Teaching Requesting to Individuals with Rett Syndrome Using Alternative Augmentative Communication (AAC) Through Caregiver Coaching via Telehealth

Abstract

Rett syndrome is a severe neurodevelopmental disorder that results in both motor and language skill regression with a wide range of severity in symptom presentation. Communication intervention may be particularly challenging for this population due to the decline in speech, motor skills, and motor planning difficulties that characterize the disorder (Townend et al., 2020), often resulting in the need for augmentative and alternative communication (AAC) technology. Very limited research has evaluated communication interventions for individuals with Rett syndrome and even fewer have targeted expressive communication, an important skill required for improved autonomy and quality of life (Sigafoos et al., 2009; Townend et al., 2020). The current study sought to systematically replicate the Simacek et al. (2017) mand training procedures to teach three girls with Rett Syndrome to use AAC to make requests through caregiver coaching by researchers via telehealth. Results suggest that mand training was successful in increasing AAC use for all three participants. Barriers to intervention for this population and implications of results for future research and clinical practice are discussed.

Rett Syndrome is a severe neurodevelopmental disorder linked to the X-chromosome affecting approximately 1 in 10,000 to 15,000 girls by age 12 (Hagberg & Witt-Engerstrom, 1986; National Institute of Neurological Disorders and Stroke (NINDS), 2021). Rett syndrome occurs almost exclusively in females and often results in multiple disabilities in motor, communication, and cognitive function that are expressed in a wide range of severity in clinical manifestations (Ricceri et al., 2008). Individuals with Rett syndrome often experience communication delays that begin during a characteristic regression period that occurs between 6 and 18 months (Hagberg, 2002). The regression period is often associated with both communication and motor skill decline, resulting in complex communication needs (CCN) for many individuals (Amoako & Hare, 2020). Despite having CCN, there is growing evidence that individuals with Rett syndrome can communicate, even if through informal means (Amoako & Hare, 2020; Lim et al., 2020; Sigafoos et al., 2011) such as eye contact, various vocalizations that do not constitute words, and other idiosyncratic responses. Many parents of individuals with Rett syndrome report their children can understand most of what they say and appear to have high levels of receptive language, despite expressive language difficulties (Bartolotta & Remshifski, 2011; Urbanowicz et al., 2016). However, potentially due in part to difficulty with verbal and motor response options, there are not directly measured studies of receptive language comprehension for individuals with Rett syndrome.

In addition to possible receptive language deficits, individuals with Rett syndrome have difficulty with expressive communication, due, at least in part, to the way motor abilities are affected by the syndrome. Thus, many individuals require aided augmentative and alternative communication (AAC) to communicate successfully (Cianfaglione et al., 2015). AAC refers to any non-vocal communication that either replaces or supports vocal speech (American Speech-Language-Hearing Association, n.d.). Aided communication involves speech that is supplemented or replaced with technology (e.g., picture cards, speech generating device, microswitch, etc.). Although AAC has the potential to make communication accessible for those who are unable to rely solely on vocal speech, simply providing an AAC system is not sufficient for individuals in the process of acquiring language. In fact, special education teachers report that the majority of their students who use AAC are not proficient in the use of their communication systems (Andzik et al, 2018). Unlike modeling and imitation of natural caregiver-child interactions that support vocal language, teaching an individual to use AAC requires some level of explicit instruction.

A large evidence base supports the use of behavioral approaches to teach early AAC skills to individuals with intellectual and developmental disabilities who have language delays or impairments (Carnett, et al., 2019; Howard et al., 2005; Iacono et al., 2016; Smith et al., 2000), with a focus on the function of language (i.e., why it occurs; Ganz, 2015; Greer & Ross, 2008; Skinner, 1957; Still et al., 2014). One of the earliest and arguably most important expressive language skills acquired is requesting (mands; Skinner, 1957; Tapp et al., 2021), which is often taught first because it’s the only type of communication that contacts direct reinforcement (i.e., the individual receives access to what they request (Sundberg & Michael, 2001) and contributes to autonomy.

Although behavioral approaches to teaching communication have been established with other populations with communication impairments, a relatively small body of research has targeted communication assessment and intervention for individuals with Rett Syndrome (Lim et al., 2020; Sigafoos et al., 2009, 2011); this is despite the importance that care providers place on communication (Larriba-Quest et al., 2020; Urbanowicz et al., 2016), and its positive impact on quality of life (Brady, et al., 2016; García et al., 2020; Markham et al., 2009). Further, many communication interventions for individuals with CCN, including individuals with Rett syndrome, focus on modeling expressive communication in natural settings (e.g., aided language stimulation; ALS) and support language comprehension. However; the interventions generally do not include explicit instruction in expressive language (Elmquist et al., 2019; Sennott et al., 2016). Although behavioral interventions to teach AAC have been effective with individuals with intellectual and developmental disabilities (e.g., Harris & Reichle, 2004), they are only recently being evaluated with individuals with Rett syndrome (Wandin et al., 2021). In addition, although modeling communication interventions with a focus on receptive language development (i.e., comprehension) are important and result in communication gains (Allen et al., 2017; Sennott et al., 2016), they do not explicitly teach independent requests. Teaching individuals to independently and expressively communicate their wants and needs is a crucial addition to communication intervention to enhance autonomy and quality of life (Iacono et al., 1998; Lindgren et al., 2020; Reichle, 1997).

In one of the earliest attempts to teach individuals with Rett syndrome to independently communicate expressively, Van Acker and Grant (1995) used mand training, within a multiple baseline across responses design, to teach three girls with Rett Syndrome to request preferred items using a touchscreen device. Two of the three participants demonstrated increases in requesting in all three targeted communication responses and demonstrated discrimination when choosing between preferred and non-preferred items. The third participant showed increases in requesting one of the targeted communication responses. Sigafoos et al. (1996) found similar results when they used mand training within a multiple baseline across participants design to teach four girls with Rett syndrome to request preferred objects. Three of the four participants demonstrated increases in requesting using a picture card across the study. These early studies demonstrated that despite CCN and severe motor impairment in many cases, individuals with Rett syndrome can be taught to expressively request preferred items.

In a more recent study targeting mands, Byiers et al. (2014) taught three individuals with Rett syndrome to use a microswitch as a competing response to interfering behavior that served the same function. Researchers conducted a functional analysis to identify the functional reinforcer maintaining the interfering behavior and then used functional communication training (FCT) to teach the individual a function- matched communicative response (pressing a microswitch that played a recorded vocal request for the functional reinforcer). Researchers used a reversal design to demonstrate experimental control and successfully taught all three participants to use microswitches to request more frequently than to engage in interfering behavior. Similarly, Simacek et al. (2017) used mand training to teach two individuals with Rett syndrome to request multiple items using speech generating devices. Researchers implemented sessions with the assistance of parents in the participants’ homes, and taught participants to activate the device with either a button press or eye-gaze to request items. Using a multiple baseline across requests design to demonstrate experimental control, both participants learned to request three preferred items. The results of the handful of studies evaluating using mand training to teach independent requesting to girls with Rett syndrome are promising but additional replications are needed.

In conjunction with further evaluation of interventions that effectively teach communication to individuals with Rett syndrome, there is a need to identify ways to make this intervention more accessible. Rett syndrome is a rare disorder that requires specific expertise in multiple domains to successfully manage (Townend et al., 2020). Yet there are relatively few clinics/professionals with the necessary expertise in relation to the number of individuals diagnosed with Rett syndrome (Rett Syndrome Research Trust, 2022; International Rett Syndrome Foundation, 2022). This means that access to expertise in Rett syndrome treatment, including communication instruction, may be limited for many individuals who need it. Telehealth technology is a promising approach for reducing geographic barriers to needed services (Akemoglu et al., 2020; Drew et al., 2022; Marino et al., 2020; Lindgren et al., 2020; Machalicek et al., 2016; Suess et al., 2016; Unholz-Bowden et al., 2020).

Telehealth refers to the use of technology to provide clinical services to individuals who are in geographically different locations than the healthcare provider (California Association for Behavior Analysis [CalABA], 2017; Kane & DeBar, 2022; U.S. Department of Health and Human Services, 2021). Telehealth, in the context of providing behavior-based services, typically involves trained therapists providing coaching to caregivers on how to implement behavioral procedures with the individuals receiving the intervention (Kane & DeBar, 2022; Neely et al., 2017; Tomlinson et al., 2018; Unholz-Bowden et al., 2020). In the case of communication-based services, telehealth allows for communication instruction to take place in natural settings, which can help promote the maintenance and generalization of skills. Previous research has demonstrated successful implementation and effectiveness of communication interventions by caregivers coached via telehealth (Neely et al., 2021; Unholz-Bowden et al., 2020). For example, in studies by Simacek et al. (2017) and Dimian et al. (2018), parents were taught to implement functional communication training (FCT) to teach children to use a speech-generating device (Dimian et al., 2018), communication cards, or microswitches (Simacek et al., 2017) to make requests. In both studies, parents were coached via telehealth to teach their children to use AAC with high procedural fidelity, with an average range of 92% to 96% across FCT sessions. In the current study, telehealth is utilized as the delivery model and the effectiveness of mand training provided by caregivers coached via telehealth on AAC acquisition is evaluated.

The purpose of the current investigation was to systematically replicate the Simacek et al. (2017) procedures to teach three individuals with Rett Syndrome to request preferred items using telehealth technology for service delivery. Our research sought to answer the following research question:

1. What are the effects of mand training (procedures similar to Simacek et al., 2017) implemented by caregivers coached via telehealth in natural settings on individuals with Rett syndrome’s use of AAC to request preferred items?

Method

Participants and Setting

Three individuals with Rett Syndrome were recruited as part of a National Institutes of Health (NIH)/National Institute on Deafness and Other Communication Disorders (NIDCD) Grant (Grant No. 1R21DC015021) from Natural History study sites across the United States, the Midwest Rett Syndrome Foundation (MRSF), the International Rett Syndrome Foundation (IRSF), and Rett syndrome community events. Detailed demographics are provided in Table 1. Cinna was a 1.75-year-old white female who was ambulatory with a walker and could self-feed. Prior to the onset of the study, she communicated with her caregivers by reaching, moving toward objects, yelling, or crying. She followed one step instructions some of the time and rarely engaged in pre-speech sounds. The target communication response for Cinna was a switch press with the vocal output messages “snack,” “toys,” and “all done.” Blythe was a 3.5-year-old white female who was ambulatory with a walker and could self-feed. Prior to the study, she communicated by reaching, looking at items, and vocalizations. Parents reported that vocalizations were pre-speech sounds used to gain attention, or indicate she wanted to stop or start an activity and that she rarely followed 1-step instructions. The target communication response for Blythe was a switch press with the vocal output messages “food,” “toy,” and “video.” Christina was a 29-year-old white female who was ambulatory with assistance and unable to self-feed. Before the study, she communicated with caregivers through infrequent loud non-speech sound vocalizations, looking at objects, eye contact, and body posturing. Parents reported that she responded to her own name by making eye contact, and sometimes followed simple instructions. The target communication response for Christina was a switch press with the vocal output messages “tablet please,” “hey there,” and “no thank you.”

Table 1.

Participant demographics. Age in years, With assist. = With assistance

Participant	Age	Diagnosis	Gender	Race/Ethnicity	Ambulatory?	Eat Indep?
Cinna	1.75	Rett	Female	White	With assist	Y
Blythe	3.5	Rett	Female	White	With assist	Y
Christina	29	Rett	Female	White	With assist	N

All sessions were conducted via telehealth video conferencing in different parts of participant homes including the living room, kitchen, and basement. Caregivers choose where videoconferencing equipment was set up and researchers made suggestions as needed to ensure observation was possible (e.g., placing the camera higher and angling it downwards).

Researchers and Interventionists

Researchers were three white, non-hispanic female Ph.D. students in the field of special education: one had their BCBA credential, and one had their BCaBA credential. All three researchers had been coaching caregivers via telehealth for at least nine months prior to initiating coaching with these participants. Both mother and father were interventionists for Cinna and Blythe. Both girls’ parents were white, non-hispanic and did not have any prior training in behavior analysis or communication training. The interventionist for Christina was her mother; she was also white and non-hispanic and did not have any prior training in behavior analysis. She had some prior communication training from a Speech Language Pathologist who worked with the family.

Materials

Videoconferencing Equipment

Researchers interacted with families via live videoconferencing using a high-speed internet connection through either Google Meet or Zoom platforms. Researchers created a Gmail address for each participant so that families could communicate and connect with the research team via Google Meet or Zoom. For most sessions, researchers used a computer (Dell OptiPlex 3010 Desktop) on campus in the laboratory with a Dell 24″ monitor. A Logitech HD Pro Webcam C920 on a Polaroid 8″ Heavy Duty Mini Tripod was used to capture image and stream video, and a Logitech ClearChat Comfort/USB Headset H390 was used to capture sound (i.e., researcher talking) and stream audio of both researcher and caregiver/participant. In a few cases, researchers used personal laptops with built-in cameras and microphones to conduct sessions. If personal laptops were used, researchers connected to the internet via the University’s virtual private network (VPN). Researchers video recorded all sessions using Zoom recording or Debut Video Capture software. All data and video recordings were stored on the University’s secure encrypted Box server. All families had high-speed internet service to their home prior to the onset of the study and most families used the video camera, speaker, and microphone on their laptop computer, electronic tablet, or telephone to connect during sessions. Researchers mailed or delivered loaner equipment to any families who did not have the required videoconferencing equipment for use during the study, depending on their home’s proximity to campus. Equipment loaned to all families included an ASUS Chromebook C200M with built-in microphone, Logitech HD Pro Webcam C920, a Polaroid 8″ Heavy Duty Mini Tripod, and microswitches for use as an AAC device.

Communication (AAC) Equipment

The communication modality for all three participants was AAC in the form of manually-activated single-message microswitches. Microswitches were chosen as the initial communication modality for a variety of reasons. First, previous research has demonstrated success using microswitches to teach with individuals with Rett syndrome to request. Second, motor abilities of the participants were taken into consideration as all three participants had limited fine motor skills. Finally, as this was the first exposure to AAC for each participant, evidence supports teaching one request at a time on a microswitch as being a simple and preliminary step to demonstrate that the girls can communicate simple requests prior to identifying an AAC device to teach more or complex communication skills (Byiers et al., 2014). In fact, two of the three participants (Cinna & Blythe) went on to learn to use an eye-gaze device after completion of the study.

Researchers mailed or personally delivered micro-switches with laminated picture cards that were attached with velcro to the top of the switch to all families for use as AAC during the study. Tangible items, preferred activities, and edible reinforcers were all individually identified within naturally-occurring routines within each participant’s home.

Dependent Variables

For all participants, the primary dependent variable included targeted AAC responses, which were selected based on preferred activities identified via preference assessments as well as parent report of any activities, such as crafts completed at a day program, that participants disliked but did not have a functional way to refuse. Operational definitions were individualized for each participant and are provided in Table 2. For all three participants, the target communication modality was a microswitch and the targeted communicative response was a microswitch press. For Blythe, the microswitch was programmed with the vocal output messages “food,” “toy,” and “video” to request access to food, toys, and videos respectively. The vocal output messages programmed for Cinna were “snack,” “toys,” and “all done” to request access to food, toys, and the termination of an activity respectively. Christina’s programmed vocal output messages were “tablet please,” “hey there,” and “no thank you” to request access to the tablet, attention, and to terminate an activity respectively.

Table 2.

Idiosyncratic and target communication operational definitions for each participant

Participants	Idiosyncratic Comm. Operational Definitions	Target Comm. Operational Definitions
Cinna	Reach: Any instance of Cinna moving her hand or arm toward an object or person Cry: Any instance of Cinna making sounds Yell: Any instance of Cinna making sounds in a volume that could be heard from the next room	Micro-switch press: Any instance of Cinna independently placing pressure on the switch with her hand or arm with enough force to play the voice output message
Blythe	Reach: Any instance of Blythe extending either both hands and arms or one hand and arm towards an object or item Vocalization: Any instance of vocal sounds, including crying, moaning, “mmm”, and “ah”. This does not include spitting/blowing sounds Look: Any instance of looking, turning eyes (if visible) or head towards parent, and pausing for at least 1 s, not including scanning room or objects	Micro-switch Press: Any instance of Blythe placing pressure on the switch with her hand or arm with enough force to play the voice output message
Christina	N/A	Micro-switch Press: Any instance of Christina placing pressure on the switch using any part of her body with enough force to play the voice output message

For two participants, Cinna and Blythe, a second dependent variable included idiosyncratic communication response(s), which were identified during pre-intervention Structural Descriptive Assessment (SDA) sessions and selected based on which responses occurred most frequently. Idiosyncratic and target communication operational definitions for each participant are described in Table 2. When multiple idiosyncratic responses occurred at similar rates, all responses were measured and combined to calculate response rates or percentage of responses. For Christina, we did not track an idiosyncratic response because while some were reported by caregivers in the initial interviews, none were not observed during the SDA or baseline sessions.

Measurement and data recording systems were determined individually for each participant. Researchers collected data for all sessions either live or after sessions from the recorded video. For two participants (Cinna and Christina), the number of trials with independent switch presses and occurrences of idiosyncratic responses were measured using event recording during trial-based instruction. For each trial, the occurrence of switch presses and/or idiosyncratic responses were coded. Percentage of switch presses and idiosyncratic responses were calculated by dividing the number of trials with occurrences by the total number of trials and multiplying by 100. Christina’s mastery criterion was set at three sessions at 80% or higher independent responding and Cinna’s was set at two sessions with 80% or higher independent responding. For Blythe, the frequency of switch presses was measured using an event recording system during 3-min sessions. Frequency of responding was measured instead of percent of independent trials because her communication goals were initially to increase response rate of switch pressing, rather than independence, to effectively compete with the rate of idiosyncratic responding. Because the goal was not independence but rather rate of responding, introducing the next message in the multiple baseline across messages for Blythe was determined based on an increase in switch pressing accompanied by a decrease in idiosyncratic responses. Intervention for the next message was implemented when researchers observed a stable level of switch pressing and no overlap with the idiosyncratic response data path for the preceding message.

Procedures

All procedures in the research project were approved by the University Human Research Protection Program/Institutional Review Board-Biomedical and Social/Behavioral Research.

For all assessment and intervention sessions, researchers provided written instructions in advance of the session. At the start of the session, researchers reviewed the procedures and parents were given the opportunity to ask questions. During sessions, researchers coached parents to implement procedures and provided direction and feedback throughout. After each session, researchers discussed what was observed during the session and briefly described the procedures for the next session.

Informed Consent

The informed consent process to join the research study was conducted via phone or videoconferencing with each participants’ parent(s) prior to the onset of the study. Researchers involved in the consent process scheduled a meeting with interested parents and provided the written consent document to review. During the consent meeting the researcher described the study in detail and asked the parents if they had any questions throughout. After all questions were answered, the researchers told the parents they could take time to consider participation, and to send back the signed form if they decided to participate. Parents were encouraged to reach out with any questions that came up after the meeting. Consent was continually assessed informally throughout participation by checking in with parents regularly on the research process, interventions, and participant progress.

Assent

Researchers were unable to obtain formal consent or assent from the participants because of limited communication repertoires, nonetheless, assent was assessed informally throughout the research process based on participants’ behaviors during sessions. Researchers stayed alert to behaviors that indicated assent withdrawal for all participants. Behaviors to be observed as indicative of assent withdrawal included repeated attempts to leave the session area even after adjustments were made, crying continuously or actively protesting, or refusing to engage with procedures. Researchers also honored all caregiver requests to terminate sessions and checked in with caregivers when any participant behavior was observed that suggested assent withdrawal behavior. When researchers observed behaviors that suggested assent withdrawal, the session was immediately terminated and rescheduled. Although data on behaviors interpreted as assent withdrawal/dissent were not systematically collected during sessions, anecdotally, assent withdrawal occurred infrequently for only one participant (Cinna), which often appeared as fussiness and may have been due to her young age. If behaviors that indicated dissent were observed for three sessions in a row, researchers scheduled a meeting with caregivers to collaborate on adjustments to the intervention procedures and procedures were adjusted accordingly. If intervention adjustments did not result in assent, researchers in collaboration with parents would have terminated participation, however this did not occur for any of the participants.

Parent Collaboration and Goal-Setting

Research, by definition, is an exploratory process, and as such, participants may not benefit from involvement. Even given this, researchers sought to make assessment and intervention procedures as relevant and contextualized to each participant as possible. Researchers worked collaboratively with participants’ parents to establish a bi-directional relationship in which they were encouraged to speak openly and provide feedback about the assessment and intervention procedures. Parents were involved in providing input on the contexts for the researchers to observe during assessment and selecting specific intervention targets and goals. Through this relationship, culturally relevant intervention targets and goals related to the current project (see Baseline section) were established collaboratively with each family.

Interviews

Researchers interviewed parents to gather information regarding the participants’ current communication forms and potential functions using the Inventory of Potentially Communicative Acts (IPCA; Sigafoos et al., 2000). During this interview, information was also gathered regarding participant interests and routines.

Technology Check

After structured interviews were completed, researchers connected with parents via the internet to test technology and address any issues with connectivity, audio, or video. During the technology check parents also gave researchers a virtual tour of the locations in the home where sessions might be conducted. Researchers offered suggestions for location of the camera to ensure appropriate angle, lighting, and video clarity. If wireless internet connection did not provide consistent connectivity, or sufficient picture or sound quality, researchers mailed the family an ethernet cable to directly connect to the internet.

Unstructured Observations

After the technology check, researchers observed family members interacting with participants during a variety of naturally occurring routines (e.g., mealtime, unstructured play, daily living routine etc.). During observations, researchers coded both the form and occurrence of idiosyncratic communicative responses during interactions with family members. Researchers also took notes on gross and fine motor ability to inform selection of appropriate AAC devices. Data collected during interviews and observations as well as parent input were used to inform the structured descriptive assessment that followed.

Structured Descriptive Assessment and Preference Assessment

After unstructured observations, researchers coached caregivers to conduct structured descriptive assessments (SDAs; Anderson & Long, 2002) and paired choice preference assessments with all participants (Fisher et al., 1992). During the SDA, researchers coached parents to manipulate antecedent variables to identify which idiosyncratic responses occurred during various situations identified collaboratively as important routines for communication. Researchers directed parents to arrange conditions and then respond to the participant the way they normally would if the coach was not observing. Conditions arranged included 1) play, 2) tasks (e.g., making trail mix), 3) preferred item out of reach, and 4) parent attending to other activity (e.g., answering emails). All SDA conditions were five minutes long and no more than 10 sessions occurred per day.

Paired choice preference assessments were conducted as described by Fisher et al. (1992). The researcher coached parents to present two items at a time until all combinations of 8–10 items were presented. The researcher coached parents to counterbalance item location by swapping items from left to right after each choice. When the participant selected an item, the parent provided the item to interact with briefly or consume (if edible) and then presented the next pair of items. Sessions were conducted until preferences were stable across sessions and a clear hierarchy of preference was demonstrated. Because Rett syndrome is sometimes associated with delays in motor planning, researchers coded how long it took participants to select a preferred item during preference assessments and used the longest length of time as prompt delays during mand training.

Baseline

Researchers discussed with parents the data gathered from assessments to select the target communication responses, the context in which mand training was conducted, and the communication modality for each participant. In making selections, researchers met with parents to discuss what was learned about their child’s current communication goals, activities identified as being highly preferred, as well as the parents’ goals for their child. Communication goals for each participant were developed by aligning assessment results and parent goals. To illustrate, Cinna’s SDA identified idiosyncratic responses resulted in access to food and toys. Communication goals related to formally requesting items from food and toy categories were developed with parent input. Further, specific foods and toys targeted for training were verified by using preference assessments, which confirmed the items Cinna would be most motivated to request. The contexts for mand training were chosen based on situations in which target communication responses were most relevant for the participants. Based on information gathered from parents and observations during the assessment, parents and researchers collaborated and agreed on the most appropriate communication modalities to teach based on the physical capabilities of the child participants.

After target communication responses were selected, researchers coached parents to implement 5-trial baseline sessions. During all baseline sessions, the microswitch was within reach of the participant. For each trial, the parent first provided access to the targeted reinforcer. After approximately one min, or the participant consumed the edible, the parent moved it out of reach of the participant and stated, “Ask if you want more.” or “If you want more, let me know.” Parents then provided the targeted reinforcer only if the participant engaged in an idiosyncratic response. If the participant did not engage in an idiosyncratic response (as was the case for all trials with Christina), the trial was reset after 20 s. If the participant pressed the switch, the parent withheld reinforcement and said, “Can you ask me another way?” and only provided the targeted reinforcer contingent on the idiosyncratic response (or if 20 s passed without a response for Christina). Switch pressing was not reinforced during baseline sessions because reinforcement is a component of mand training and we did not want a component of intervention present in baseline. Researchers conducted baseline probe sessions with all participants on all communication responses from the onset of the study until intervention was introduced to the next response. Researchers continued baseline probe sessions for the first response taught for at least three sessions or until stable responding occurred. For each subsequent response taught, researchers continued baseline probe sessions until the previous response in intervention was mastered, staggering intervention application across responses.

Mand Training/Independent Variable

After baseline, researchers coached parents to implement mand training (similar to the procedures conducted in Simacek et al., 2017) for the target communication response of using the microswitch as the independent variable. For all three participants, preferred items/activities identified in the preference assessments were selected as initial mand targets. For Cinna and Christina, a protest response was also targeted. Sessions for young children (Cinna and Blythe) were shortened due to their age. Cinna’s sessions lasted approximately 5 min but ranged depending on responding (r = 4 min 50 s-6 min 41 s) and Blythe’s lasted for 3 min. Given that Christina was an adult, sessions lasted until 5 trials were completed. Christina’s session lengths ranged from 2 min 20 s to 10 min 9 s.

The goal of mand training is to establish the response-reinforcer relation for the targeted mand, in other words, to bring the target mand into frequent contact with contingent meaningful reinforcement. The target mand (request) via AAC is honored by providing the reinforcer. This is taught systematically through initially prompting the AAC response and then fading prompts over time. Researchers coached parents to implement procedures similar to those described in Simacek et al. (2017). Specifically, researchers coached parents to implement the following steps: 1) Set up motivation for the targeted response, 2) provide the opportunity for the participant to respond, 3) prompt the participant to respond or provide error correction if necessary and, 4) provide the targeted items contingent on the target response. Access to reinforcers for requests was provided for approximately 30 s for tangible items and until the edible item was consumed. For example, if the target response was pressing a microswitch to compete with the idiosyncratic response of reaching to request a cookie, parents began trials by providing a small amount of the cookie to establish motivation. After the bit of cookie was consumed, parents removed the remainder of the cookie and placed it out of reach but within sight of the participant and waited for the participant to respond. When the participant pressed the microswitch that activated the vocal output “cookie,” parents provided the cookie. This process was repeated for all remaining trials in the session. Sessions continued in this fashion until participants reached the mastery criterion described above. For Blythe, intervention was applied to the second target response after differentiation between idiosyncratic and target responses occurred during intervention on the first, and so on until all responses showed differentiation in intervention. Detailed information on prompting procedures for each participant are displayed in Table 3. The prompt hierarchy used by parents were least to most for Cinna and Blythe and most to least for Christina and included a full physical and a partial physical prompt.

Table 3.

Prompting procedures and topographies

Participant	Prompting Procedure	Prompt Topographies
Cinna	Least to Most	Point, Partial Physical, Full Physical
Blythe	Least to Most	Point, Partial Physical, Full Physical
Christina	Most to Least	Full Physical, Partial Physical

Coaching

All mand training was implemented by the parents in their home while being coached via telehealth by the researchers. Prior to the first session, parents were provided with written instructions describing mand training procedures. Parents were given time to read the instructions prior to the first session and time to ask questions. When parents met with researchers remotely for live sessions, researchers vocally reviewed the instructions and, if needed, provided a model, before parents implemented the procedures. As parents conducted mand training during sessions, researchers provided vocal prompts if parents needed help with specific steps of the procedures. Researchers also provided performance feedback both during sessions when parents implemented steps correctly, as well as following sessions in the form of praise for what went well and any improvements for future sessions. As parents became more fluent in the procedures, researchers were able to provide fewer prompts and less feedback.

Maintenance/Follow Up

Follow up sessions were conducted one and three months after mastery with Cinna and once a month for 4 months after mastery with Christina. Follow up sessions were identical to mand training sessions. Although we reported on independence for Cinna and Christina during maintenance/follow up sessions, parents provided a prompt after a delay to use AAC to make a request. Researchers did not provide corrective feedback in these cases (i.e., request parents withhold the promt) to preserve social validity because families naturally responded by prompting the child when the child did not produce a request and the parent expressed discomfort with withholding prompts.

Procedural Fidelity

Observers collected procedural fidelity data on an average of 67% (R = 33%−97%) of recorded baseline and mand training conditions across the three participants. Researchers created a procedural fidelity checklist that included the following: 1) setting up motivation for the response, 2) providing an opportunity for the participation to respond, 3) followed prompt hierarchy when necessary (intervention), and 4) providing the reinforcer for either the idiosyncratic response (baseline), or the target response (intervention) and 5) withholding the reinforcer for the idiosyncratic response (intervention) or target response (baseline). Researchers then used the checklist to evaluate procedural fidelity. Prior to coding experimental data, observers practiced using the procedural fidelity checklist to code baseline and mand training sessions from recorded sessions from a different study until they met a 90% accuracy criterion. Average procedural fidelity was high across all participants with an average of 96% (R = 92%−99%). For Blythe procedural fidelity averaged 99% (R = 98%−100%), for Cinna it averaged 92% (R = 75%−100%), and for Christina it averaged 96% (R = 60%−100%).

Interobserver Agreement

An independent observer who was trained on session procedures and measurement of responses coded at least 25% of all mand training baseline and intervention sessions and data were compared to determine interobserver agreement (IOA). An agreement was recorded when both primary and secondary observers coded an occurrence/nonoccurrence of idiosyncratic and target response for each trial in a session for trial-based sessions. For response-based sessions, (Blythe) the observation period was divided into 10 s intervals and occurrence/nonoccurrence for each response type was compared for each interval. Any trials or intervals where occurrence/nonoccurrence did not match across observers was coded as a disagreement. Session IOA was calculated by dividing the number of trials/intervals with agreement by the total number of trials/intervals (agreements + disagreements) and multiplying by 100 to obtain a percentage of agreement. Average agreement for the three participants across baseline and intervention sessions was 96% (R = 67%−100%).

Experimental Design and Analysis

Multiple baseline (Cinna and Blythe) and multiple probe (Christina) designs across communication messages were used to demonstrate experimental control for all three participants. For Blythe, a reversal design was embedded within the first communication response taught (i.e., toy). Parents logged on to telehealth meetings with researchers two or three times per week and during each meeting, they ran one to seven sessions. Research sessions spanned two months for Blythe, 5 months for Cinna, and nearly 7 months for Christina, including follow up sessions. Researchers graphed all session data and phase change decisions were made based on visual inspection of the graphs. For Christina and Cinna, after a target communication response was mastered, researchers introduced intervention to the next response until mastery criterion for that response was met, and so on, until intervention was completed for all responses. For Blythe, intervention was introduced for the second and third responses after clear differentiation appeared between idiosyncratic and target responses in the preceding responses (first and second responses, respectively).

Acceptability

Researchers assessed the acceptability of intervention procedures by providing caregivers with the Treatment Acceptability Rating Form- Revised (TARF-R; Reimers et al., 1991) after the investigation was complete with their child. The TARF-R requires the respondent to rate statements about feasibility and preference for treatment on a 7-point Likert scale (1 = low, 7 = high except for 5 items which are reverse-scored).

Results

Results of mand training for each participant are depicted in Figs. 1, 2, and 3. For Blythe, across all three communication responses during baseline sessions when parents reinforced idiosyncratic responses, idiosyncratic responses were more frequent than switch pressing. By contrast, during intervention sessions when use of AAC produced reinforcement and idiosyncratic responding did not, switch pressing occurred more frequently than idiosyncratic responding. In the first leg of the multiple baseline across requests design (toy), when the parent returned to reinforcing idiosyncratic responding, that response occurred at higher levels (M = 4.8, R = 4–5) than switch pressing (M = 2.4, R = 0–5). Finally, when the parent reinforced mands rather than idiosyncratic responding, use of the microswitch rapidly became more frequent (M = 4.2, R = 3–5) than idiosyncratic responding (M = 0.6, R = 0–1). In the second leg of the multiple baseline design (videos), when parents reinforced idiosyncratic responses in baseline, they occurred at higher levels (M = 4.5, R = 4–5) than switch pressing (M = 1, R = 0–4). When parents began reinforcing mands instead of idiosyncratic responding, higher levels of mands occurred (M = 4.5, R = 2–5) with idiosyncratic responses remaining low (M = 1, R = 0–2). In the third leg of the multiple baseline design (food), switch pressing occurred at a higher rate (M = 7.7, R = 4–10) than idiosyncratic responses (M = 3.3, R = 3–4) within the first three sessions. Researchers hypothesized this may be due to the placement of the switch on the food tray close to her body and directly below her arms. Blythe engaged in a variety of hand movements throughout sessions, and pressing appeared to be accidental. When the switch was moved to the table next to Blythe during session 4, switch pressing decreased and remained low throughout the remainder of baseline (M = 2.7, R = 0–10). Importantly, after parents began reinforcing switch pressing, mands increased without adjusting the placement of the switch (M = 4.1, R = 3–5), suggesting switch pressing early in baseline was likely accidental.

Fig. 1.

Frequency of idiosyncratic responses (squares) and mands (circles) across three requests for Blythe. * = moved switch to table

Fig. 2.

Frequency of idiosyncratic responses (squares) and mands (circles) across three requests for Cinna

Fig. 3.

Frequency of mands across three requests for Christina. * = 10 s prompt delay ** = 15 s prompt delay, *** = increased reinforcement for independent responses

For Cinna, across all three communication responses, when idiosyncratic responses were reinforced with the preferred item in baseline, idiosyncratic responses remained high and switch presses remained low. In the first leg of the multiple baseline (“snack”), when parents implemented mand training, independent switch pressing increased (M = 78.5%, R = 0%−100%) and idiosyncratic responses decreased (M = 28.3%, R = 0%−57.1%) relative to baseline but remained variable until differentiation occurred in the last two sessions. Cinna met mastery criteria for “snack” in 12 sessions. High levels of independent switch pressing to request “snack” maintained at both 1- (M = 89.8%, R = 78.6%−100%), and 4-month (M = 82.0%, R = 50%−100%) follow ups. Similarly, when parents implemented mand training for the second response (“no thank you”), clear differentiation occurred quickly, with high levels of independent switch pressing (M = 95.6%, R = 83.3%−100%) and lower levels of idiosyncratic responses (M = 44.9%, R = 16.7%−71.4%). She met mastery criteria for the “no thank you” response quite quickly, within 4 sessions. High levels of independent responding remained at both 1- (M = 92.9%, R = 100%) and 4-month (M = 100%, R = 100%) follow ups. Finally, in the third leg of the multiple baseline (“toy”) high independent levels of switch pressing (M = 42.2%, R = 0%−100%) occurred during three sessions of baseline sporadically but reduced significantly prior to intervention. When parents began mand training, independent switch pressing increased rapidly (M = 88.9%, R = 66.7%−100%) and remained stable; Cinna met mastery criteria for “toy” in only three sessions. High levels of independent manding maintained at 1- (M = 91.7%, R = 75%−100%) and 4-month (M = 88.9%, R = 66.7%−100%) follow ups.

Christina’s switch presses in baseline for the first leg of the multiple baseline (“tablet”) remained at zero for three consecutive sessions. Upon implementation of mand training, independent switch presses increased (M = 24.9%, R = 0%−80%) but remained variable for 99 sessions. During intervention, procedures were modified on three different occasions in an attempt to increase independent responding. In session 15, the prompt delay was increased from 5 to 10 s; a second increase from 10 to 15 s was implemented on session 42. On session 102, reinforcement provided for correct responses was increased from 30 s access to the tablet to 60 s of access. Immediately after this change, independent switch pressing increased (M = 70%, R = 40%−80%) and mastery criteria was met within 4 sessions (106 total sessions). Independent switch pressing for “tablet” remained high after an initial drop during 3 months of maintenance (M = 76.7%, R = 40%−100%). Parents implemented the increased access to reinforcement procedures when implementing mand training for the 2^nd and 3^rd responses. For Christina’s second communication response (“hey there”) independent switch presses remained at zero throughout baseline and increased steadily after mand training was introduced (M = 38.8%, R = 0%−100%). She met mastery for the “hey there” response after 33 sessions and high independent responding maintained for the following month (M = 100%, R = 100%). For the third response (“no thank you”) baseline levels of switch pressing remained low (M = 6.7%, R = 0%−20%) until she began to increase independent responding on the second response. At this point, independent switch pressing increased (M = 80%, R = 60%−100%) and within three baseline probes she had 100% independent switch presses. Independent responding decreased to 40% on session 138 and as a result mand training was initiated. Independent switch pressing increased immediately following intervention and mastery criteria was met within 3 sessions (M = 93.3%, R = 80%−100%). After intervention, researchers coached parents to conduct novel activity probes using a 2^nd non- preferred activity (stirring brownie mix). Independent switch pressing remained at 100% for all three novel activity probes.

The TARF-R was returned by two of the three parents. The results of TARF-R suggested that overall families found intervention procedures highly acceptable with an average rating of 6.5 out of the highest possible score of 7 across participants. Scores by participants were 6.29 for Blythe, and 6.8 for Cinna.

Discussion

The results of the current study successfully replicated the mand training procedures utilized in Simacek et al. (2017), and add to the evidence base that mand training procedures are an effective approach for teaching individuals with Rett syndrome to independently request preferred items. The goal of mand training was to teach the participant a formal communicative response to use more often than the idiosyncratic communication response. All three participants in the study acquired three functional mands during training, with one participant (Christina) acquiring the third mand without explicit training. These results are consistent with previous research (Simacek et al., 2017) and further demonstrate individuals with Rett syndrome can learn to expressively communicate independently with the aid of AAC and systematic instruction. Not only were both young children (Cinna and Blythe) able to acquire all mands taught, but they acquired each relatively efficiently with an average of 5.5 sessions (range: 3–12) to mastery for Cinna or until stable differentiation between responses (i.e., 3 sessions differentiated in a row) for Blythe. Although acquisition of targeted responses took longer for Christina, the only adult participant (m = 52 sessions; range = 17–106), she appears to have demonstrated generalization of the teaching procedures for the 3^rd response taught. This is encouraging and suggests that programming for generalization may increase efficiency of acquisition for learners with Rett syndrome. Although generalization was not explicitly targeted or measured in the current study, future research should explore the role of programming for generalization on the acquisition of communication skills. It’s encouraging that after acquisition of the first response, the number of sessions before acquisition of the second and 3rd responses were significantly less (33 [“hey there”] and 17 [“no thank you”] respectively). Another factor that may have facilitated the acquisition of the “no thank you” response without explicit training may have been Christina’s motivation to end the undesirable activity. Prior to the introduction of the switch, parents reported that Christina’s response to arts and crafts activities in school and her day program was to look away and slump her shoulders, which parents interpreted as her protesting. This response was not often noticed or interpreted as protesting by staff, so it’s likely Christina completed those activities often prior to communication training. Further, anecdotally the latency to pressing the switch for the arts activity was much less than the other two responses and in the few sessions before mastery Christina began grunting when the activity was presented. Protest responses are acquired early for neurotypical learners, and yet, to our knowledge have not been targeted in previous mand training studies with individuals with Rett syndrome. Teaching a protest response is important for an individual to develop agency and should be targeted as a communication response in future research.

Although these results are encouraging, much more research is needed on teaching individuals with Rett syndrome to communicate. First, given the importance of independent expressive communication on autonomy and quality of life (Wehmeyer & Schwartz, 1997), future research should continue to investigate the utility of explicit instruction procedures to teach individuals with Rett syndrome. These investigations might include evaluating gains made in expressive language after mand training when compared to other approaches that do not include explicit expressive communication instruction (e.g., Aided language stimulation) or compared to other common early AAC support approaches implemented with girls with Rett syndrome (e.g., partner assisted scanning or modeling alone). Teaching both language comprehension and independent expressive language are important for individuals with Rett syndrome to gain autonomy and research that adds to our understanding of which combination of approaches are most successful is extremely important. Second, more research is needed on how we identify communication targets for this population. The current study targeted simple requests based on caregivers’ report that participants lacked this skill, and low levels of requesting in baseline. This is consistent with the current literature on independent expressive communication for individuals with Rett syndrome which includes only a handful of studies and is focused almost exclusively on teaching requesting (Amoako & Hare, 2020). This is unsurprising given the importance of this early communication skill, however, future research should investigate procedures to expand repertoires to include complex communication, such as conversations, social interactions, and academic behavior.

One challenge to communication intervention for individuals with Rett syndrome, and particularly expanding communication skills, is determining where to start. Assessing language skills and cognitive abilities can be difficult because standard language assessments require vocal or motor responses, both of which may be impacted by the disorder (Byiers & Symons, 2012; Ward et al., 2021). Although we did not assess receptive language in the current investigation, future research should seriously consider developing these assessments for several reasons. Without reliable and valid language assessments, implementing communication intervention with confidence can be more difficult. Teaching communication beyond requesting is likely to require an accurate assessment of communication skills in Rett syndrome. Although there are some assessments that evaluate communication as one portion of the assessment (e.g., Vineland Adaptive Behavior Scales), very few assessments involving direct observation of communication skills for individuals with Rett syndrome have been developed (The GAIRS checklist, Fabio et al., 2022, and Non-Speech Test, Huer, 1983 [revised to be the TECEL: Test of Early Communication and Emerging Language; Huer & Miller, 2011], are exceptions; Townend et al., 2020). Even fewer provide results that allow practitioners to target specific communication behaviors or align well with an explicit instruction or behavioral approach to communication intervention. Assessments that involve direct observation of behavior are important for most efficiently identifying appropriate targets and overcoming a number of concerns related to data obtained by individuals self-reporting what they’ve observed after the fact. Behavioral assessments that assess communication in other populations (e.g., ASD/DD; VB-MAPP) require significant adaptations for use with individuals with Rett syndrome. Further, given the unique characteristics of the syndrome, dynamic assessments (Haywood & Lidz, 2006), or those that are interactive and conducted at multiple points throughout time, should also be evaluated (Townend et al., 2020). As such, additional research on adaptations of communication assessments for individuals with Rett syndrome, are sorely needed (Ward et al., 2021). Direct observation assessments of current communication levels would allow for more individualized communication intervention and have implications on the approach as well as language practitioners use while teaching expressive communication. For example, an individual with Rett syndrome with high levels of receptive communication may find the mand training techniques used in the current investigation demeaning because mand training is typically conducted with and geared toward early learners. Further, knowing an individual has strong receptive language skills (even with relatively limited expressive language) allows for modification of the teaching procedures for expressive language to be more efficient (e.g., providing vocal instructions on how to use a device versus using gestural prompts). Future research on language assessments is an important next step in advancing communication research with individuals with Rett syndrome.

Given the limited expressive communication of many individuals with Rett syndrome, formal consent or assent to participate in research is difficult to obtain. We attempted to address this by collaboratively determining assent and assent withdrawal behaviors for each participant and implementing specific procedures when assent withdrawal behaviors occurred (e.g., terminate the session, revisit procedures or goals, etc.); however, we did so informally without explicitly tracking assent or assent withdrawal behaviors. Future research should formalize procedures for tracking and responding to assent withdrawal behaviors when formal assent and assent withdrawal directly from participants is not possible. This might include determining operational definitions for assent and assent withdrawal behaviors collaboratively with caregivers at the outset of the research relationship and re-evaluating those behaviors to determine if any changes to definitions are needed throughout the study. Researchers could also create data collection systems to track when assent withdrawal behavior occurs and what changes were made in response to them. Identifying assent withdrawal behaviors and designing explicit procedures in response to observation of those behaviors during instructional sessions is an important addition to our research procedures to ensure we are treating our participants with dignity and respect. This is important for any population who may not have the communication skills to verbally withdrawal assent.

The results of the current study also add to the existing literature (e.g., Akemoglu et al., 2020; Unholz-Bowden et al., 2020) supporting the use of parent training/coaching via telehealth as a viable service delivery model. This delivery model is particularly important for individuals with Rett syndrome for whom specialized care is needed, and access to highly trained professionals and clinical services is limited (International Rett Syndrome Foundation, 2022). In the United States, there are 19 clinics that specialize in services for individuals with Rett syndrome that serve approximately 15,000 individuals (Rett Syndrome Research Trust, 2022; International Rett Syndrome Foundation, 2022). Further, these specialty clinics are in large urban centers with most states limited to one throughout the entire state and many states without a clinic at all. This makes traveling to physical sites to access care difficult if not impossible for individuals who live outside those areas. Our results add to the literature that demonstrates telehealth as a feasible delivery model allowing for expansion of services to more individuals and their families.

All sessions in the current study were conducted via telehealth and caregivers obtained high levels of procedural fidelity in implementation of intervention procedures. Using telehealth as a service-delivery mechanism for communication interventions in this study came with several benefits: (1) We were able to reach and work with families who we would otherwise be unable to provide services for due to geographical barriers, (2) parents were provided with systematic instruction as part of the coaching for implementing the communication intervention, and are therefore more equipped to continue teaching communication skills to their children on their own, and (3) through running the communication intervention in participant’s homes, the girls were able to acquire communication skills within a more natural context, which may help facilitate long-term maintenance of those skills and generalizing to other natural settings. A couple of barriers noted were (1) the potential for parents to be distracted within their home, such as by other family members in need of attention, and (2) the potential for families to experience technology difficulties resulting in a delay of intervention sessions. However, with these barriers came the opportunity for us as researchers and therapists to become more flexible, adaptable, and naturalistic in how we coached parents to implement intervention procedures, as well as more skilled in supporting parents if they struggled with technology or running interventions with distractions. Noting the anecdotal nature of these reports, identifying effective methods for delivering interventions via telehealth that maximizes the benefits and minimizes the barriers continues to be an important area for future research with this population.

Some limitations regarding the study need to be noted. First, the third participant (Christina) acquired the third communication response without explicit teaching, limiting the experimental control of that response within the multiple probe design. Although a limitation to experimental control, it appears acquisition of the third response may have been due to generalization of acquisition of the other two communication responses. Generalization is plausible because baseline levels of responding remained low until after acquisition of the first response and teaching began on the second. Acquisition of new responses through generalization (without explicit teaching) is a goal of communication intervention, allows for increased efficiency, and should continue to be embedded in future Rett syndrome communication research. Given generalization probes were not conducted in baseline, we cannot know for sure if generalization occurred. Researchers did not formally program for, or assess the generalization (e.g., to other settings, novel communication partners, etc.) of the communication intervention and future research should more explicitly examine this effect. In addition, the limitation of the design when generalization across responses is likely may be addressed by future research by using other research designs or included embedded reversals as was done with Blythe. Second, the sample of individuals who participated in the study is small, and all three of them were white, potentially limiting our ability to generalize these findings to more diverse populations with Rett syndrome. Third, the communication responses targeted in this study required the participant to engage in only a simple discrimination, and did not include discrimination between responses, or more complex expressive communication targets necessary to develop an extensive repertoire of expressive language. Whether similar explicit teaching procedures would be successful in teaching individuals with Rett syndrome more complex responses (e.g., discrimination, device navigation, abstract communication messages like feelings, etc.) should be evaluated in future research. The results of research that used explicit teaching procedures to develop complex communication with other populations with similar language deficits (e.g., autistic individuals; Bak, et al., 2021; Jennings et al., 2021) provide some evidence that application with individuals with Rett syndrome may be successful. Similarly, the procedures used to teach mands via a microswitch are relatively simple, so more research is needed to evaluate the feasibility of parent coaching via telehealth when parents are required to implement more complex procedures. Finally, researchers evaluated long-term maintenance of communication responses for only two participants (Cinna and Blythe), and maintenance beyond 4 months was not assessed. Future research should examine the durability and maintenance of communication responses for all participants and across longer periods of time (e.g., 1 year). One way future research might examine the durability of communication is to examine repair strategies used when an initial communication attempt is not successful.

The current study provides additional evidence of the effectiveness of mand training and explicit instruction procedures for parents who are being coached via telehealth technology to teach their children with Rett syndrome to use AAC to make requests and protest undesired activities. The evidence base is currently small, and we have barely scratched the surface in terms of establishing a strong empirical basis for providing effective and accessible expressive communication instruction to individuals with Rett syndrome. Continued research is needed so all individuals with CCN, including those with Rett syndrome, can lead full and self-determined lives where their wants and needs can be communicated and therefore met.

Data Availability

Data are available upon written request from first author.