences in child language outcomes post-intervention. We will calculate standardized treatment effect sizes (ES) for all analyses to allow comparison across outcomes with different measurement scales. We will not adjust for covariates in the primary analyses. However, in sensitivity analyses, we will use bias-corrected ITT analyses to examine the robustness of treatment effects in the setting of non-adherence, missing data, or study attrition. To assess attrition bias, we will explore outcomes using Inverse Probability Weighting (IPW), with weights determined by the probability of having missing data or being lost to follow-up based on the RCT stratification variables and demographic and clinical data [39]. We will also conduct sensitivity analyses to examine the effects of variation in treatment fidelity using a weighting approach. Finally, we will examine the potential effects of children’s participation in other interventions in the community (e.g., speech and language therapy) using censored normal regression (CNR) [40].

Longitudinal analyses examining changes in child language during book-sharing and caregiver behaviors (i.e., quality and frequency of book-sharing) will use a GEE approach to linear regression with robust standard error estimates. This approach accounts for the correlation of observations over time and allows for the use of partial data on subjects who are lost to follow-up or non-compliant. Estimates will be adjusted for the RCT stratification variables, demographic confounds, and clinical characteristics (e.g., cleft type, age at cleft repair). Smoothing splines will be fit to examine changes in parents’ reading behaviors over time, including non-linear trends.

We will use a multi-step linear regression approach to examine whether changes in parents’ reading behaviors mediate differences in children’s language outcomes. Mediators include: (1) shared reading frequency, (2) overall quality of parent-child interactions, and (3) parents’ shared reading skills, which will be assessed throughout the intervention period. Mediators will be analyzed separately to determine their individual effect on child outcomes and to avoid a washout effect from including all mediators in a single model. We will not adjust for multiple comparisons for this aim. Each regression model will follow the same basic approach to analyzing the effects of adjustment for a given mediator (Z) on the association between the intervention (X) on the child language outcome (Y). We will run two models per outcome: (1) Y = α + βX + e1, and (2) Y = A + BX + CZ + e2. The terms e1 and e2 represent model errors. Model 1 describes the unadjusted effect of the intervention on the outcome. Model 2 illustrates the combined effect of intervention and mediator on outcome. A comparison of the coefficients β and B indicates the degree of mediation. For example, β = B >0 suggests a positive intervention effect that is not mediated by Z (i.e., the null hypothesis of no mediation), whereas β >0 and B = 0 reflects a positive effect fully mediated by Z. Intermediate values reflect the degree of mediation (e.g., B = β/2 represents 50% mediation). Confidence intervals for the degree of mediation will be obtained using the bootstrap methods.

Finally, we will explore demographic and clinical characteristics associated with "good" versus "poor" treatment responses. In these analyses, we will examine parent reading behaviors and child outcomes in relation to clinical and demographic characteristics (e.g., language spoken in the home, family composition, parents’ education level, the child’s cleft type, child’s hearing status, sex of the child, study center/interventionists). We will use a two-step process for model selection and analysis. First, we will identify children who score > 0.5 SD above or below the mean relative to their intervention group. Second, we will examine the distribution of each characteristic of interest and identify those that (a) differ in distribution between those who score above-expected values, below-expected values, and as-expected and (b) have sufficient cell sizes (n >30). Selected variables will then be included in multivariate linear regression models to evaluate whether parents’ reading behaviors and child language outcomes vary across subgroups. We will calculate correlations between each variable before fitting any model, combining or excluding variables to avoid multicollinearity. To prevent overfitting, no more than 12 terms will be included in the model.

Study monitoring, ethical considerations, and dissemination

Ethical considerations and consent/parent permission

A possible breach of confidentiality is a particular concern for our study, given the use of in-home video recordings and study visits. We have taken several steps to protect confidentiality. In our video recording instructions for participating parents, we ask that they ensure that recordings do not capture anyone else in the home. We ask that parents record in a private setting and inform others in the house when they begin recording. We developed an app for data collection that allows caregivers to review all videos before sharing them with our team. Videos are securely shared to a file that is only accessible to research team members and saved in a password-protected file. Parent-child book-sharing videos will be coded and retained until child participants turn 18. Any future use would require separate IRB approval. Once participants turn 18, we will permanently delete recordings. Study visits are conducted using a HIPAA-compliant platform (Zoom). Sessions are video recorded for review of treatment fidelity, and videos will be permanently deleted once the recordings are coded. Other steps to protect confidentiality include using study identification numbers instead of identifying information. Only certain study team members can access the "link" between these study identification numbers and participant identifiers. During consent, we inform parents of the potential limits to confidentiality (e.g., instances of suspected abuse or neglect).

We established single IRB reliance for all sites, with Seattle Children’s being the primary IRB for the study. The Seattle Children’s IRB approved the protocol, informed consent form(s), recruitment materials, and all participant materials on November 7, 2024 (IRB# 00004565). Other participating centers are were activated on May 10, 2024 (Nationwide Children’s Hospital) and May 16, 2024 (Children’s Hospital Los Angeles). An IRB-approved consent and parent permission form describing the study procedures and risks will be sent or given to participants in their preferred language (English or Spanish). Our team includes multiple bilingual members, and we will ensure that a team member fluent in Spanish is available to review the consent and answer any questions for Spanish-speaking families. Before completing any study-related procedures, we will document consent in writing or using e-consent.

Per NIH guidelines, we will make de-identified data available to other research groups with appropriate IRB approvals, and results will be submitted to ClinicalTrials.gov. We have established an analysis and dissemination plan for our team members who are interested in new analyses. This includes generating a brief (1-page) proposal to be reviewed by the PI and site PIs to assess the scientific merit of the analysis proposal and any potential conflicts with other work in progress or planned analyses. Once approved, analysis proposals will be sent to all investigators to invite participation as co-authors on the writing team. Site PIs can nominate other trainees, staff, or collaborators who can be invited to participate in dissemination activities. We will provide potential co-authors with guidelines from the International Committee of Medical Journal Editors (ICMJE) and ask for confirmation that they meet the criteria for co-authorship. Authorship order will be determined based on contribution [41]. We will prioritize opportunities for new investigators and trainees to serve as first authors and site-PIs to serve as senior/last authors.

Limitations

Anticipated limitations and potential pitfalls for our study include limited generalizability to patients with clefts in conjunction with known genetic syndromes and other malformations. These groups are known to be at greater risk for poor neurodevelopmental outcomes, and we anticipated that their needs and response to intervention would likely differ. It is also possible that some patients will receive an exclusionary diagnosis (e.g., ASD) after study completion since these conditions may not be identified by 24- to 32-months. Developing similar interventions for children with syndromic conditions is a high priority. However, in this initial study, a more clinically homogeneous patient group was needed to establish intervention efficacy. Including English and Spanish-speaking participants complicates our outcomes assessment, given potential differences in language development for multi-lingual children. We have accounted for this by including language as a stratification variable in randomization, helping to ensure that the groups are balanced. The intensity of our intervention is another consideration, and we acknowledge that three 50-minute visits are modest. The results of meta-analyses have been somewhat contradictory on the minimum duration needed for similar interventions, with some indication that interventions of 90 minutes or greater produce significant treatment gains and others suggesting that longer duration is necessary for populations at significant risk for language delays [13, 14]. Ultimately, a lower-intensity intervention was needed to deliver the program at scale while minimizing participant burden and maximizing subsequent adaptation into standard cleft care. Finally, given the young age of our participants, there is a risk that our outcome measures will not be sensitive enough to detect treatment gains. Our multi-method assessment, including caregiver reports, clinician-administered measures, and observational measures of outcomes proximal to the intervention (e.g., changes in caregivers’ reading behaviors and changes in observed child language), is intended to address this concern.

Conclusion

The BOOST study will contribute to the limited literature on parent-focused interventions to promote language development in children with CP±L. The study will also advance knowledge of remote intervention delivery, with applications for geographically dispersed patient populations that would be difficult to serve with in-person developmental interventions.