Parent-Implemented Positive Behavior Support Strategies for Young Children on the Autism Spectrum: A Pilot Investigation

Abstract

Purpose:

Parents of children on the autism spectrum enrolled in early intervention often receive coaching to address both social communication and disruptive behavior, which are the two most frequently reported concerns by parents. Intervention techniques for both are often recommended to be implemented across daily routines and require the parents to learn new ways of interacting with their child. A sequential approach to instructing parents in these key intervention targets may reduce burden and increase adherence.

Method:

This multiple-baseline design pilot study included three mother–child dyads who received instruction in a disruptive behavior intervention immediately following a social communication intervention. Maternal maintenance of social communication strategies and child disruptive behaviors were measured during probes throughout the study.

Results:

Results indicate that although mothers readily learned to implement the techniques, fidelity of implementing social communication strategies declined after introduction of the positive behavior support strategies.

Conclusions:

A sequenced approach to parent-mediated intervention is feasible and acceptable. Clinical implications and future directions are discussed.

Supplemental Material:

https://doi.org/10.23641/asha.19528978

The core diagnostic features of autism include specific impairments in social communication and the presence of repetitive interests or behaviors (American Psychiatric Association, 2013), which are detectable by 24 months of age (Charman & Gotham, 2013). Although disruptive behaviors are not considered a core feature of autism, disruptive behaviors are common in this population (Gadow et al., 2004; Lecavalier, 2006). Furthermore, social communication and disruptive behavior are parents' two greatest concerns prior to receiving an autism diagnosis (De Giacomo & Fombonne, 1998; Kozlowski et al., 2011), with the majority of parents identifying their first concerns as disruptive behaviors (Guinchat et al., 2012). Taken together, speech-language pathologists (SLPs) are often challenged with addressing disruptive behaviors to optimize social communication outcomes for young children on the autism spectrum.

Both social communication and disruptive behavior are critical targets to address early in life due to their strong association with positive long-term outcomes and influence on family's participation in typical activities (Billstedt et al., 2007; Gillespie-Lynch et al., 2012). Both targets are frequently addressed using a parent-mediated approach. As such, the sequence of delivering intervention approaches must be simplified for individual families to optimize engagement, reduce burden, and individualize based on child and parent needs. However, little is known about how to sequence and integrate multiple parent-mediated interventions to meet families' specific needs.

Understanding how social communication and disruptive behavior influence one another is critical to understanding how best to sequence interventions that often address each intervention target separately. A dual deficit in social communication and disruptive behavior may occur via their reciprocal influence on each other. For example, when a child has difficulty communicating, he may become frustrated and exhibit disruptive behaviors, such as screaming. Likewise, a child with significant disruptive behaviors may not be able to engage easily with social communication partners, thereby impeding social communication development. Disruptive behavior and social communication skills may be linked in children with autism (Park et al., 2012). However, this association may weaken in older children as verbal ability becomes more associated with adaptive behaviors and coping skills (Williams et al., 2018). Therefore, addressing social communication alone may not be sufficient to prevent long-term disruptive behavior, and targeted skill building intervention strategies for disruptive behavior may be necessary for some children.

Early intervention is a critical period for teaching social communication skills to maximize gains in outcomes in continued language development (Thurm et al., 2007), and parent-mediated interventions are effective at improving social communication skills in young children with and at increased likelihood for autism (Roberts et al., 2019). Naturalistic Developmental Behavioral Interventions (NDBIs) include teaching parents communication support strategies to build a foundation of joint engagement for language teaching in naturally occurring activities (Schreibman et al., 2015). Although NDBIs have been identified as being the most effective approach for improving many outcomes in young children with autism (Tiede & Walton, 2019), the intervention packages focus strategies toward building joint engagement and social communication and notably few include manualized strategies toward addressing disruptive behaviors. While NDBIs are not intended to address disruptive behaviors specifically, a few manualized programs include strategies that target disruptive behaviors. For example, one NDBI, Project ImPACT, provides a brief and optional single-session module on managing disruptive behaviors (Ingersoll & Dvortcsak, 2019). NDBIs may address disruptive behavior through different pathways including this type of optional direct instruction or through improving alternative behaviors such as play and positive social communication. However, a targeted set of tools designed to specifically address disruptive behavior may be necessary. Disruptive behavior and social communication have an influence on one another through intervention; children with significant language delays, including autism, have demonstrated significant improvements in disruptive behavior when gains in social communication are achieved (Curtis et al., 2018, 2019). Additionally, children who received parent-mediated interventions targeting disruptive behaviors also demonstrated improvements in social communication (Bearss et al., 2013). These results indicate that the link between disruptive behaviors and communication abilities may be used to maximize gains in outcomes across intervention targets. While the reciprocal influence of disruptive behaviors on social communication may mitigate the need for explicit disruptive behavior strategies for some parents, the added benefit of integrating positive behavior support strategies and sequencing these strategies with social communication strategies for parents during NDBIs has yet to be evaluated.

Early intervention is also a critical period for addressing disruptive behaviors, particularly because these behaviors interfere with learning across intervention targets, increase parent stress, and impact overall family quality of life (Bohadana et al., 2019; Costa et al., 2017; Lecavalier, 2006). This is especially important in early childhood where caregiver strain has been found to be particularly driven by disruptive behaviors (Bradshaw et al., 2020). Additionally, when parents learn behavior management strategies, their children have fewer disruptive behaviors (Bearss et al., 2013; Brookman-Frazee et al., 2020; Ginn et al., 2017; Postorino et al., 2017). Parent instruction programs targeting disruptive behavior specifically focus on providing parents with strategies for improving the family's quality of life through prevention, response, and teaching strategies. One such intervention, developed through the Research Units in Behavioral Intervention (RUBI) Autism Network, provides a manualized weekly curriculum including 11-core skill building sessions to teach parents to understand, prevent, and respond to disruptive behaviors as well as strategies for teaching replacement behaviors (Bearss et al., 2018). These strategies are designed to improve joint engagement between the parent and child by teaching parents to observe disruptive behaviors, determine the function of the behavior, select appropriate prevention and response strategies, and teach skills and verbal requests to replace and prevent disruptive behaviors. The skill building approach to RUBI is designed to address ongoing disruptive behaviors as well as build skills in the parent and child to improve parent–child joint engagement through preventing and addressing future disruptive behaviors. This approach includes components of evidence-based behavior practices that are recommended by the American Speech-Language-Hearing Association (ASHA, 2018) such as Functional Behavior Assessment (Neidert et al., 2013) and Functional Communication Training (Reichle & Wacker, 2017), which are taught as feasible strategies for parents to implement in everyday settings. Although this intervention approach does teach communication skills as a replacement behavior, these communication opportunities occur in a narrow context and are limited in form. The RUBI intervention has been found to be effective for not only improving parent use of strategies but also for improving child disruptive behavior outcomes (Bearss et al., 2015). However, the RUBI program exclusively focuses on reducing disruptive behavior and improving adaptive skills, and although some children additionally demonstrated gains in social communication, there are no existing recommendations for how to sequence the RUBI intervention with other critical parent-mediated interventions to optimize social communication for children on the autism spectrum.

An adaptive intervention is a model that uses treatment options to systematically individualize and accommodate the specific and changing needs of individuals (Collins et al., 2004). An adaptive intervention is needed when parents are implementors of intervention strategies, given its potential to reduce costs, minimize burden placed on parents, and effectively focus intervention content around the needs of each parent's child. Additionally, children on the autism spectrum often require support and intervention throughout early childhood, such that intervention approaches are often sequenced throughout the prolonged diagnosis and initial intervention phase (Green, 2019). Simultaneous delivery of multiple intervention approaches may introduce unnecessary complexity and may be confusing for the parent. Because of the reciprocal nature of these two important targets, SLPs are challenged with addressing both in a feasible and reasonable manner for the family.

Study Purpose

The purpose of this study is to provide a pilot investigation of the feasibility and initial effects from one possible sequence of interventions to address parents' primary concerns (i.e., social communication and disruptive behavior) in early intervention: implementing a parent-mediated positive behavior support intervention (RUBI) immediately following an 8-week parent-mediated responsiveness intervention, targeting social communication. Although the RUBI intervention has a growing evidence base (Bearss et al., 2013, 2015), this article serves as an initial step toward refining an adaptive intervention through more carefully considering the individual components of the RUBI intervention, the variability in implementation across strategies and across parent/child profiles, the acceptability and feasibility of sequencing interventions, and the impact of sequential training on parent implementation over time. Sequential implementation of NDBI and RUBI interventions is one possible approach in working toward an adaptive intervention model. Our goal is to design an adaptive approach that is responsive to child and parent characteristics and changing needs throughout intervention, and this sequential approach was therefore selected as a first step to understanding the components and decisions needed to build a truly adaptive intervention model. A single-subject design, multiple-baseline across behaviors, was selected to answer the following research questions: Do parents of young children with autism (a) implement positive behavior support strategies with high levels of implementation (≥ 80% fidelity) immediately following parent instruction, (b) maintain use of social communication strategies after the introduction of positive behavior support strategies, (c) rate the sequence of interventions as favorable and acceptable, and (d) maintain and generalize newly learned skills over time?

Method

Participants

This study included three preschool-aged (33–40 months) children diagnosed with autism and their mothers. All mothers had received professional training, graduate education or higher; one mother was a special education teacher. The children included one female and two males (see Table 1). The mother–child dyads are referred to by the following pseudonyms: Adele and Addie, Brianna and Brian, and Carla and Charlie. Children had not received prior applied behavior analysis intervention. All three children had participated in state-funded early intervention programming, and two parents reported receiving tips or suggestions about interacting with their child from this early intervention (Adele and Carla).

Table 1.

Pilot study participant characteristics.

Variable	Adele and Addie		Briana and Brian		Carla and Charlie
Mother's age (years)	40		32		34
Family size	5		4		3
Family income	$70,000–$100,000		> $100,000		> $100,000
Mother's race	White		White		White
Mother's ethnicity	Hispanic or Latino		Not Hispanic or Latino		Not Hispanic or Latino
Child's age in months
RUBI intervention baseline	33.87		39.83		34.61
Child's gender	Female		Male		Male
Child's race	More than one race		White		White
Child's ethnicity	Hispanic or Latino		Not Hispanic or Latino		Not Hispanic or Latino
Child's target behaviors	Elopement, aggression (hitting, biting, pulling hair, scratching)		Elopement, tantrums (throwing body on ground), yelling, throwing objects		Elopement, throwing objects, whining, standing on chairs
Cognitive skills (MSEL Visual Reception NVIQ/AE)	85.71/24		55/18		83.33/25
Autism calibrated severity score (ADOS-2)	7		10		7
Child language age equivalence (PLS-5)	RUBI baseline	RUBI posttest	RUBI baseline	RUBI posttest	RUBI baseline	RUBI posttest
Auditory comprehension (months)	23	30	18	17	22	24
Expressive communication (months)	20	30	20	24	21	21
Child disruptive behavior profile
ABC Irritability	17	6	8	11	5	14
ABC Hyperactivity	19	6	28	28	13	27
HSQ Total Number of Problem Settings	23	14	21	21	13	13
HSQ Mean Severity Score	7.38	0.92	3.46	4.13	2.71	1.50
DB-DOS Behavior Regulation (raw score) a	16	2	15	20	14	6
DB-DOS Competence (raw score)	11	16	1	14	13	23
DB-DOS Problems in Anger (raw score) a	23	10	0	16	16	9

Note. RUBI = Research Units in Behavioral Intervention; MSEL = Mullen Scales of Early Learning (Mullen, 1995); NVIQ = Nonverbal Intellectual Quotient; AE = Age Equivalent; ADOS-2 = Autism Diagnostic Observation Schedule–Second Edition (Lord et al., 2012); PLS-5 = Preschool Language Scales–Fifth Edition (Zimmerman et al., 2011). ABC = Aberrant Behavior Checklist–Community (Aman et al., 1985). HSQ = Home Situations Questionnaire (Chowdhury et al., 2016). DB-DOS = Disruptive Behavior Diagnostic Observation Schedule, Examiner context.

^aScores are reverse coded such that higher scores indicate greater impairment. RUBI baseline and posttest were 12 weeks apart.

Children who had recently completed an 8-week social communication intervention with their caregiver as part of an ongoing randomized controlled trial (R01DC014709) and who had a confirmed autism diagnosis were recruited for this institutional review board–approved study. Parent–child dyads were included if the parent indicated any concern about their child's disruptive behavior and if they consented to the additional training in strategies to address disruptive behavior. Autism diagnosis was confirmed by a research reliable assessor using the toddler module (Gotham et al., 2007) of the Autism Diagnostic Observation Schedule (ADOS-2; Lord et al., 2012) and clinical judgment. Children presented with a range of developmental levels that fell within 1 SD below the mean (Nonverbal Intellectual Quotient; NVIQ range: 55–87), as measured by the Visual Reception subscale of the Mullen Scales of Early Learning (VR-MSEL; see Table 1; Mullen, 1995). NVIQ was calculated using the Visual Reception subscale age equivalent scores divided by the participant's chronological age and multiplied by 100 (Bishop et al., 2011). Clinicians included research personnel who had reached high levels of intervention fidelity (> 90%) prior to the beginning of the study and were two licensed SLPs and one speech-language pathology student in training.

Setting and Materials

All intervention sessions and outcome measures were conducted in the family's home, with the exception of the generalization measure, which was administered in a clinic setting. The family selected a location in their home where they typically spent play time with their child, and all known distractions were minimized. Families selected open living rooms, play rooms, and bedrooms. Televisions were turned off, additional family members left the room, and distractions such as tablets or preferred toys/items that are difficult to share were placed out of sight. Weekly intervention sessions occurred in a comfortable space in the home with familiar toys available where the clinician and mother could have a conversation with minimal distractions.

Experimental Design and Procedures

This study employed a multiple-baseline design across parent behaviors with an embedded multiple probe design across parent behaviors for child disruptive behavior, and this was replicated across the three mother–child dyads (Gast & Ledford, 2014). Mother–child dyads participated in each phase until all relevant lessons had been delivered and a stable performance level was observed across three consecutive sessions. If the mother had completed all the lessons for a particular phase but had not yet demonstrated meaningful improvement, the clinician delivered an additional booster session. Booster sessions included previously taught content that the mother had yet to master.

Social Communication Intervention

Prior to this study, mother–child dyads participated in an 8-week NDBI intervention that focused on coaching the mothers to respond to all of their child's social communication attempts. This intervention included all of the responsive interaction components of Enhanced Milieu Teaching and did not include communication temptations or prompting procedures (Kaiser & Hampton, 2016). Mothers were taught to build joint engagement through turn taking, responding to communicative attempts, pausing between comments, and reducing questions/demands (Roberts & Kaiser, 2015). Social communication intervention sessions occurred with the clinician, mother, and child. The clinician used a teach–model–coach–review format to instruct the mother on each of the target strategies (Roberts et al., 2014). Intervention sessions were each 1-hr long, resulting in a total dosage of 8 hr of parent training in social communication strategies. Mother–child dyads enrolled in the second stage disruptive behavior intervention within 4 weeks of completing the social communication intervention.

Baseline

Dyads participated in 20-min baseline observations until a stable performance level was observed over three consecutive sessions. Following this initial stability, mothers engaged in the first session with the clinician to learn about defining disruptive behaviors. During this first session, parents received instruction on defining and observing disruptive behaviors using an antecedent–behavior–consequence format. During this functional behavior assessment activity during the initial session, the clinician and parent identified the function of the target behavior and all future sessions incorporated this perspective into how strategies were implemented. Because this session did not include specific intervention strategies, this initial session conceptually continued baseline. Mothers continued data collection until a stable level and trend was observed, and then the second session was presented, initiating the first phase of the disruptive behavior intervention. For each participant during the first phase, baseline lasted three sessions before the introduction and three sessions following the introduction.

Disruptive Behavior Intervention

The RUBI disruptive behavior intervention (Bearss et al., 2018) was divided into four conceptual phases by the research team, which included one of the developers of the disruptive behavior intervention. The four phases, Prevention, Reinforcement, Responding, and Teaching (see Table 2), were considered to be similar in amount of content and difficulty for implementation. Each session was approximately 1-hr long, and with the addition of booster sessions, this resulted in a total dosage of 13–14 hr of disruptive behavior instruction. Each phase consisted of two to three sessions, which occurred between the mother and clinician alone. During these dyadic sessions, clinicians followed the scripted RUBI protocol that includes discussion of the content, video examples, role play of the targeted intervention procedures, and review of homework assignments. Clinicians assigned homework assignments each week, which included targeted strategy practice in daily routines and data collection on strategy use. Following each session, clinicians rated parent completion of the previous week's homework on a 12-point scale with regard to parent's collection of data, accuracy of practice strategy implementation, and frequency of strategy practice. Across all sessions, mean clinician rating for parent homework completion was 10.21 (SD = 1.56) across participants with some variability across participants (Adele: M = 8.67, SD = 2.06; Brianna: M = 11, SD = 1.49; Carla = 10.8, SD = 1.14).

Table 2.

Intervention sequence.

Phase	Length	Content
Social communication intervention	8 weeks	Responsiveness, joint engagement, turn taking
Baseline: Principles	2 weeks	Defining disruptive behaviors and overview of procedures
Disruptive behavior intervention:
Phase 1: Prevention	2 weeks	Prevention strategies and daily schedules
Phase 2: Reinforcement	2 weeks	Reinforcement through praise and engagement
Phase 3: Responding	2–3 weeks	Planned ignoring and compliance training
Phase 4: Teaching	3 weeks	Functional communication training and teaching skills
Maintenance	2 weeks	Telephone booster

Additionally, during each weekly dyadic session, the mother and clinician worked together to build a Behavior Support Plan (BSP) by adding individualized strategies related to that week's session and the identified target behaviors (see Table 1). By the end of the 12-week intervention, the mother and clinician had co-created a complete BSP for the child's target disruptive behaviors and the mother had the tools to continue to adapt the plan as necessary. An overview of the phases is outlined below, and specific details necessary for replication are available in the published manual.

Phase 1. Prevention strategies. Phase 1 consisted of teaching the caregiver to implement RUBI Prevention Strategies. These included avoiding difficult situations, controlling the environment, changing the order of events, addressing setting events, doing things in small doses, responding to early signs, increasing choices, and using auditory cues, timers, first–then statements, and visual schedules. Across the two Phase 1 sessions, mothers were given materials for a visual schedule and practiced the skills for at least two additional prevention strategies that were selected by the clinician and mother together. During this initial phase, mothers also reviewed data collection from the initial session of the antecedent–behavior–consequence sequences that they had observed. The clinician and parent worked together to refine the target disruptive behavior definitions and hypothesized about the function of the behavior(s) based on these data. This conversation was guided by the trained clinician who had also reviewed baseline assessments to aide in forming the hypothesis.

Phase 2. Reinforcement and engagement. Mothers learned how to identify, select, and deliver specific praise and tangible reinforcement for both appropriate behavior and engagement in play. Mothers were also coached to follow their child's lead, reduce demands, and describe play actions as natural reinforcement for engagement in appropriate play. These additional strategies shared some features with the social communication intervention; however, the purpose of implementation and focus of application differed (i.e., supporting engagement in play to find opportunities for reinforcement rather than to build a foundation for turn taking or communication).

Phase 3: Responding. Instruction focused on how mothers should respond when disruptive behaviors occur and included three primary strategies: compliance training, redirection, and planned ignoring. Compliance training was defined as following through on task directions (using hand-over-hand prompting) given to the child that the parent reasonably expects the child to carry out. For example, if the child had mastered the skill of sitting when directed, the mother was instructed to follow through when giving the task direction “sit down,” and immediately provide reinforcement. Mothers were taught to implement planned ignoring of disruptive behaviors by avoiding eye contact and touch, walking away, maintaining a neutral facial expression, avoiding talking to the child or responding to requests in any way, and by ensuring the attempt to ignore is obvious, abrupt, and exaggerated. Mothers were also instructed to implement planned ignoring in two additional formats, as necessary: ignoring the child but not the behavior when the behavior is unsafe and ignoring the behavior but not the child by redirecting the child when the behavior is particularly persistent but not unsafe. Redirecting was defined as guiding the child to re-engage with the current activity or routine using neutral statements and physical guidance as needed.

Phase 4: Teaching. The final phase focused on instructing caregivers on how to teach their child new skills. Content included functional communication training, teaching a chained behavior, and teaching discrete skills. Functional communication training involved teaching the parent to prompt for a replacement communicative behavior and praise use or attempts at use of a replacement communication behavior. The mother and clinician worked together to select a chained behavior and a method (forward, backward). Chained behaviors, those with multiple steps, included handwashing and dressing and were identified as target behaviors at the start of the RUBI intervention. Mothers were also taught to teach their child discrete single-step skills, such as cleaning up a toy, using a similar procedure.

Maintenance

Data collection continued for 2 weeks following the completion of Phase 4. During this time, mothers received a scheduled phone call to discuss any challenges or questions and to confirm final assessment logistics. Although some minimal content was discussed, these calls primarily served as a check-in to ensure continued participation and data collection.

Fidelity of Procedures

Fidelity of intervention implementation was rated for 40% of total sessions, randomized across clinician–mother dyads and phases of intervention. Clinicians rated one another's sessions using the RUBI fidelity checklists. Reliability was completed on 25% of these selected sessions by a second clinical rater; reliability with second rater coding was 100% across the double-coded sessions. Additionally, an expert RUBI clinician (Bearss) viewed fidelity sessions and consulted on implementation to ensure high quality of intervention implementation.

Forty percent of all intervention sessions were scored for fidelity of implementation, randomized within (a) interventionist/dyad and (b) phase of intervention. Fidelity was measured using manualized RUBI fidelity checklists (Bearss et al., 2018), individually tailored to each session's content. Intervention fidelity for Dyad 1 was 100% across all sessions observed in each phase. Intervention fidelity for Dyad 2 was greater than 94% across all sessions observed in each phase (range: 94.44%–100%). Intervention for Dyad 3 was greater than 91% across all observed sessions in each phase (range: 91.67%–100%).

Dependent Measures

Assessments were administered prior to participants beginning the social communication intervention (T0), between the end of the social communication intervention and the beginning of the disruptive behavior intervention (T1), during the course of intervention, and immediately following the disruptive behavior intervention (T2).

Pre–Post Measures

Immediately before and after the disruptive behavior intervention, generalized child disruptive behavior was characterized using three measures: (a) The Aberrant Behavior Checklist (ABC; Aman et al., 1985), (b) the Home Situations Questionnaire for Autism (HSQ-ASD; Chowdhury et al., 2016), and (c) The Disruptive Behavior Diagnostic Observation Schedule (DB-DOS; Wakschlag et al., 2008), with adaptations for young children with autism (Hampton et al., 2021). The DB-DOS assesses disruptive behavior across parent and clinician contexts during which the child is exposed to standardized compliance tasks, each designed to elicit mild frustration or disappointment. The examiner responds to disruptive behavior according to a hierarchy of responses providing increasing support as needed. The DB-DOS scores for child disruptive behavior include Anger Modulation, Behavior Regulation, and Competence. The ABC and HSQ-ASD are parent survey measures designed to capture the child's overall disruptive behavior and the impact of these behaviors on everyday living. The ABC includes five subscales: Irritability, Social Withdrawal, Stereotypy, Hyperactivity, and Inappropriate Speech, rated on a 0–3 scale of severity (Aman et al., 2010). Total subscale scores for Irritability and Hyperactivity are presented to demonstrate the variability of disruptive behaviors across the subtypes most expected to change in this study. The HSQ-ASD is a measure of noncompliance across daily routines and situations presented as a mean severity score with higher scores indicating greater severity (Chowdhury et al., 2016). We also report the number of problem areas (out of 24) to indicate the pervasiveness of the noncompliance that each family reports. When interpreting the results of the ABC, a 25% decrease in score on the ABC or the HSQ-ASD is considered a meaningful clinical improvement (Aman et al., 2009; Bearss et al., 2015).

Parent Implementation of Intervention Procedures

During the course of the 12-week intervention, mothers recorded home videos of a disruptive behavior parent–child interaction (DB-PCX) 3 times per week (i.e., were instructed to complete three recordings between each intervention session). The 20-min semistructured interaction included four specific contexts designed to elicit opportunities for the mothers to respond to and prevent disruptive behaviors: snack, play, a self-care skill, and clean up. Mothers and clinicians selected a standard set of materials and activities at the beginning of the study. Target self-care activities were selected with support from the clinician to ensure that each activity was appropriately challenging. Everyday materials, already available in the family's home, were used for all activities. Mothers were instructed to complete the four contexts in any order they preferred, such that the order of activities varied across recordings, and were coached to conduct each context for about 5 min. Recordings were collected using secure video conferencing software, and mothers were taught how to record these videos prior to the first data collection session. The video conferencing software retained all recordings, such that mothers could not select an optimal video to submit. Recordings that were longer than the allotted time were shortened to the middle 20-min segments where the parent and child were interacting, removing any long periods (> 1 min) of off-camera time. Videos that were shorter than 20 min were rated in full, and the parent was given feedback to extend the future recordings. Short videos occurred for 11/114 of the home-recorded sessions, and the length of shortened recordings ranged from 17:45 to 19:55.

Parent–child interactions were coded by research assistants for the parent's quality and frequency of implementation of the positive behavior support strategies using a 20-item checklist with a 0–2 rating scale for each item across the four intervention phases (Supplemental Material S1). The primary coder of the DB-PCX was aware of the study purpose and phase of the study; however, all secondary coders were masked to study phase. DB-PCX sessions were randomly assigned such that the secondary coders did not rate them in chronological order. The checklist was developed by the study team including one of the RUBI intervention developers (Bearss). Scores were summarized as percent correct strategy use ([total raw score]/[total possible score] × 100) with 80% considered mastery of the intervention strategies.

Parent Generalization of Behavior Support Strategies

Generalization was measured using the DB-DOS. The parent context of the DB-DOS is unscripted to provide an opportunity to observe child self-regulation and mothers' natural responses. The interactions are coded using quality ratings for three subscales: Quality of irritability, Noncompliance, and Aggression. For this study, we also applied the DB-PCX coding scheme to the parent context of the DB-DOS as a generalization probe of the mothers' ability to implement the disruptive behavior intervention strategies in a novel context. The DB-DOS was conducted prior to baseline and following the completion of the maintenance phase of intervention in a clinic setting.

Parent Implementation of Social Communication Strategies

Once, during each phase of intervention, parent–child dyads participated in an additional 10-min parent–child interaction that was specifically designed to elicit the parent's use of the social communication intervention strategies. These interactions were filmed by a research assistant in the family's homes using a researcher-provided standard set of toys, and each session was transcribed by a coder blind to time point. Parent implementation of social communication strategies during social communication parent–child interactions (SC-PCX) is presented as a composite score from coded transcripts of the interaction, rated by coders trained to greater than 80% point-by-point agreement. The composite responsiveness score from the SC-PCX context reflects use of four-core responsive strategies, adjusted for child language level. These four strategies, weighted equally, are temporal contingency (i.e., responding to child communication within 3 s and leaving time for balanced turns), topic contingency (i.e., modeling language based on the child's attentional focus), mirroring and mapping (i.e., imitation and labeling of child actions), and target level speech (i.e., modeling utterance length that matches child language level (Sone et al., 2021).

Child Rate of Disruptive Behaviors

Approximately 25% of the DB-PCX recordings in each phase and for each participant were also coded for the presence of disruptive behaviors using a 15-s partial interval code. Disruptive behaviors included tantrum behaviors, aggression, and noncompliance. These behaviors were specifically defined for each participant based on the baseline assessments (see Table 1). Target disruptive behaviors (as outlined in Table 1) remained the focus of the intervention for each child throughout the study.

Social Validity

Mothers rated postintervention satisfaction (Supplemental Material S2) with the intervention procedures using a 22-item questionnaire. Questions were rated on a 3-point scale of agreement, each probing about the perceived helpfulness and value of the disruptive behavior intervention structure, content, and materials. Mothers were also asked to provide open-ended responses about specific likes and dislikes related to the intervention. Additionally, we asked the mothers to both rate and provide an open-ended description of their relative satisfaction with and perceived efficacy of the social communication and disruptive behavior interventions and the sequence.

Interobserver Agreement and Reliability

Interobserver agreement (IOA) for parent use of intervention strategies was completed on a randomized 20% of all DB-PCX sessions. For each dyad, sessions selected for IOA coding were randomized within intervention phase, such that overall, double-coded sessions were balanced across participants and phases. Raters of the IOA sessions were masked to study phase but were aware of the overall study purpose. Overall, exact agreement for parent use of intervention strategies, across participants and phases, was moderate (M = .73, SD = .10; see Table 3). Agreements that fell below 70% were reviewed by an expert coder, consensus scores were used in final reporting, and retraining occurred (N = 8). Although item-level agreement of the exact score (0, 1, or 2) occasionally fell below optimal levels, disagreements could primarily be categorized as unitizing errors rather than classifying errors. In other words, disagreements at the item level did not result in significant overall differences in the primary outcome (i.e., the sum across items within phase constructs). For example, one rater may have provided a score of 1 for inconsistent use of a strategy on one item, and another rater may have applied this deduction to a different item, but if both items fell within the same overall strategy, then the total score would be the same. An intraclass correlation (ICC) demonstrates how similarly raters ranked sessions within phases, indicating moderate-to-strong agreement between raters (.82–.91; see Table 3). This ICC score demonstrates that although overall item level exact agreement on scoring may have had some variability, overall scores remained similar across coders. Total IOA coding of child behaviors during DB-PCX sessions was also completed on a randomized 17% of the DB-PCX sessions that were coded for presence of child behaviors (IOA M = 83, SD = 7.4).

Table 3.

Reliability of outcome disruptive behavior parent–child interaction (DB-PCX) scores.

Strategy	ICC	Confidence interval
Prevention1	.86	.68–.94
Reinforcement1	.90	.78–.96
Responding1	.91	.79–.96
Teaching1	.82	.60–.92
	M (SD)	Range
Adele and Addie2	.79 (.05)	.70–.85
Briana and Brian2	.73 (.05)	.70–.85
Carla and Charlie2	.80 (.06)	.70–.90

Note. Interobserver agreement presented in two formats.

¹Phase-level reliability (classifying) across participants for DB-PCX.

²interobserver agreement (unitizing) on item-level scores for DB-PCX. ICC = intraclass correlation.

Analysis

Visual analysis was applied to the graphed data of the parent implementation ratings across all three participants, in line with What Works Clearinghouse criteria for evidence-based practice evaluation in single-case design research (Kratochwill et al., 2010). Visual analysis methods were employed by observing changes in the data between baseline and the introduction of each phase of intervention (Ledford et al., 2018). We observed multiple aspects of the data to determine if a functional relationship exists between the introduction of the interventions and the parent's fidelity of implementation. These aspects included an observation of a change across at least 5 data points in overall level, overall trend or slope of the data, and consideration of a change in variability. Second, we considered overlapping data points between baseline and the intervention, consistency across participants, and the immediacy or delayed impact of the effect where observed.

Results

All three parent–child dyads participated in all four phases, totaling 12 sessions each. All three mothers completed Phases 1 and 2 without additional booster sessions. All three mothers received an additional booster session following Phase 3, due to not reaching a high level of strategy use (i.e., ≥ 80% implementation during DB-PCX). Adele and Carla also received a booster session at the end of Phase 4. Mothers recorded 37–39 video probes for data collection.

Figures 1, 2, and 3 present the graphed data for the three parent–child dyads during baseline and introduction, intervention, and maintenance conditions for each of the four phases. Data are presented across observations, which occurred approximately 3 times per week. Child-disruptive behaviors from observational probe data are also presented in the first graph for each figure. Data were not recorded during parent–clinician training sessions. All data are presented without trend lines, mean lines, or effect size metrics to facilitate visual analysis by the readers and avoid the introduction of any potential bias (Wolery et al., 2008).

Figure 1.

Adele and Addie. DB-PCX sessions occurred approximately 3 times per week. DB-PCX= disruptive behavior parent–child interaction.

Figure 2.

Brianna and Brian. PCX-DB sessions occurred approximately 3 times per week. PCX-DB = parent–child interaction for disruptive behavior.

Figure 3.

Carla and Charlie. PCX-DB sessions occurred approximately 3 times per week. PCX-DB = parent–child interaction for disruptive behavior.

Disruptive Behavior Strategies

During the control baseline conditions, all three caregivers demonstrated moderate–low (i.e., < 50% strategy use) implementation of positive behavior support strategies across types. Following the introduction of the intervention, a clear and immediate shift in level and overall trend was observed for Prevention Strategies immediately following the introduction of the intervention. This effect was observed consistently for all three dyads with fewer than 5 total points of overlap observed across participants between baseline and intervention.

A small immediate increase in level and overall increase in trend was observed for implementation of the Reinforcement Strategies immediately following the introduction of the intervention consistently across all three dyads. However, the third mother implemented Reinforcement Strategies with relatively high levels (60%–80% strategy use) of implementation during baseline. Thus, the magnitude of the immediate effect for this participant is smaller with the majority of intervention points overlapping with baseline. However, the overall trend still improved and the behavior was more consistent for this participant over time.

All three mothers presented a clear increasing trend in implementing responding procedures either immediately following the intervention or immediately following the booster session. However, outcome data in the Responding phase presented with some overlap between baseline and intervention, and the trends for all three participants decreased during the last three to five sessions.

Finally, the first two mothers demonstrated a clear and immediate shift in level and trend during the Teaching phase whereas the third mother presented with some variability. The lack of consistent effect across participants and overlapping data between baseline and intervention result in reduced confidence in the effect between intervention and baseline for this strategy across participants.

All three of the mothers maintained similar levels of use of intervention strategies 2 and 4 weeks following the end of the RUBI intervention. Maintenance was greatest for the Prevention Strategies across all three mothers, compared with the maintenance of the other strategies in the RUBI intervention. However, we observed variability and greater overlap between baseline and intervention data in the third dyad, likely due to maintenance as this mother had already demonstrated some of these key skills during baseline.

Overall, the mothers learned the intervention strategies, although implementation was variable at times and for certain skills. It should be noted that all mothers demonstrated improved implementation immediately following booster sessions, indicating that adding booster sessions toward improving implementation of learned strategies before presenting novel strategies may be an effective tool for improving outcomes.

Social Communication Strategies

Caregivers demonstrated moderate–high levels of responsiveness (.55–.80) following the social communication training, which decreased during the initial phases of the RUBI instruction (see Table 4). Across the RUBI intervention, responsiveness levels dropped for all mothers. Although responsiveness has decreased during follow-up observations in previous studies (Hampton et al., 2020; Roberts et al., 2014), this decrease is typically moderate and does not decrease to baseline levels, indicating that the decrease in responsiveness in this study may be greater than expected by the passage of time.

Table 4.

Mothers' use of social communication intervention strategies.

Participant	Strategy	SC baseline	SC posttest	RUBI Week 2	RUBI Week 4	RUBI Week 6	RUBI Week 8	Maintenance
Adele	Responsive strategy use	.54	.80	.51	.46	.30	.45	.48
Brianna	Responsive strategy use	.37	.64	.33	.43	.47	.41	.39
Carla	Responsive strategy use	.29	.55	.68	.42	.58	.46	.31

Note. Mothers' implementation of social communication strategy use measured during repeated 10-min semistructured mother–child interactions. Social communication strategy use is a composite score of four responsive strategies: temporal contingency, topic contingency, mirroring and mapping, and target level speech. SC = social communication; RUBI = Research Units in Behavioral Intervention.

Generalization

Parent use of RUBI strategies was observed in the DB-DOS parent context. Mothers used low rates of the strategies before instruction. Following the intervention, mothers demonstrated variable rates of RUBI strategies (see Table 5). Adele used more strategies in the generalized context than baseline, but fewer than 50% of disruptive behavior strategies were implemented correctly. Brianna increased her use of reinforcement in the generalized setting. Carla demonstrated the greatest generalization to the DB-DOS context, with high rates of implementation (i.e., ≥ 80%) for all strategies except for teaching. However, it should be noted that there are few opportunities for teaching in the DB-DOS protocol.

Table 5.

Mothers' generalized use of disruptive behavior strategies during DB-DOS parent context.

Participant	Rate	RUBI baseline				Maintenance
		Prevention	Reinforcement	Responding	Teaching	Prevention	Reinforcement	Responding	Teaching
Adele	% correct DB-PCX	0%	0%	10%	0%	30%	50%	30%	0%
Brianna	% correct DB-PCX	30%	50%	60%	0%	30%	70%	50%	0%
Carla	% correct DB-PCX	20%	20%	10%	0%	50%	80%	90%	10%

Note. Disruptive Behavior Diagnostic Observation Scale (DB-DOS; Wakschlag et al., 2008; with autism adaptations; Hampton et al., 2020) parent context scored using the disruptive behavior parent–child interaction (DB-PCX). RUBI = Research Units in Behavioral Intervention.

Child Disruptive Behavior

All three children engaged in variable rates of disruptive behavior across the DB-PCX observations. Addie engaged in the highest rates of disruptive behavior during DB-PCX observations (50% of intervals), and although her response during intervention was variable, Addie's disruptive behavior decreased meaningfully during the final phase (< 35%) and during maintenance (< 20% of intervals). Brian and Charlie presented with low levels of disruptive behavior during the baseline observations (DB-PCX), and overall levels of disruptive behavior were variable but remained low overall. Intensity of the disruptive behavior was not rated during the DB-PCX. Overall child disruptive behavior meaningfully improved for two participants from RUBI baseline to posttest with a reduction of 25% or greater in the ABC Irritability subscale (for Addie) and the HSQ Mean Severity Score and relatively large improvements during the DB-DOS examiner context (for Charlie and Addie; see Table 1).

Social Validity

Overall, all three mothers agreed or strongly agreed that the sessions were an appropriate length, the amount of instruction was acceptable, the video examples and handouts were helpful, and the setting was helpful. Additionally, the mothers agreed or strongly agreed that each of the different strategies were helpful. All three mothers reported frequent use of the strategies (much to most of the time) and that the strategies have increased their confidence in managing current and future disruptive behaviors. The percent of items rated as satisfied on the 3-point scale indicates high rates of satisfaction among the mothers (Adele = 100%, Brianna = 80%, and Carla = 97%). When asked which intervention was most beneficial, Adele reported the disruptive behavior intervention was most beneficial for her and her child, Brianna reported the social communication intervention was most beneficial, and Carla reported that both were equally beneficial. Interestingly, Carla noted, unsolicited, that she appreciated learning the social communication strategies before disruptive behavior strategies.

Discussion

The purpose of this study was to evaluate the feasibility and initial effects of the RUBI disruptive behavior parent-instruction intervention as implemented immediately following parent instruction for social communication support strategies. This is the first study to evaluate RUBI intervention by specific strategy (i.e., Prevention, Reinforcement, Responding, and Teaching). This is important because we see that caregivers were able to successfully implement the RUBI skill building strategies across different caregiving routines with their child. While mothers in this study implemented RUBI strategies with levels of fidelity in line with previous RUBI evaluations (Bearss et al., 2015), caregivers did not maintain the same rates of responsiveness demonstrated during the social communication intervention with their children during and following the RUBI intervention. This may be due to a lack of maintenance or perhaps interference of the new strategies with the previously learned strategies. For example, parents who learn to regularly praise appropriate play and engagement may need strategies to balance praise with responsiveness strategies including pausing between utterances. Additional strategies may be required to successfully sequence these intervention models for caregivers to optimize implementation, maintenance, and integration of intervention models over time, and this may also be true for other similar responsive parent-mediated interventions. How, what, and when we teach parents is critical for SLPs working with young children on the autism spectrum who are likely to be simultaneously receiving behavioral services in-home (Xu et al., 2019).

During the introduction of the fourth set of strategies, mothers' strategy use did not increase as quickly as the first three components, indicating that perhaps further practice and feedback during this fourth phase may be necessary for some parents. Variable outcomes across DB and social communication strategies may be due to a few factors. Both Carla and Adele discussed, anecdotally, their resistance candidly during the sessions over the course of intervention, and increased buy-in seemed to relate to the improvement in strategy use over time, although we have no clear measure of this over time. Additionally, the mothers in this study appeared to decrease the use of the social communication intervention strategies over the course of the RUBI intervention, which may suggest that additional support to integrate these two types of strategies is necessary.

Clinical Implications

Disruptive behavior presents an important barrier to social communication acquisition over time for young children on the autism spectrum. SLPs should consider the added benefit of a parent-mediated intervention for skill building to improve disruptive behaviors, such as RUBI, to improve and support a parent toward creating moments of joint engagement. However, parent response to the individual components of the RUBI intervention was variable across the intervention period in this study. Clinicians who implement RUBI in a clinical setting might consider progress monitoring tools to evaluate when a booster session may be optimal before moving on to novel strategies. The data from this study may also suggest that clinicians may consider routine booster sessions during Phase 4 instruction of RUBI as all the mothers in this study demonstrated additional difficultly in acquiring these skills compared with the first three phases. Furthermore, SLPs who implement RUBI in a clinical setting should work with the parents to maintain responsiveness strategies throughout implementation.

It is important to note that while educating parents and supporting social communication development in early childhood fits squarely within the SLPs scope of practice (ASHA, 2018), RUBI in isolation may not be readily covered by insurance reimbursement for SLPs. However, this fully manualized, relatively brief, intervention may be used in conjunction with social communication interventions to maximize language learning opportunities by reducing disruptive behaviors that may interfere with social communication intervention strategies. As such RUBI provides a useful solution for SLPs who often encounter disruptive behaviors with minimal training in this area (Chow & Wallace, 2021). SLPs are often the first interventionist to serve children with autism before, during, and after a diagnosis, and parents' first concerns are often disruptive behaviors; RUBI therefore offers a feasible and acceptable strategy for SLPs to address this important need during the early stages of intervention. Future work is necessary to understand how SLPs may use RUBI to transition a family to more intensive behavioral intervention, if needed, or under which conditions an SLP should implement RUBI. This deeply tailored approach to early intervention is necessary to identify these conditions, and future research should include nuanced approaches and analyses to address this need (Hampton & Chow, 2021).

Limitations

These results should be considered in light of a few important limitations. First, although there is merit and strength in the home data collection procedures, this data collection method did introduce some additional variability. For example, caregivers recorded at different times of the day and, sometimes, in different locations despite researchers requesting and discussing consistent times and locations. This method also put a large burden on the families to record the frequent observations. The results should also be interpreted considering the limitation of a narrow sample of well-educated middle-class families. Future replications should recruit a diverse sample across education and income levels to increase generalizability of these findings.

Rating positive behavior support strategies from naturalistic observations presents a number of challenges that impacted our raters' ability to maintain high levels of item level agreement. The individualized implementation of many of the strategies are dependent on the child's particular disruptive behaviors or on the parent's target goals, which may change throughout the course of intervention such that parents may ignore one set of behaviors, which they later respond to based on the child's changing needs. Additionally, the opportunities to practice certain strategies (e.g., extinction procedures) may vary greatly from session to session. Parents often obtained lower scores during sessions when children engaged in high rates of disruptive behavior. For example, Brian often demonstrated lower rates of disruptive behavior during our observations compared with what his mother anecdotally reported during the day. This can also be observed in the survey measures of disruptive behavior that we collected (see Table 1).

Another important limitation is that Charlie entered the study with an age-equivalent receptive language score of around 18 months, which is the recommended minimum receptive language level for the RUBI intervention. Although the strategies may still be effective and his mother was able to learn them, some variability in the results for this participant may be due to areas where Charlie may have benefited from additional support or modified strategies. Future research should consider adaptations of RUBI specifically designed for children in the earliest stages of language learning. Additionally, Brian and Charlie entered both with relatively low ABC scores, which may be an important consideration when determining which intervention strategy a parent might need to learn first.

It is important to note that observing disruptive behaviors during a semistructured observation (DB-PCX or DB-DOS) is complex and nuanced. Disruptive behaviors can be subtle, a trigger or antecedent for a disruptive behavior may be highly context-specific, and a parent's use of strategies may vary widely based on the opportunities for practice. For example, a parent may have fewer opportunities to practice extinction procedures if a child displays a low-rate, although severe, disruptive behavior. Alternately, a child who engages in a mild but high-rate disruptive behavior may reduce a parent's ability to find moments for praise. Therefore, there is great need for observational tools to quantify and qualify rates of child disruptive behaviors over time. For this study, we developed a novel disruptive behavior observation tool (DB-PCX) that appears to be sensitive to change in parent implementation and changes in child disruptive behavior. Observed changes on the DB-PCX over time are similar to the observed changes on the other standardized measures of disruptive behavior over the course of the intervention (i.e., HSQ, ABC).

Future Research

This study represents a first step toward identifying a more integrated approach to parent-mediated interventions for optimizing social communication outcomes for young children on the autism spectrum. While this study focuses on a sequential approach, future research should focus on building an intervention model that is adaptive to individuals changing needs by identifying the most efficient mode for combining parent mediated interventions including multiple sequences or integration of the two models. When caregivers begin early interventions for their children with autism, they are met with multiple approaches and clinicians. Although families of children with autism benefit from a wide variety of service delivery options, there is little known about how to best support caregivers to implement early intervention strategies at high levels of implementation fidelity (i.e., ≥ 80%) across intervention targets. This is likely due to the fact that parent instruction often occurs across disciplines, which may not collaborate around parent instruction. Thus, future research should specifically focus on developing parent instruction models that span different disciplines and consider optimal sequencing or integration where appropriate. A sequential multiple random assignment trial (Almirall et al., 2012) is a useful tool for optimizing intervention packages and decisions and may be particularly useful in developing intervention sequences in special education (Chow & Hampton, 2019). An adaptive intervention model could better address questions raised in this study, such as, Which children on the autism spectrum need a 12-week dedicated behavior intervention such as RUBI? or Does RUBI improve disruptive behaviors to a greater degree compared with NDBIs alone? Due to the variability observed in this study, future research should be especially mindful of how we specifically sequence or integrate the intervention approaches and tailor these approaches based on child and parent characteristics.

Conclusions

Parents of children with autism are often taught to use multiple strategies and interventions over time. To optimize child outcomes, it is important to design intervention approaches such that parents will be able to not only maintain strategy use over time but also integrate strategies across developmental intervention targets. This study demonstrates that social communication and disruptive behavior interventions can feasibly be sequenced for toddlers with parent acceptability. While the intervention did overall improve parent implementation of positive behavior supports, these results are tempered by variability in some outcomes and the decrease in social communication strategies by parents while learning positive behavior support strategies. Taken together, there is great need for future research to evaluate adaptive intervention packages that tailor intervention options based on parent characteristics, child characteristics, and in response to the changing dyad over time.

Funding Statement

This study was funded by the National Institutes of Health (R01DC014709) awarded to Megan Y. Roberts.