Exploring social network factors impacting the implementation of communication supports designed for minimally verbal autistic preschool children: a study protocol Authors: Sarah Dufek, Aubyn C Stahmer, Giacomo Vivanti, Elizabeth McGhee Hassrick Published: 2025 ABSTRACT AbstractBackgroundAutism is a lifelong condition and can lead to particularly severe disability for minimally verbal (MV) children. Prior research suggests that early intervention is critical for autistic children; yet long-term outcomes have been variable. This variability may be due to poor implementation of evidence-based strategies within the child’s community, however research examining this idea is limited. This study protocol aims to see how collaborative dynamics within the team of caregivers can impact intervention effectiveness in MV preschoolers with autism.MethodsUsing social network analytics, we will be looking into collaboration within and between home and school teams for the adoption of Alternative and Augmentative Communication (AAC) strategies. This study will involve 40 MV autistic children from two sites and their respective caregivers/identified providers. We will collect social network data from each child’s parent and preschool teacher to assess collaboration across home, school and community. Data collection will be both quantitative, using social network surveys, and qualitative, using focus groups, to track the social dynamics of adoption and diffusion of AAC knowledge among caregivers, school staff members, community providers and supporters.DiscussionOur goals include finding an association between team density and the adoption of AAC strategies across settings, identifying barriers to collaboration among home/school team members, and exploratory child outcome analysis. We expect that stronger network collaboration between all caregivers will correlate with a better implementation of AAC strategies and improved child outcomes. We hope that these findings can better inform communication interventions for MV autistic children, enhancing their quality of life and easing the burden of care on loved ones.Keywords: Autism, Social network analysis, Alternative and augmentative communication (AAC) strategies, Caregiver collaboration, Childhood intervention, Implementation science, Egocentric networks Introduction Autism prevalence impacts approximately one in 36 children in the US[1]. Autism can result in lifelong disability, which can contribute to physical, emotional and economic burdens for autistic people and their family members [2]. Intervening in childhood can change life course trajectories of autistic individuals [3] and result in significant cost savings [4]. However, even when empirically supported strategies are deployed, outcomes for autistic children receiving intervention are variable, ranging from dramatic to modest improvements, with consequences on long-term well-being and quality of life [5, 6]. Differential response to treatment is particularly concerning for minimally verbal (MV) children with autism who do not improve in their communication skills during school years. MV children with autism are more likely to develop severe behavioral disturbances, resulting in greater safety risks, tremendous emotional, and economic burdens for their caregivers as well as enormous societal costs [7, 8]. One factor that impacts variability in intervention outcomes is poor community implementation of evidence-based strategies [9–11]. In particular, it is plausible, based on research from other fields, that poor collaboration and coordination of care contributes to variability in intervention response for this population [12]. However, no research has addressed the role of collaboration dynamics in autism intervention. Collaboration may be especially important in autism due to the need for integration of multiple strategies for care, the importance of caregiver implementation of strategies, and the multidisciplinary nature of the team required for addressing the children’s complex needs. Literature shows that both caregivers and intervention providers value collaboration[13–15], and preliminary evidence suggest that caregiver-professional collaboration affects treatment quality [15] and goal attainment [16] for children with autism. However, when different stakeholders [caregivers and professionals (including teachers, occupational therapists, speech pathologists, behavior analysts)] involved in the intervention – from now named as the “network members” - seek to collaborate, they face many challenges. Collaboration among network members requires robust financial, logistical and social resources which are often scarce in schools[17], and caregivers and providers often do not have the time, support, and training they require to work together efficiently[18], even though legislation mandates caregiver involvement [19]. Additionally, evidence suggests that current practices to establish and maintain collaboration within intervention network members are often ineffective or mismatched to caregivers’ unique needs [20]. In autism, treatment is often provided by multiple agencies and many disciplines whose theoretical orientation can at times be in conflict. While it is plausible that these issues affect autism outcomes, research is limited, partly because of the challenges inherent to the measurement of collaboration dynamics within and between home and school contexts. Theoretical framework and approach Recent research has shown that collaboration can be successfully measured using social network analysis, which derives social structures using networks and graph theory [21]. Much social network analysis suggests that collaboration within and between contexts[22, 23], sometimes referred to as ‘bridging and bonding social capital’[24], can impact access to resources and the diffusion and adoption of practices [25]. However, to our knowledge, no researchers (except for previous projects conducted by PI McGhee Hassrick and colleagues) have used social network analytics to study network collaboration among caregivers, community members and professionals as a potential mechanism of action for improving treatment outcomes for children on the autism spectrum. Preliminary findings from their research are promising, suggesting that collaboration among network members can impact child outcomes. For example, trust among network members for autistic children changing schools is associated with transition success, immediate post-transition [26]. Also, the size of caregiver networks is associated with greater access to services for autistic children from lower resources families [27]. Additionally, the social networks of educators was found to shape the experiences of autistic children in school settings. Teacher social network size was associated with the successful preparation for transition to a new school setting for autistic children [28]. Additionally, educators who provide professional development for teachers of autistic children are more effective when their networks are larger and have more network members with expertise in evidenced-based practices [29]. However, no studies have yet measured the impact of the networks and the adoption and diffusion of evidenced based practices. To address this gap in the field of autism, we propose an innovative approach based on traditional social network analysis [21] to evaluate whether collaboration dynamics among network members contribute to variability in implementation outcomes (our primary objective) and child treatment outcomes (our exploratory objective). In this study, we measure collaboration dynamics in two ways: (1) network density within contexts (bonding) and between contexts (bridging), defined as actual collaboration connections/total number of possible collaboration connections and (2) diffusion and adoption of EBP, defined as the proportion of network members who know about/adopt EBP with child. The project adds critical specificity by tracking the diffusion and adoption of one particular EBP strategy- the use of Alternative Augmentative Communication (AAC: see Fig. 1) strategies for MV preschoolers with autism. The rationale for focusing on this group is twofold (1) MV children are at heightened risk for poorer outcomes and (2) the most frequently implemented EBP for minimally verbal children involve the use of AAC strategies, which require teams to work collaboratively to ensure consistent use of specific devices and procedures (e.g., knowledge on how to correctly use a speech generating device or a picture communication system). Research suggests that poor treatment responses in this population might be due, in part, to poor or inconsistent implementation of such techniques among the team of professionals and family members who interact daily with the child [30]. This might be due to insufficient information sharing (diffusion) about the child’s use of an AAC intervention, as well as insufficient adoption of the child’s AAC intervention among team members. To investigate the role of collaboration dynamics on implementation and child outcomes for MV preschoolers receiving AAC intervention, we are using the EPIS implementation framework[31], as recommended by the California Evidence-based Clearinghouse, illustrated in Fig. 2. We have adapted EPIS for AAC as our targeted EBP. ACC examples EPIC Model Our overarching research questions are: To what degree is the density of home/school collaboration associated with the diffusion/adoption of AAC strategies for minimally verbal preschoolers with autism over time? We expect that high density between home/school (bridging) and within home/school (bonding) team collaboration will enable more people on the team to learn about the child’s AAC needs and intervention procedures (diffusion) and use similar AAC strategies to use with the child (adoption). To what degree is the density of home/school collaboration associated with the diffusion/adoption of AAC strategies for minimally verbal preschoolers with autism over time? We expect that high density between home/school (bridging) and within home/school (bonding) team collaboration will enable more people on the team to learn about the child’s AAC needs and intervention procedures (diffusion) and use similar AAC strategies to use with the child (adoption). What are the barriers and facilitators to collaboration among home/school team members and how does collaboration influence the diffusion and adoption of AAC strategies for preschoolers with autism? What are the barriers and facilitators to collaboration among home/school team members and how does collaboration influence the diffusion and adoption of AAC strategies for preschoolers with autism? (Exploratory): Are home/school collaboration dynamics associated with child intervention outcomes? (Exploratory): Are home/school collaboration dynamics associated with child intervention outcomes? Pilot test of social network survey with parents and school staff of autistic students To test our hypotheses, we will adapt a social network survey, developed and successfully tested for feasibility in prior research study by McGhee Hassrick and colleagues [32, 33]. The prior social network survey measured social capital, defined as access to resources, including autism information, intervention resources and autism training knowledge, embedded in parents/school networks. The pilot study successfully mapped (with 70% or above team participation) parent and school collaboration for 23 children on the autism spectrum in two low-income urban public schools (both schools were > 70% free/reduced lunch). Findings indicated that parents at one school (Aspire Academy) had access to more sources of autism information and intervention resources through their coordination, problem solving and good ideas networks, as compared with parents at the other school (Flores Elementary), suggesting that the social dynamics in school contexts potentially influence home/school collaboration and access to autism resources. Between school differences were also identified in bridging and bonding collaboration about the strategies and interventions for autistic children among school providers and between school providers and parents. Social networks in Flores Elementary were characterized as having small, well-connected networks among school staff (high bonding network density within school), with few connections to parent (low bridging network density between home and school) while Aspire Academy social networks consisted of high bridging between school staff and parents and high bonding networks among school staff [32]. While the pilot test of the social network survey provided a useful preliminary overview of the importance of school context and the differences between bridging/bonding networks among parent/school providers, a critical limitation was the lack of reliable information about the degree to which team collaboration helped or hindered the diffusion and adoption of evidence-based interventions for each child. Another key gap was lack of data about home bonding and bridging among other providers at home in addition to the parent and in the community. Finally, we did not measure child outcomes, due to diversity in outcomes across the autism spectrum. To address this, we adapted the social network survey to use egocentric network survey approach, where parents and teachers act as informants and report all known networks members at home, school and community. To address the wide range of variability among individuals on the autism spectrum for our proposed study, we narrow our investigation to a sub-group of autistic individuals (minimally verbal preschoolers). To cope with the associated empirical heterogeneity of evidence-based interventions implemented for each autistic child, we constrained our investigation to one type of evidence-based practice with high potential impact for later child outcomes (AAC device implementation). Methods Participants Ultimately, we will recruit 40 MV school-age children with autism (20 per site) between the ages of 2–10 years old and the key caregivers (N = 40) and providers (N = 40) involved in their interventions at home and school. We expect approximately 40% of our projected families will be African American, 40% will be Hispanic and 20% will be white; 60% will be from low-income families and 40% from medium to high income. We will employ financial incentives to compensate families and providers for their time. Children participating in the study will have an autism diagnosis and will be MV. For this study, MV status is defined as an expressive language level below 18 months as measured through the Vineland Scales of Adaptive Behavior-II (VABS) [34], which reflects lack of functional phrase speech. This approach to operationalize the construct of MV status is based on the work by Hus Bal[35], which has been shown to provide a stringent criterion to identify children with a severe language impairment in the autistic population. Autism diagnoses will be ascertained by district documentation in the child’s IEP. Recruited caregivers and identified providers must be active teachers or caregivers for enrolled children. In addition, to gather richer, more detailed information regarding barriers and facilitators to collaboration we will also recruit 10 additional caregivers and 10 additional providers for focus group interviews. A moderator and note taker from the research team will facilitate each focus group interview, utilizing a standard protocol that includes specific prompts about AAC diffusion and adoption dynamics (see Fig. 2 [31, 36]) as well as open-ended prompts for the inclusion of broader themes. Focus groups will be audio recorded and transcribed. Procedures/measures Child outcome measure Change in child communicative functioning is measured by re-administering the Brief Observation of Social Communication Change (BOSCC) [37]. The BOSCC demonstrates high to excellent inter-rater and test–retest reliability and shows convergent validity with measures of language and communication skills. The BOSCC is administered at baseline (T0) and end of preschool year (T2) to detect changes in the communication subscales. Outcome analysis is exploratory, and child cognition calculated with the Mullen Scales of Early Learning (MSEL) [38] is used only as a control. Team network measures At T0, 3-months post school start (T1) and T2, egocentric network surveys are administered face-to-face or virtually using zoom to the caregiver and the preschool teacher of each child. The egocentric network survey will use a three-step process to measure home, school and community connections. Step 2: In the egocentric network survey, we ask participants (caregivers and teachers) to identify each member of the child’s intervention team, called ‘alters’. The network generator prompt is “Who are the people who interact with your child at home, in the community, or in the healthcare setting right now?” Follow up prompts are provided to assist with memory recall, including “Are there any more people that interact with your child in your family? In your community? Any more preschool staff? Any more service providers?” A final network generator is asked to include key people who are not interacting directly with the child, “Who are the people who do not interact directly with your child but who provide you with help and support as you help your child communicate?” A final prompt is offered, “Are there other people that you have not yet listed?” Step 2: We then ask them to provide demographic attributes for each identified ‘alter’, including role, gender identity, age and ethnicity/race identity. We next ask them to identify AAC adoption and diffusion expertise and activities for each alter (see Team Diffusion and Adoption Rating Measure). Step 3: We then conduct a dyad census, where, for each unique pair of ‘alters’ named, they answer the following question: “Does (Person A) communicate about [CHILD]’s AAC device with (Person B)?” Frequency options are: A Lot, Some, A Little, Not at All. The question is designed to solicit report of observable behavior [32]. The egocentric network survey uses traditional egocentric network data measures adapted from standard industry approaches used routinely in publications over the past 40 years for several major social science surveys, including the NSF-funded General Social Survey, the National Longitudinal Study of Adolescent to Adult Health, and the National Social Life, Health and Aging Project. We increase accuracy by calculating the average caregiver and teacher report of density, rather than single participant report, and calculate network measures including all informant network data [33] that consider all reported ties among the identified network members. Team diffusion and adoption rating measure At Time 0 and Time 2, we ask participants (child’s caregiver and teachers) to report on AAC knowledge (diffusion) and use (adoption) for themselves and for each identified network member. We will ask participants to rate expertise in AAC device use for each alter by asking, “What is the level of expertise that (alter name) has in using [name of child’s primary AAC device] in general, not specifically with [CHILD]?” We will ask participants if each alter has AAC training by asking, “Has this person received training (e.g. instructions, demonstrations, informal coaching) in how to use [name of child’s primary AAC device] with [CHILD]?” We will ask participants if each alter knows about the child’s AAC device by asking, “Does (alter name) know that the child is using [name of child’s primary AAC device] to communicate?”. We will ask participants if each alter is using the AAC device with the child by asking, “Does the child use the primary communication device to communicate with this person? (e.g., communicating with the child using the communication tool or helping the child use the communication tool)”. We will ask participants if it is easy for alters to use the ACC device by asking them if the following statement is true or false for each alter: “It is easy for this person to use the child’s primary communication device with the child.”We will ask participants how frequently they collaborate with each alter by asking, “How often do you collaborate with this person to figure out how to improve your child’s use of the primary communication device? (e.g., provide advice, coaching, solve problems together, trouble shooting in the moment)”. We will ask participants about the belief that the child’s AAC device is better than alternative approaches by asking them if the following statement is true or false for each alter: “This person believes that [name of child’s primary AAC device] helps [CHILD] to communicate needs better than alternative approaches.” An AAC adoption indicator (belief in AAC, child’s AAC device is easy to use, expertise in AAC device use and reported use of child’s AAC with child) and an AAC diffusion indicator (AAC training, know about child’s AAC device, collaborate about child’s AAC device) will be calculated using composite scores. For the AAC adoption composite, each alter in the social network will have a score of 0–4 for belief in AAC and ease of use of child’s AAC device and expertise about AAC device use score range will be 0–3. For the AAC diffusion composite, AAC training and knowledge of the child’s AAC device will have a range of 0–1 while collaboration about child’s AC device will have a range of 0–5. Composite scores for each indicator will be summed up for each alter in the network. Mean composite diffusion and adoption scores will be calculated based on the average score of the people in the network. Team bridging/bonding density measure At T0,T1 and T2, we will use the social network survey to measure the (1) overall density of team ties, calculated as the number of actual ties of all providers/possible ties; (2) bonding domain degree proportion, calculated as the number of ties each team member has with other team members from the same context/total # of team members identified from that context; and (3) bridging domain degree proportion, calculated as the number of ties each team member has with other team members from each different context/total number of team members identified from each different context. Bonding ties include interactions within home, school and community team members identified by informants and bridging ties include interactions between home, school and community team members identified by informants. Each team member will have a bridging/bonding value provided by each informant. We calculate mean values across all informants for each network member. Focus group qualitative measures for implementation outcomes To determine barriers and facilitators to collaboration, we are conducting focus groups and individual interviews [39, 40]. Following general principles, we recruited 6–10 ‘homogenous strangers’[41] to participate in small focus groups or individual interviews, including caregivers and providers at each study site. A moderator and note taker from the research team facilitated each focus group or interview, utilizing a standard protocol that includes specific prompts about AAC diffusion and adoption dynamics (see Fig. 2 [31, 36]) as well as open-ended prompts for the inclusion of broader themes. Focus groups were recorded and transcribed. Data analysis Our research questions will be tested using a mixed-methods approach. An overview of the approach for each aim is illustrated in Table 1. All instruments will be pilot tested (n = 5 caregiver/teacher informant pairs, recruited at research sites). Table 1ProceduresRationaleAim 1• Recruitment and measurement of baseline characteristics (SES, severity, IQ, demographics) of 40 autistic preschoolers starting an early intervention program• Testing the hypothesis that teams with high bridging/bonding density across home/school will diffuse more information about AAC techniques and, as a result, more successful treatment techniques will be adopted by within/between home/school providers, thus resulting in more consistent and frequent use of successful treatment techniques over time.• Measurement of bridging/bonding density of the intervention team at baseline (T0) 3-months post school start (T1) and end-of-school year data (T2) using the social network survey• Rate the diffusion and adoption of AAC strategies at T0,T2.Aim 2• Focus group interviews with the parents and teachers who provide AAC strategies for minimally verbal autistic preschool children about factors listed in Aim 2. Transcribed data will be qualitatively coded using NVIVO and analyzed using grounded theory[42].Determining barriers and facilitators to the diffusion of information about what works or does not work for the child, and the resulting adoption of successful treatment techniques and discontinuation of unsuccessful treatment techniques, over time.Aim 3• Measurement of intervention outcomes, i.e., changes in child communication functioning as indexed by changes occurring from T0 to T2 in BOSCC, VABS 2• Measurement of bridging/bonding density of the intervention team (T0,T1,T2) using the social network survey• Exploratory analysis on the association between bridging/bonding density and variability in child intervention outcomes• Preliminary examination of how team density is associated with variability in child intervention outcomes• To determine key parameters for multiple study components (e.g. innovative ecosystem social network data collection, MV recruitment dynamics, gender disparities, effect size parameters) Table 1 • Measurement of intervention outcomes, i.e., changes in child communication functioning as indexed by changes occurring from T0 to T2 in BOSCC, VABS 2 • Measurement of bridging/bonding density of the intervention team (T0,T1,T2) using the social network survey • Exploratory analysis on the association between bridging/bonding density and variability in child intervention outcomes • Preliminary examination of how team density is associated with variability in child intervention outcomes • To determine key parameters for multiple study components (e.g. innovative ecosystem social network data collection, MV recruitment dynamics, gender disparities, effect size parameters) Aim 1: to determine associations between bridging/bonding density and the diffusion/adoption AAC strategies for preschoolers on the autism spectrum For Aim 1, outcome measures, social network measures, and demographic and covariate data will be collected for each team via Qualtrics surveys from youth, caregiver and key preschool providers. Linear Network Autocorrelation Models (LNAM) [43, 44] will be deployed to test the degree to which bridging/bonding density contributes to diffusion and adoption of AAC within each team. Aim 1 will characterize the implementation network structure of collaboration for MV autistic children in preschool settings and examine whether variation in network structure is associated with the diffusion and adoption of AAC strategies. Aim 2: to determine barriers and facilitators to collaboration among home/school network members and investigate how collaboration influences the diffusion and adoption of AAC strategies for preschoolers on the autism spectrum Aim 2 addresses “why” diffusion and adoption might vary across network members, allowing us to investigate what hinders and facilitates implementation of AAC. Based on the EPIS implementation model [31] (Fig. 3) and other autism implementation research[45], we will focus on the inner context of the child’s team, including school and family collaboration/skill/culture, and their use of AAC. Focus group interviews with team members and school leaders will investigate fit and complexity as defined by (1) the perceived relative advantage of using AAC strategies over other treatment techniques, (2) the compatibility of AAC strategies with existing values, needs and experiences and (3) the perceived complexity of AAC strategies. See Fig. 3. Additional barriers/facilitators related to implementation outcomes that will be examined include acceptability, feasibility, penetration and sustainability, and barriers related to (a) school norms regarding AAC, (b) organizational routines that shape teacher practices, (c) the number of different providers and agencies involved in the child’s intervention team, and (d) differences in team members’ theoretical orientations and training. We will audiotape and transcribe focus groups and we will code transcripts using deductive (theory-based codes) and inductive codes, using grounded theory[42], an inductive qualitative method for uncovering dynamics that shape implementation processes. The qualitative data will provide detailed information about how family and school barriers and facilitators shape acceptability, feasibility, penetration and sustainability of AAC strategies. The qualitative findings will inform instrument revisions for Aim 1&3 data collection. Finally, we will triangulate Aim 2 findings with Aim 1 network analyses by investigating the alignment between qualitative and quantitative data sources. Example of implementation factors Aim 3: to determine whether home/school collaboration dynamics is associated with child intervention outcomes Aim 3 involves an exploratory analysis examining the link between the diffusion and adoption of AAC strategies for a particular child and child intervention outcomes. Aim 3 will provide pilot data and a feasibility test for a later appropriately powered study involving a larger sample that controls for nested effects, within level network effects, and other key controls, such as SES, IQ, and severity of autism symptoms. For our analysis, we will use Linear Network Autocorrelation Models (LNAM) [38, 39] to test whether changes in child communicative functioning are predicted by bridging/bonding density measures above and beyond other relevant factors (including language, IQ, symptom severity, family SES, which will be entered as covariate terms in analyses). Given the small sample employed in this pilot study, effect sizes will be used to evaluate the contribution of putative predictors to magnitude of treatment-related changes and interactions with relevant variables. This pilot study will provide the groundwork needed to pursue optimally powered research. Study status The Institutional Review Boards of each participating university have approved all study procedures. At the time of submission of this manuscript, we have already enrolled participants for data collection. Clinical trial number: not applicable. Discussion This study protocol looks at a crucial gap in the literature for children with autism spectrum disorder and how social network research can elucidate ways to improve a child’s outcomes. By implementing a specific evidence-based practice, Alternative and Augmentative Communication (AAC) strategies, we believe this study can show how collaboration among caregivers and providers impacts the effectiveness of AAC interventions. For this study we address this gap in the literature through an adaptation of the social network survey aimed at analyzing social networks for minimally verbal (MV) autistic children who would benefit from the use of a particular EBP- an Augmentative and Alternative Communication system (AAC). This will allow us to obtain parent and teacher ratings for AAC knowledge and AAC use, measured by asking parents and teachers to identify (a) if team members knew about AAC techniques for the child, (b) if they had implemented AAC strategies for the child, and (c) if the intervention team adopts AAC techniques over time. We will combine these data with the egocentric network data to identify whether AAC treatment techniques tend to be diffused and adopted by intervention teams with higher bonding/bridging network density at home/school (Aim1) and if high team collaboration predicts improved child outcomes (Aim 3). For Aim 1&3, we will collect social network data from each child’s parents and key preschool providers to identify all providers at home and school and measure about how much they (1) coordinate (feedback system); (2) use and promote successful treatment techniques (champions); and (3) bridging/bonding network density over time (team restructuring). We anticipate that our findings will highlight the need for stronger and more synergistic collaboration between a MV autistic child’s care groups to facilitate the adoption of AAC strategies. As a result, this may lead to improved communication outcomes for these children, increasing the likelihood of contributing to what Prasad [46] (2021:3–4) identified as problem solving sociology, defined as “sociology that attempts to use traditions of sociology research to solve real world problems, and uses the attempt to grapple with real-world problems as a way of reformulating understandings of society and renew or reinvent those traditions.” Abbreviations Alternative and Augmentative Communication Minimally Verbal Exploration, Preparation, Implementation, Sustainment Brief Observation of Social Communication Change Vineland Adaptive Behavior Scales Child Development Inventory Mullen Scales of Early Learning Socioeconomic Status Individualized Education Program Evidence-Based Practice Authors’ contributions Sarah Dufek: Conceptualization, Methodology, Funding acquisition, Writing - Original Draft, Writing - Review & Editing, Visualization. Aubyn Stahmer: Conceptualization, Methodology, Funding acquisition. Writing - Review & Editing. Giacomo Vivanti: Conceptualization, Methodology, Funding acquisition, Writing - Original Draft, Writing - Review & Editing, Visualization. Elizabeth McGhee Hassrick: Conceptualization, Methodology, Funding acquisition, Writing - Original Draft, Writing - Review & Editing, Visualization. Funding This research was supported by the National Institute of Mental Health under Award Number 1R21MH121731-01 A. Data availability No datasets were generated or analysed during the current study. Declarations Ethics approval and consent to participate The study was approved by the Drexel IRB (Protocol # 1907007312) and University of California, Davis has an approved IRB (Protocol # 1907007312) as well as Simons Foundation Autism Research Initiative (SFARI) Recruitment Project Approval to recruit participants from Simons Powering Autism Research (SPARK) participants, in accordance with the ethical standards as laid down in the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards. All participants gave written informed consent before participating in the study. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement No datasets were generated or analysed during the current study.