The prevalence of autism spectrum disorder (ASD) in the U.S. has increased 114% in the past decade, from 1 in 150 (Centers for Disease Control and Prevention, CDC 2007) to 1 in 68 children diagnosed with the disorder (CDC 2014). ASD is a developmental disability characterized by difficulties in social-communication skills and engagement in repetitive and restrictive behaviors and interests (RRBI). It is a lifelong disorder for which there is no cure. The increasing prevalence of this disorder has provoked widespread public concern and led to issues related to treatment of individuals with ASD (McDonald et al. 2012). In particular, the increasing prevalence of ASD has generated a gap between available resources and consumer demand (Wainer and Ingersoll 2013). Long waitlists and a shortage of trained professionals result in delayed or denied services with drastic impacts on the well-being of individuals with ASD (Machalicek et al. 2016; Meadan et al. 2016).

This service–need gap is a continual problem in health care and education, and various delivery models have been researched to address this gap (Nelson and Palsbo 2006). For decades, the medical field has investigated the use of telemedicine as a means of extending the reach of health care providers (Augestad and Lindsetmo 2009). Telemedicine is defined as the use of telecommunication and online technologies to provide health care at a distance. Following the success of telemedicine, the field of telehealth emerged to expand the use of distance technology beyond medicine to the dissemination of other services, such as counseling and psychotherapy (Elford et al. 2000). The field of education has also begun to leverage distance technology to engage in telepractice. Telepractice involves the use of online instruction and videoconferencing to connect interventionists (e.g., parents, educators, therapists) with an expert at a distance (Symon 2001). In the past 10 years, researchers have begun to utilize telepractice to coach individuals to implement effective assessment and intervention practices for persons with ASD (Boisvert et al. 2010). A review by Boisvert and colleagues in 2010 identified eight studies that utilized telepractice to deliver services to individuals with ASD. Of those eight, five utilized telepractice to coach interventionists while implementing assessments and interventions for individuals with ASD. Since that review, a number of studies were published that further the literature base and investigate the use of telepractice to teach a varied population of interventionists including parents of individuals with ASD (e.g., Heitzman-Powell et al. 2014; McDuffie et al. 2013), behavioral therapists, (e.g., Wainer and Ingersoll 2013), and educators of students with ASD (e.g., Hay-Hansson and Eldevik 2013). There is now converging literature identifying the telepractice delivery model as not only effective, but cost efficient (Lindgren et al. 2015), time efficient (Wacker et al. 2013a), and resource efficient (Lindgren et al. 2015; Wacker et al. 2013a).

In a broader view, telepractice is emerging as a promising delivery model to connect autism experts with consumers needing services. However, on an individual level, the outcomes for the consumers and individuals with ASD are mixed. For example, Meadan et al. (2016) examined the effects of telepractice training and coaching on a parent-implemented communication intervention. The authors found that parent skill varied after the first post-training phase and performance feedback was necessary for some parents to reach desired fidelity outcomes. In addition, the impact on child outcomes has been mixed. For example, Wainer and Ingersoll (2014) trained five mothers of children with autism to implement reciprocal imitation training (RIT) using a self-directed telepractice training program followed by individual coaching and follow-up. Although four out of the five mothers reached criterion in implementing RIT during individual coaching, there were mixed impacts on the children with three of the five children showing no effects.

There are several factors that might mediate parent-, educator-, or therapist-directed treatment outcomes. Two major variables are intervention fidelity and implementation fidelity (Barton and Fettig 2013). Implementation fidelity refers to the training procedures used to coach interventionists in implementing the procedures. Intervention fidelity refers to the fidelity of procedures as implemented by the interventionist. Intervention fidelity and implementation fidelity are both essential in ensuring positive outcomes for the individual with ASD (Barton and Fettig 2013). To further the current literature base, this review aims to synthesize fidelity features in telepractice studies focused on parent-, educator-, and therapist-directed (e.g., “interventionists”) treatments for individuals with ASD. In particular, this review aims to capture the telepractice delivery models utilized and subsequent impacts on intervention fidelity (Barton and Fettig 2013). A review of this nature is intended to offer future directions for researchers and offer direction to practioners interested in utilizing telepractice technologies to connect high quality ABA coaches with ASD interventionists.

Method

Literature Search Procedures

A systematic search was conducted in the following online databases: ERIC (EBSCO), Medline Complete, Academic Search Complete, Psychology and Behavioral Sciences Collection, and PsycINFO. Publication year was not restricted, but results were limited to peer-reviewed research. Search terms to describe individuals with an ASD were combined with terms to describe telepractice. The terms for individuals with an ASD included “Asperger,” “autis*,” “developmental disab*,” “ASD,” and “PDD-NOS.” The search terms to describe telepractice included “telehealth,” “telepractice,” “videoconferenc*,” “telemedicine,” “distance train*,” “distance education,” and “teleconference”. This initial search was conducted in October 2014, updated in January 2017, and identified 329 studies once duplicates were removed. Following the database search, an extended search of the reference lists of included articles was conducted and a forward search was conducted by searching the “cited by” function in Google Scholar. Additionally, a hand search of the references from two relevant reviews (e.g., Boisvert et al. 2010; Meadan and Daczewitz 2015) was conducted. The extended searches identified an additional 63 articles for a final total of 392 articles. Figure 1 documents the search procedures and inclusion/exclusion process.

Fig. 1
figure 1

Search graphic

Full size image

Inclusion Criteria

To be included in this review, articles had to meet the following criteria: (a) include at least one participant with ASD (inclusive of ASD, pervasive developmental disability, or Asperger’s syndrome); (b) have at least one dependent variable focused on fidelity of implementation of an intervention or assessment procedure by an interventionist who was at a distance from the specialist/expert for the duration of the training; and (c) use a form of telepractice to connect an expert or specialist to an interventionist. For the purpose of this review, telepractice was defined as the application of communication technologies (i.e., online instruction, videoconferencing software, or computerized software; Boisvert et al. 2010; Nelson and Palsbo 2006) to connect an expert to an interventionist across some distance. Studies that combined in-situ instruction with telepractice instruction, or in which the effects of the telepractice instruction could not be isolated, were excluded (e.g., McDuffie et al. 2013). Also excluded were studies that did not report fidelity outcomes (e.g., fidelity, accuracy.) (e.g., Ingersoll and Berger 2015; Wacker et al. 2013b). A total of 19 articles were identified after applying the inclusion criteria to all identified articles.

Descriptive Synthesis

Each included study was summarized according to the following variables: (a) characteristics of the participant with ASD (i.e., age, diagnostic information, gender), (b) characteristics of the interventionist (i.e., relationship to participant with ASD [teacher, parent, etc.], age, gender, and previous experience with the target assessment or intervention), (c) dependent variables for the interventionist(s), (d) dependent variables for the participant(s) with ASD dependent variable(s), (e) telepractice delivery methods utilized (i.e., online module, videoconferencing), (f) description of the training procedures (e.g., video models, written instruction, verbal instruction), (g) duration of training, (h) outcomes for the interventionist(s), (i) outcomes for the individual(s) with ASD, (j) fidelity of independent variable implementation, (k) study design, (l) generalization, (m) maintenance, and (n) social validity.

Establishing Inter-Rater Reliability

Inclusion Criteria

A second independent rater reviewed 48% (n = 189) of the studies during the title/abstract search. The second rater read each title and abstract and rated them as “1” for potential inclusion in the review or “0” for articles that did not meet criteria for inclusion in this review. Resulting inter-rater reliability (IRR) was calculated as the number of agreements divided by the sum of the agreements plus disagreements and multiplied by 100 to obtain a percent. The resulting IRR was 93% for the title/abstract review. Following the title/abstract review, a comprehensive list of articles was developed for a total of 52 articles resulting from the initial search and 63 from the extended search.

The 115 articles resulting from the initial title/abstract review were systematically rated for potential inclusion in this review. IRR was established for 73% of the articles (n = 84). Each study was reviewed based on inclusion criteria and assigned a rating of “1” (meets criterion) or “0” (does not meet criterion). IRR was calculated using a percent agreement measure by dividing the total agreements by the total sum of items reviewed and multiplying by 100. The agreement for whether or not to include an article was 99%. Following the calculation of IRR, the two raters reviewed the discrepancies and came to a collaborative consensus for a final IRR of 100%.

Descriptive Synthesis

A second independent rater coded 63% of articles (n = 12) for a measure of IRR. There were a total of 168 opportunities to establish agreement (i.e., 12 articles with 14 variables). IRR was calculated by dividing the total number of agreements by the sum of the agreements plus disagreements and multiplying by 100 to obtain a percentage. There were 14 disagreements for a total IRR of 92%. Upon instances of disagreements, the raters reviewed and came to a collaborative decision for a final IRR of 100% on the extracted data.

Results

Descriptive Review

Ten journals published the 19 articles included in this review. The highest concentration (n = 5) was published in the Journal of Autism and Developmental Disabilities. Publication dates ranged from 2009 to 2016. The resulting study summaries are presented in Table 1.

Table 1 Descriptive synthesis of included studies
Full size table

Participant Characteristics

A total of 155 interventionists participated across the 19 studies. Five of the 19 studies taught teachers of individuals with ASD (26%), three taught therapists of children with autism (16%), and 10 included parents of a child diagnosed with ASD (53%). One study taught both therapists and teachers of individuals with ASD (5%).

Fourteen of the studies reported the gender of their participants, with 114 of the 155 interventionists being female (74%) and 16 of the 155 interventionists being male (10%). Five of the studies reported ages for their interventionists with an average age of 32 years (range 20–47 years). Fifteen out of the 19 studies (79%) reported whether the interventionists had prior knowledge on the targeted skills prior to their study. Three of the studies provided descriptive data by stating the participants had no prior experience. Eight provided results from a skill assessment (i.e., pre-test or baseline performance data) prior to the introduction of the training program. Four provided both descriptive data and assessment data regarding interventionists’ prior knowledge.

Sixteen of the studies (84%) also included a total of 128 individuals with ASD as participants. Eleven of the studies reported the gender for 75 of the 128 (60%) participants with ASD with 65 male and 10 female participants. Twelve of the studies reported the age of their participants with ASD (86%). The average reported age of the participants was 3.83 years (range 1.3–16 years).

Dependent Variables

Across the 19 studies, four prepared interventionists to implement assessments, 13 studies focused on behavioral or teaching interventions, and two conducted an assessment with a follow-up intervention. Of the six studies that included assessments, two taught interventionists to conduct a preference assessment, and four taught interventionists to conduct a functional analysis (FA) of challenging behavior. A total of six different strategies were taught across the nine studies focused on behavioral intervention and teaching strategies: functional communication training (FCT), discrete trial teaching (DTT), Early Start Denver Model (ESDM), reciprocal imitation training (RIT), incidental teaching (IT), and other behavioral teaching strategies (e.g., prompting, shaping, reinforcement procedures).

Fourteen of the 19 studies (74%) reported outcomes for participants with ASD. Over half of the 14 studies (n = 9; 64%) focused on social communication behaviors (e.g., spontaneous verbalizations, prompted verbalizations, joint attention). Four of the 14 studies (29%) collected data on the participants’ challenging behavior (e.g., elopement, aggression, property destruction), and one study (7%) reported outcomes of preference assessments.

Telepractice Delivery Method

The 19 studies used a combination of four different delivery methods for their training programs: online modules, videoconferencing, online modules with videoconferencing, and DVD with videoconferencing. Half of the studies (n = 9) used videoconferencing only to prepare interventionists. The other seven studies were split between online modules (n = 1), videoconferencing with online modules (n = 7) and DVDs with videoconferencing (n = 2).

Description of the Training Program

The procedures used to teach interventionists varied across the 19 studies in regard to duration and instructional elements. Although all of the studies utilized telepractice to deliver the instruction, 17 of the studies provided one-on-one instruction (89%), and two provided group instruction (11%). The number of instructional sessions varied between studies, with 18 of the studies reporting the total duration of the training program. Reported instructional times ranged from 40 mins to 44 h. Most of the programs included more than one session (n = 18, 95%). Although all of the programs included some form of didactic instruction, nine studies delivered the didactic instruction via videoconferencing (43%), eight used online modules/website (42%), and two provided DVDs to participants prior to videoconferencing. For example, Wacker et al. (2013a, 2013b) met with their interventionists for a 1 h pre-assessment meeting, prior to coaching, in order to conduct an FA. During this pre-assessment meeting, the interventionists were provided verbal and written instruction regarding behavioral assessment rationale and procedures. In contrast, Heitzman-Powell et al. (2014) had the interventionists complete an online module prior to meeting with the trainer via videoconferencing.

Instructional elements included a combination of (a) verbal instruction, (b) written instruction, (c) modeling, (d) role-play, (e) performance feedback, (f) question and answer, (g) video examples, and (h) interactive learning activities (e.g., assessing others’ ability to implement RIT). Seventeen of the studies used verbal instruction (89%) and 18 incorporated written instruction (95%). Verbal instruction typically included the rationale of the intervention or assessment, introduction to the components of the intervention, or prompting during the implementation of the assessment or intervention. Written instructions included instructions outlining the implementation of the assessment or intervention (e.g., checklists, step-by-step instructions) and instruction on the rationale and support for the intervention or assessment.

Also commonly used by a majority of studies was performance feedback (n = 17; 89%). Across the 17 studies, 15 of the studies provided immediate one-on-one targeted performance feedback to the interventionists after viewing a live demonstration of the skill with a child participant. The other studies relied on delayed feedback for at least one of the participants (Wainer and Ingersoll 2014) or delayed video-based feedback (Neely et al. 2016). One study provided performance feedback in a group setting rather than one-on-one. Less commonly used instructional elements included modeling (n = 5, 26%), role-play (n = 3, 16%), interactive learning activities (n = 6, 32%), built–in question and answer opportunities (n = 8, 42%), and video examples (n = 9, 47%).

Duration of Training

Duration of training was reported for a majority of the included studies, however, the completeness of the information varied. For studies that included an online module or website prior to video conferencing, duration of training via the online module was reported in eight of the 10 studies (80%). The duration of engagement with online content ranged from 0.67–16 h. All of the studies using videoconferencing reported the duration of videoconferencing sessions. Durations ranged from 45 mins–21 h. One study reported a total duration of training of 40 h (Fisher et al. 2014) but did not differentiate between time to complete online module and time spent videoconferencing.

Fidelity of Training Program

Ten of the 19 studies (53%) collected implementation fidelity for their training program. In four of the studies, the coaches were trained to criterion on the training procedures prior to the intervention, and fidelity data were collected throughout the study (Heitzman-Powell et al. 2014; Vismara et al. 2012; Vismara et al. 2013; Vismara et al. 2016). During the studies conducted by Machalicek et al. (2010) and Neely et al. (2016), the first author implemented the intervention and a second rater evaluated her adherence to performance feedback procedures throughout the study. For Meadan et al. (2016), doctoral students served as the coaches and a second member of the team evaluated their fidelity of coaching. For those utilizing online modules, studies reported adherence data through reported duration to complete online module content (e.g., Ingersoll et al. 2016; Vismara et al. 2013), access of online module content (e.g., Vismara et al. 2013; Vismara et al. 2016), or verification of completion of online module content through a knowledge assessments or interactive activities (e.g., Heitzman-Powell et al. 2014; Neely et al. 2016).

Outcomes for the Interventionists

All of the studies reported interventionists were able to implement the assessment or intervention with increased fidelity following the training program. Twelve of the studies established a pre-set performance criterion for their interventionist. Pre-set performance criteria ranged from 80%–90% intervention fidelity (e.g., as measured by a fidelity rubric; Neely et al. 2016). Six of the studies reported that all of the interventionists met the performance criterion (50%). Six studies reported that some interventionists did not meet the performance criteria. For example, Vismara et al. (2016) reported that only five of the 14 parents enrolled in the treatment group met fidelity criterion at post-test. During follow-up, only nine of the 14 parents met fidelity. Similarly, in Vismara et al. (2013) and Vismara et al. (2009), only some of their participants met the fidelity thresholds following coaching. For Wainer and Ingersoll (2014), there were mixed results for parent intervention fidelity following the online module learning phase, but four of the five parents reached criterion with the videoconferencing feedback phase. Similarly, Meadan et al. (2016) saw mixed results for their parents following the initial didactic training phase, but all parents met criteria following the videoconferencing feedback phase. Finally, Alnemary et al. (2015) provided training in a group format and indicated that, although all educators improved their fidelity of implementing FAs, only one met the performance criteria.

Outcomes for the Individual with ASD

Although a majority of the studies reported that data were collected on outcomes for participants with ASD (n = 15; 79%), outcomes in two studies were either not reported or could not be isolated for the participants with ASD.

For example, Vismara et al. (2009) assigned 10 interventionists to two groups and compared the effects of a training program delivered through telepractice versus on-site. Although they collected data for the participants with ASD, the results were aggregated for the two groups and reported results were combined. Therefore, the results for the telepractice group could not be isolated.

Of the thirteen studies where the outcomes for the participants with ASD could be isolated, nine of the studies reported improvements in the targeted behaviors for all participants (69%), and two reported clear assessment outcomes (15%). Two studies reported mixed results with some participants demonstrating improvements and some maintaining pre-intervention levels (8%) or participants demonstrating improvements but no significant effects (8%). One study reported clear assessment outcomes for 18 of their 20 participants in terms of clear functions of their challenging behavior (8%).

Experimental Design

Although all 19 articles reported outcomes for interventionists’ regarding their implementation of an assessment or intervention, only 15 employed an experimental design to systematically manipulate the independent variable (i.e., the training program). Of the 15 studies, 40% (n = 6) used group design methodology and 60% (n = 9) used single-case design. Two studies employed a pre-experimental non-randomized pre/post design, and four utilized a randomized group assignment design with pre/post analysis. For the nine studies utilizing single-case methodology, the majority (n = 8; 89%) employed a multiple-baseline design across interventionists. The remaining study utilized a multi-element design without a baseline phase to evaluate parent’s implementation of FCT during sessions coached via videoconferencing versus sessions implemented independent of coaching (Suess et al. 2014).

Maintenance and Generalization

Nine of the 19 studies (47%) collected maintenance data on interventionists’ implementation fidelity. Following the conclusion of the intervention, follow-up probes ranged from 1-week to 4-months. Six of the studies reported skills maintained above baseline levels at a 6-week follow up for ESDM, 2-month follow-up for DTT, 3-month follow-up for the ESDM, and at 2-, 3- and 4-month follow-up for naturalistic communication interventions. Three of the studies reported mixed results, with some interventionists returning to baseline levels for conducting an FA at 1- to 3-week follow-ups and implementing RIT at 1- to 3-month follow-ups, and some interventionists returning to baseline for social-communication skills (Meadan et al. 2016).

Three studies evaluated the generalization of the interventionists’ skills. Hay-Hansson and Eldevik (2013) provided descriptive information indicating two of their three interventionists generalized DTT skills to the new child. Meadan et al. (2016) reported positive results for implementation of skills when no coaching was provided. Alnemary et al. (2015) found one of their four special educators met the preset criterion and collected generalization data with a targeted student. Results indicate that he generalized acquired assessment skills to the targeted student.

Social Validity

Social validity of the training programs was reported for 11 of the 19 studies. Ten of these studies (91%) utilized a Likert-type questionnaire. Six studies used open-ended questions either in addition to a Likert-type questionnaire or as the primary means of evaluating social validity of the telepractice program. Results were positive across the studies with high acceptability for online modules and videoconferencing delivery methods. One study reported mixed results (Alnemary et al. 2015) with some interventionists indicating issues with technical difficulties. In addition, Vismara et al. (2009) reported results of a social validity questionnaire in terms of variability between the groups assigned to the on-site training program versus the telepractice program. They found no difference in the satisfaction between the two groups. Vismara et al. (2016) reported validity for both the therapist-assisted and community based group with higher satisfaction noted for the therapist-assisted group. Of note, the community-based parents indicated in the open comments a need for additional videoconferencing time with the coach.

Responses to the open–ended questions found that interventionists found the video examples to be most helpful for learning the targeted intervention. Interventionists also identified performance feedback as a highly useful training procedure. Vismara et al. (2012) also found that, although interventionists were initially concerned about the level of support available through telepractice, by the end of the study, all of the interventionists reported that telepractice was as informative and valuable as face-to-face delivery methods. Although interventionists in Wainer and Ingersoll (2014) and Vismara et al. (2009) did indicate that there were some technology issues throughout the studies, they reported these issues were easily remedied. Overall, interventionists from Vismara et al. (2009) and Vismara et al. (2012) indicated that they would recommend telepractice approach to other parents of children with ASD.

Discussion

This review synthesized 19 studies identifying the impacts of telepractice training programs on interventionists’ treatment fidelity for individuals with ASD. The 19 telepractice studies delivered training programs to 155 interventionists. A variety of assessments and interventions were taught including preference assessment, FA, FCT, DTT, incidental teaching, RIT, and the ESDM. Training procedures also varied in both duration and elements include verbal and written instruction, modeling, role-play, performance feedback, question and answer, video examples, and interactive learning activities. Training was delivered via online modules, videoconferencing, and DVDs. All of the studies reported interventionists were able to implement the assessment or intervention with increased fidelity following the training program. However, results were mixed in four studies as some participants needed additional training beyond the intervention to meet preset performance standards. In addition, coaches’ adherence to training procedures was not widely reported. Overall, this literature base is best described as limited due to the small number of studies, variability in training procedures, and mixed results for some participants.

The first aim of this review was to synthesize the training programs and subsequent impacts on intervention treatment fidelity. The most common training procedure involved verbal and written instruction and performance feedback. Previous research supports the use of this training package as an effective means of training ASD interventionists (e.g., Ward-Horner and Sturmey 2012). Of the 19 studies included in this review, only two (Wacker et al. 2013a, 2013b; Wainer and Ingersoll 2013) did not specify the use of performance feedback as a component of their intervention. These results suggest performance feedback may be an active element necessary for effective training of ASD interventionists via telepractice. These results confirm previous conclusions linking performance feedback to improved fidelity for interventionists working with individuals with disabilities (e.g., Brock and Carter 2016). However, the effect of the individual training elements cannot be isolated due to a variety of procedures. Isolating effective training elements is essential to improving usability and effectiveness of telepractice training programs (Ingersoll and Berger 2015). Future research might address this issue by conducting component analyses to isolate the active training elements.

Overall, there were positive results for the impact of the training programs on interventionist intervention fidelity. However, there were a couple of notable mixed results. First, the two studies that employed group training via videoconferencing reported mixed results (Alnemary et al. 2015; Vismara et al. 2009). Although these results could be mediated by the complexity of the skill (e.g., FAs were taught in Alnemary et al. 2015), these results challenge the validity of group training via videoconferencing. Future research might investigate ways to increase the effectives of group trainings as group trainings may be more economically feasible for resource constrained settings (e.g., schools). For example, schools may consider a workshop with follow-up feedback in the classroom (Brock and Carter 2016). Second, mixed results were noted for the effects of online modules on intervention fidelity (e.g., Wainer and Ingersoll 2014). In a couple of studies, content delivered via online modules did not equate to improved intervention fidelity until performance feedback was provided (e.g., Meadan et al. 2016). These results further support previous conclusions identifying performance feedback as an essential training component (Brock and Carter 2016; Rispoli et al. 2011).

The second aim of this review was to synthesize the reporting of implementation fidelity as potential moderator of subsequent intervention fidelity. Implementation fidelity was collected through reported duration of the training program and coaching fidelity. A strength of this literature base is a majority of the studies reported the duration of the training program. Although durations varied with the complexity of the skills being taught, the median duration of training was approximately 5.5 h for online modules and 8.5 h for videoconferencing. Although it has been shown that training duration does not necessarily impact training effectiveness (Brock and Carter 2016), training duration can potentially impact acceptance of procedures and subsequent intervention fidelity (Ingersoll and Berger 2015). Future studies should report access statistics when using a telepractice delivery model.

A major weakness of this literature base is the reporting of implementation fidelity. Only 10 of the 19 studies reported fidelity of the coaching procedures. For those using videoconferencing, procedures included a second rater observing the videoconference or a video-taped version of the videoconference and rating coach adherence to the coaching procedures. For online modules, studies either reported engagement through duration spent with the material or completion of a knowledge assessment. As intervention fidelity can moderate intervention fidelity and subsequent outcomes for individuals with ASD, reporting of implementation fidelity is vital. With less than half of the studies reporting implementation fidelity, the evidence supporting use of the telepractice delivery model as a means to facilitate high-fidelity intervention is limited.

Although all of the studies provided some demographic information regarding the interventionists, the descriptions were limited and typically did not include the age and gender of the participants, previous educational experience, or their previous experience with the skill being taught. As participant characteristics and previous experiences are likely to affect the success of the training, these descriptions are necessary to identify the populations for which the effects might generalize (Wainer and Ingersoll 2014). Interventionists’ previous experiences may also correlate to intervention fidelity and the ease with which they acquire the skills (Vismara et al. 2013). Future researchers should provide comprehensive descriptions of interventionist participants to promote the external validity of this literature base and to serve as potential moderators of training effectiveness.

Limitations and Directions for Future Research

Although the literature base has advanced within the past 5 years, there are a number of limitations that may serve as suggestions for future research. First, a majority of the studies provided training using one-on-one instruction. Of the studies that used group training, there were mixed results suggesting that one-on-one instruction may be an essential component. However, this dependence on a specialist to deliver individualized training may delay interventionists’ access to quality training programs and leave interventionists susceptible to controversial or ineffective treatments. Future research might investigate ways to maximize small group instruction to promote effectiveness of the instruction.

Second, although results suggest that telepractice did lead to initial acquisition of skills, the results were mixed for the maintenance and generalization of the skills. Of the studies that collected maintenance data, 33% reported that the skills did not maintain (e.g., Machalicek et al. 2010; Meadan et al. 2016; Wainer and Ingersoll 2014). In addition, only three studies reported generalization data with one report of descriptive data indicating that the interventionists generalized skills across their students (Hay-Hansson and Eldevik 2013), one reported positive results for implementation of skills without coaching support (Meadan et al. 2016), and one reporting generalization of skills for only one out of four interventionists (Alnemary et al. 2015). Although some skills may not be intended to generalize or maintain (e.g., parents conducing FAs; Wacker et al. 2013a, b), some assessments (e.g., preference assessments) or interventions (e.g., incidental teaching), would be intended to be used without ongoing coaching or support. However, the results from these studies do not indicate sustained behavior change. Fortunately, telepractice service models may allow for repeated follow-up assessments and coaching. In addition, telepractice can allow for more opportunity for coaches to follow-up with interventionists in the natural environment in which they will use the skills to help facilitate maintenance of skills (e.g., home for parent or school for educator). However, the current literature base is limited and future research should investigate ways to maintain and generalize skills learned via telepractice technologies

Although this review highlights the potential use of telepractice to facilitate early intervention, the fact that a majority of the participants with ASD were preschool or elementary age limits the generalizability of the outcomes. Although one study did include adolescents and teenagers with ASD (Machalicek et al. 2016), the literature is limited, and it is not possible to conclude that similar results would be obtained for interventionists working with adolescents or adults with ASD. Therefore, future research might consider replicating or extending the previous research to include interventionists working outside of early childhood.

Finally, this review was limited to the information provided by the authors of the included articles within the confines of the published article. It is possible that the authors may not have provided all the information pertinent to the intervention. For example, some articles provided detailed descriptions of the participants with ASD and limited descriptions of the interventionists. These omissions may be due to publication restrictions or the availability of the data. Therefore, the conclusions of this review are constrained by the information provided.

Implications for Practice

This review demonstrates promise for the use of telepractice technology in practice. Overall, the use of telepractice to facilitate parent-, educator-, and therapist-directed intervention fidelity was linked with positive outcomes for the interventionists. In addition, improved behaviors were also noted for a majority of the participants with ASD. Therefore, the preliminary results suggest that telepractice may be an effective means of preparing ASD interventionists.

Of note, of the studies that assessed outcomes for the individuals with ASD, all included preschool and elementary aged children. These results are encouraging as previous research identifies that early intervention is correlated to improved functioning for individuals with ASD (Kuppens and Onghena 2012; Makrygianni and Reed 2010; National Research Council 2001). Telepractice may facilitate early intervention by allowing specialists to serve those populations who were previously inaccessible due to the barriers of distance, time, and money. However, practioners should exercise caution when using telepractice to prepare interventionists working with adolescents or adults with ASD.

With respect to the training components used, practioners may consider the use of didactic instruction including verbal/written instruction and individualized performance feedback. The combination of these training components appeared in the majority of the studies and was linked to increased intervention fidelity for the interventionists. In particular, individualized performance feedback may be an active element to effective training programs and verbal/written instruction may be necessary for more complex skills.

Although this literature base demonstrates promise, it is still developing. Practitioners should take care to evaluate the effectiveness of their program through continued progress monitoring. Practioners should also plan to embed some form of planned generalization (Gianoumis and Sturmey 2012; Stokes and Baer 1977) and to provide ongoing support to ensure adherence to intervention procedures. In addition, although the current literature base did provide some preliminary support for the use of telepractice, there is a need to advance the research in this area. Therefore, practioners should continue to rely on face-to-face training when feasible and supplement with telepractice where necessary.

Conclusion

The use of telepractice as a service delivery model has promise. However, there is still great variability in the training components employed with subsequent varying effects on intervention fidelity. In addition, there was limited focus on implementation fidelity in the existing literature base. Therefore, telepractice as a delivery model requires additional investigation before qualifying as an evidence-based delivery model.