Scoping Review on Play-Based Interventions in Autism Spectrum Disorder

Play as a therapeutic strategy can help to improve daily functioning in children and adolescents with autism spectrum disorder (ASD). Play-based intervention can thus be an optimal option for treatment of this population. Our aim was to describe play-based interventions used in children and adolescents with ASD. We conducted a scoping review. A peer-reviewed literature search was conducted on PubMed, Scopus, EMBASE, Web of Science and PsycINFO databases. We included experimental studies which analyzed play-based interventions in children and adolescents with ASD, which were published in English/Spanish with full text available. We used three tables elaborated a priori to perform data extraction by two authors. Fifteen studies were included, mostly conducted in Australia and published during the past 10 years. Play-based intervention was categorized into three groups: new technologies, symbolic play or official techniques. Interventions lasted approximately 11 weeks, contained one weekly session of 30–60 min and were delivered by parents and teachers. Play-based interventions using new technologies were the most used. Intervention duration and number of sessions varied between articles. Further studies are needed to create play-based intervention protocols which can be implemented in clinical practice with children and adolescents with ASD, thus promoting evidence-based interventions in this field.


Introduction
Autism spectrum disorder (ASD) is a group of neurodevelopmental disorders characterized by restrictions in social interaction, alterations in communication, a lack of cognitive flexibility, the presence of restrictive interests and stereotyped behaviors [1]. The cause of ASD is unknown, although epigenetic and environmental agents have been described as etiological factors [2]. These disorders are diagnosed four times more often in boys than girls [3], usually at nearly 4 years old [4]. The worldwide prevalence of ASD is approximately 1 in every 160 children and adolescents [5,6], although some studies have reported prevalence of up to 1 in every 138 [7,8].
ASD symptomatology affects the occupations of children and adolescents with these disorders, limiting their autonomy and independence [9]. Children and adolescents with ASD have difficulties initiating and maintaining interactions with peers, especially during play and social activities [10]. Play is the most important occupation for all children since it helps them to develop the necessary skills for the autonomous performance of daily life activities (ADL) [11]. However, the ability to play naturally is limited in children with ASD due to communication difficulties, alterations in social relationships and restricted interests [12]. This can alter their development, leading to disruption in the performance of #2 TITLE-ABS-KEY (("game-based" OR "play-based")) #3 TITLE-ABS-KEY ((asd OR autism OR autistic OR asperger OR rett OR pervasive OR disintegrative) AND ("game-based" OR "play-based")) #2 ("game-based" OR "play-based") #3 (ASD OR autism OR autistic OR asperger OR rett OR pervasive OR disintegrative) AND ("game-based" OR "play-based") PsycINFO #1 (ASD OR autism OR autistic OR asperger OR Rett OR pervasive OR disintegrative) #2 ("game-based" OR "play-based") #3 (ASD OR autism OR autistic OR asperger OR rett OR pervasive OR disintegrative) AND ("game-based" OR "play-based")

Inclusion and Exclusion Criteria
We established the following inclusion criteria: (a) experimental studies with at least 10 participants: randomized and non-randomized clinical trials, pilot studies, exploratory studies and quasi-experimental studies; (b) child or adolescent population (under 18 years old); (c) population diagnosed with ASD: Asperger's syndrome, Rett syndrome, disintegrative disorder, classic autistic disorder and pervasive developmental disorder; and (d) studies using play-based intervention. We also established the following exclusion criteria: (a) studies published in a language other than English and/or Spanish; and (b) studies with no full text available.
We did not apply any filters by time or study type in any of the databases consulted. All inclusion and exclusion criteria were applied manually.

Study Selection
The review and screening of the articles identified in the literature searches of the five databases was carried out using Microsoft Excel. We downloaded all article titles from each database onto an Excel sheet, from which duplicates were removed. Once duplicates were removed, we performed the screening of the remaining studies in three phases: by title, by abstract and by full text. One researcher (LCLN) created the database in Excel and removed duplicates, two researchers (LCLN and LMCG) performed all screenings independently, and a third researcher (LTC) resolved discrepancies between the other two researchers during the screening processes.

Data Extraction and Synthesis
In order to make data synthesis more objective, we created three tables for the information to be collected from each included article, before starting the data extraction. These tables were elaborated by consensus of all the researchers and were based on the Cochrane Handbook recommendations [22]. The first table included items related to the main characteristics of the studies: author/year, study design, sample/country, participants, intervention/comparator, assessment and main study variables [24]. The second table included items related to the specific characteristics of the play-based interventions: author/year, participants and diagnosis, intervention, intervention duration, number of sessions, intervention manager and main results. The last table included items related to the risk of bias of the included articles: author/year, main limitations, funding sources and declarations of interest [25]. Two researchers (LCLN and LMCG) were in charge of data extraction, and a third researcher (LTC) was responsible for resolving possible discrepancies during data extraction. We present a narrative description of the results in this scoping review and we have used tables and figures wherever possible.

Quality Assessment
We did not assess the quality of the included articles as it is not a mandatory process in scoping reviews [26][27][28]. However, as recommended in the Cochrane Handbook, we have included a table on the risk of bias in each of the included articles to alert readers to aspects closely related to low quality such as limitations, funding sources and declarations of interest [22]. In addition, we have added a subsection on results providing a summary of the main limitations reported in the included studies, which can also be an indicator of the quality of the studies.

Results
We obtained a total of 1202 articles from the literature searches conducted in the five databases. After eliminating duplicates, we screened 661 articles by title, then the remaining 571 were screened by abstract, leaving 181 which were screened by full text. Of these 181 articles, only 15 met the inclusion criteria and were included in this scoping review ( Figure 1).
Most of the control groups were characterized by not having received any intervention (n = 4) [32,38,39,42], or by having received either: the same intervention as the intervention group but with a different session duration (n = 1) [35], the same intervention as the intervention group but with different intervention duration (n = 1) [30] or a different intervention to the intervention group (n = 1) [34] (Table 2).

Study Variables and Measurement Instruments
Significantly, few articles (n = 3) used playing as a study outcome. In this sense, pretend play was assessed using the Child-Initiated Pretend Play Assessment (ChIPPA) [29], interactive peer play skills were assessed using the Penn Interactive Peer Play Scale (PIPPS) [29], playtime interaction was assessed by the Peer Interaction Paradigm (PIP) [35] and behavior during play was assessed by video analysis [40] (Table 2).
Behavior was the most studied outcome among the included studies (n = 7) ( Table 2). Usually, only one questionnaire or battery was used per article to assess behavior. The evaluation tools were: the Conners Comprehensive Behavior Rating Scales (CCBRS) [31,33], the Early Head Start (EHS) 24-Month 3-Bag Scales [36], the Functional Emotional Assessment Scale (FEAS) [37], the Eyberg Child Behavior Inventory-Parent (ECBI) [34] and the Achenbach System of Empirically Based Assessment (ASEBA) [40]. However, one included article [42] used the following five questionnaires to assess behavior: the Behavior Rating Inventory of Executive Function (BRIEF), the Conners Rating Scale-Short Version (CRS-3), the Behavioral and Emotional Rating Scale 2nd edition (BERS-2), the Social Skills Rating System (SSRS) and the Gilliam Autism Rating Scale 2nd edition (GARS-2).
Language was the second most studied outcome among the included studies (n = 5) ( Table 2). Four studies assessed language as a global function, using the following tools: The Preschool Language Scale 4th edition (PLS-4) [29], the Children's Communication Checklist 2nd edition (CCC-2) [31,32], the Expressive Vocabulary Test 2nd Edition (EVT-2) and the Elaborated Phrases and Sentences subtest of the Test for Auditory Comprehension of Language 4th Edition (TACL-4) [32,33]. Another four studies assessed specific aspects of language such as literacy skills using the SEMA-TIC [38], French reading skills using the Alouette Reading Test [38], pragmatic language using the Pragmatics Observational Measure (POM) [31,33], reception and expression of prosody using the Profiling Elements of Prosody in Speech Communication (PEPS-C) [31], language performance during social interaction using Pragmatics Observational Measure 2 (POM-2) and Social Emotional Evaluation (SEE) [32,33]. Another study assessed language development disorder using the Children's Communication Checklist 2nd edition (CCC-2) [33].
Other study outcomes on aspects such as attention [39], working memory [43], executive functions, stress [35] and gross motor skills [30] were assessed in isolation in some studies using a variety of questionnaires (Table 2).

Play-Based Interventions
The duration of interventions used in the included studies was 11 weeks, except for one that lasted 32-48 weeks [37] and one for which the duration is unknown [36] ( Table 2). The sessions usually took place once a week (n = 8) [31][32][33][34][35]38,40,41], although in some studies they were carried out twice [29,36], three [30,42] or five times a week [39]. In one article the number of sessions varied between two and five times a week [43], and in another article the number of sessions was not reported [37].
We classified the play-based interventions used in the included studies into three main groups: play-based interventions using new technologies, play-based interventions using symbolic play and play-based interventions using official techniques.  [36] 18, ASD and TD Two play sessions, each taking place in a distinctly different setting. The social-play-based setting consisted of common toys typically used in a natural or free-play setting such as toy cars or imaginative play with blocks. The motor behavior-based setting: situated in a large play space in addition to common equipment to encourage motor behavior, such as play on stairs or a tricycle. The sessions were recorded.      Peer-mediated, play-based pragmatic language intervention, which consisted of 30 min child-directed free play with therapist and peers, video feedback play and home training. The play in the clinic was recorded and the images formed the video-feedback for the following week. In each session mothers learn one technique at home, with feedback from therapist; then the therapist performs a demonstration of play with child; and mother interacts with child, with guidance from therapist. 12 60 min, 1 times/week OT, mothers Increase in ABC score (p < 0.001) and ATEC score (p < 0.001) in postvs. pre-intervention Increase in mother-child bond (p < 0.001) in post-vs. pre-intervention Decrease in mothers' disconnection (p = 0.001) and stress (p = 0.004) in post-vs. pre-intervention No difference in child self-efficacy (p = 0.318) in post-vs. pre-intervention
Three different computer games were used as an intervention strategy. Firstly, Serret et al. [38] used the SEMA-TIC which is a computerized game based on non-verbal cognitive skills and consists of ten sets of ten games and a dictionary in order to teach the player how to use a computer mouse. In this study, participants were aged from 6 to 11 years and they underwent 23 weeks of intervention with one weekly 4 h session. The intervention was conducted by speech therapists, psychologists, teachers and the children's parents. Secondly, Beaumont et al. [34] used the Secret Agents Society (SAS), a computer resource for people with communication difficulties with the aim of improving face-to-face communication by overcoming different "game missions", which facilitates training in social skills. In this study, participants were aged from 4 to 6 years and they underwent 10 weeks of intervention with one weekly 30 min session. The intervention was conducted by occupational therapists and the children's parents. Thirdly, Macoun et al. [42] developed and used the Caribbean Quest, a cognitive training program based on virtual games with the aim of improving attention and executive function skills through repetitive and hierarchical games based on neuroplasticity, using cognitive strategies to generalize the use of these abilities to the real world. In this study, participants were aged from 6 to 12 years and they underwent 8 weeks of intervention with three weekly 30 min sessions. The intervention was conducted by teachers.
Three other interventions were carried out using virtual games through digital devices such as the Xbox Kinect device (Microsoft, Redmond, WA, USA) and an eye-tracker. Edwards et al. [30] carried out an intervention using three sports video games through the Xbox Kinect device. This device has a sensor which allows full body movements with a free range of motion, and was therefore used to improve gross motor abilities and the control of objects. In this study, participants were aged from 6 to 10 years and they underwent 2 weeks of intervention with three weekly 45-60 min sessions. The intervention was conducted by the children's parents. The Xbox Kinect device was also used by Mairena et al. [40] but this time with the Pico's Adventure game, in which the child must help an alien that has just landed on Earth to address social initiation behaviors. In this study, participants were aged from 4 to 6 years and they underwent 4 weeks of intervention with one weekly 60 min session. The intervention was conducted by psychologists. The last devices used in the included studies was the Tobii I-Series, an eye-tracker, which is a device that registers eye movements and sends them to specific software to generate gaze data. Fabio et al. [39] used this device with the Sensory eye FX program, which consisted of a set of thirty applications to analyze attention and motivation during digital games play. In this study, participants were aged from 2 to 33 years and they underwent 16 weeks of intervention with five weekly 30 min sessions. The intervention was conducted by the researchers.
The smartphone was the last technology used in play-based interventions. Wagle et al. [43] used five different smartphone games for mobility management training which contained sensorimotor and behavioral components. In this study, participants were aged from 6 to 13 years and they underwent 4 weeks of intervention with 2 to 5 weekly 30 min sessions. The intervention was conducted by teachers.
Stagnitti et al. [29] carried out the group-based intervention Learn to Play Program, which consisted of transport, construction and household games, as well as caring for dolls, in order to develop play and simulation skills. In these games, the child was in charge of taking photographs while playing, in order to talk about them later and, in this way, work on language skills. In this study, participants were aged from 5 to 6 years and they underwent 24 weeks of intervention with two weekly 60 min sessions. The intervention was conducted by teachers, speech and language therapists and occupational therapists.
MacDonald et al. [36] carried out a play-based intervention in two different scenarios. The first was based on social play, in which cause-effect toys, construction, cars, miniatures and imaginative play were used. The second was based on motor abilities in which stairs, mats, slides, see-saws and balls were used in an open space. Authors instructed children to "play together as usual", and then they were recorded. In this study, participants were aged from 2 to 7 years and they underwent two days of intervention with one daily 10 min session in each scenario. The intervention was conducted by the researchers.
Parsons et al. carried out three studies in 2019 [31][32][33] in which a pragmatic language intervention based on play was delivered with different elements such as sandboxes, animal toys, dolls clothes, board and card games or building blocks. In this intervention, children were asked to play in pairs, putting children with ASD together with typically developing children. The sessions were recorded and the participants rated the game as "green" or "red" according to whether they thought it was favorable for both members of the pair or unfavorable for one or both, respectively. In these studies, participants were aged from 6 to 12 years and they underwent 10 weeks of intervention with one weekly 50-65 min session. The intervention was conducted by speech therapists, occupational therapists and parents.
The last study in which play-based intervention using symbolic play was used was a study carried out by Fu et al. [41]. They used a play-based communication and behavioral intervention including various elements such as cause-effect toys, stories, balls, card games, dolls or building blocks. In this intervention, each week occupational therapists taught a new technique to mothers of children with ASD to practice with their 1-2-year-old children. In this study, participants were aged from 1 to 2 years and they underwent 12 weeks of intervention with one weekly 60 min session. The intervention was conducted by occupational therapists and the children's mothers.

Play-Based Interventions Using Official Techniques
In two of the included articles, published in 2007 [37] and in 2017 [35], a play-based intervention using official or registered techniques was carried out (Table 3).
Corbett et al. [35] used a program called SENSE Theatre ®® which involved observing, interpreting and articulating thoughts and feelings through improvisation, role play, scripted interaction and acting. This program is aimed at improving social skills in children and adolescents with ASD through theatrical techniques delivered in a supportive and peer-mediated intervention. In this study, participants were aged from 8 to 14 years and they underwent 10 weeks of intervention with one weekly 4 h session. The intervention was conducted by clinical researchers.
Solomon et al. [37] carried out the PLAY Project Home Consultation (PPHC) program, which is a home-based program using the DIR ®® /Floortime approach. This home-based counseling program was created to improve parent-child interaction by training parents of children with ASD in skills and techniques to use during play. A professional visited the families each month to give advice on the play sessions, which were videotaped. In this study, participants were aged from 1.5 to 6 years and they underwent 32-48 weeks of intervention in which parents were encouraged to play 15 h per week with their children. The intervention was conducted by social workers, recreational therapists, home-based consultants and the parents of children with ASD [37].

Main Results of the Included Studies
All studies showed improvement in most of their outcomes (Table 3). However, three of the included studies reported negative results [33,36,40] such as less parental involve-ment during play in the group of children with ASD compared to typically developing children [36], more maladaptive behaviors during video game intervention (intervention group) compared to free play (control group) [40] and worse post-intervention scores on discourse coherence in children with ASD [33].
In nine of the included studies, no significant differences between groups or pre-post intervention were found [30,32,33,[35][36][37][41][42][43]. In one of these studies, there were no differences between the intervention and control groups in anxiety and cortisol levels [35]. In the remaining eight studies, there were no differences before and after intervention in different outcomes such as object control [30], child's negativity towards the parents during social play [36], effectiveness of parents' involvement with their children [37], language performance during social play [32], receptive syntax and expressive language [33], selfefficacy [41], divided and sustained attention [42] and working memory [43].

Discussion
Play-based interventions for children and adolescents with ASD were mainly characterized by the following characteristics: a duration of approximately 11 weeks, one weekly session of 30-60 min, conducted at school or at the participants' home, and were delivered by the parents of children with ASD or teachers. In addition, most play-based interventions included new technologies or symbolic play, and rarely included registered or official techniques.
Most of the articles included in this review have been published in the past ten years and the majority of studies have been carried out in Australia. Rice et al. attributes this increase in research during the past few years to several causes, such as increased awareness of ASD, changes in diagnostic criteria of ASD and the need to know the risk factors associated with this disorder in more depth [44]. The incidence of publications on ASD has increased in recent years, as has the incidence of the disorder itself, especially in underdeveloped and developing countries [6]. In this sense, one of the continents with the highest prevalence of ASD is Oceania, where 697 out of every 100,000 children receive this diagnosis [7]. This fact could explain why Australia was one of the countries in which most studies were carried out.
The participants in the included studies were mostly male and around 6 years old. ASD is known to be more prevalent in boys than in girls [45,46]. Duvekot et al. try to explain this fact, stating that boys tend to show more symptoms that are characteristic of ASD such as repetitive and restricted behaviors (e.g., playing with dinosaurs or means of transport), while girls show more emotion-related symptoms and behavioral problems [47] and have a greater tendency to disguise their symptoms or difficulties [48]. As a result, ASD in girls may be commonly mistaken for other pathologies or mental health problems [47][48][49]. All these reasons can make the diagnosis of ASD in girls more difficult, and can, in some way, contribute to the low prevalence in girls. Finally, the fact that the most common age of the participants among the included studies was 6 years could be because ASD diagnosis is based on a series of symptoms which stand out more in a school setting, and therefore coincide with the beginning of primary school, at around 6 years [50].
The most used strategy in play-based interventions studied in the included articles was the use of new technologies. Moreover, it should be noted that all the play-based interventions using symbolic play included some elements that could be considered technological, such as digital photographs or video cameras. The wide use of technology may be due to the fact that children and adolescents currently belong to the digital era, i.e., technology is on the rise and is part of their daily lives [51] and, therefore, the use of new technologies can motivate them to follow the intervention. Another reason that can explain the frequent use of new technologies in play-based interventions is that nowadays many technological devices are easy to acquire and are commonly found in private homes, schools and clinics, especially computers, tablets and/or smartphones, increasing their accessibility [52]. In addition, children and adolescents with ASD present a clear preference for games related to new technologies [53], and therefore, their use in interventions may help professionals to increase these children and adolescents' motivation and interest in rehabilitation, which may result in better adherence to treatment [52,54].
Play-based interventions were mostly delivered by parents, and to a lesser extent, by teachers. On the one hand, this result may be justified by the results shown by Althoff et al. in their recently published systematic review [55]. They showed that parent-mediated interventions for children and adolescents with ASD lead to an improvement in several outcomes related to children, such as attention, language, non-verbal communication, social communication, interaction, play, adaptive functioning and ASD symptoms [55]. On the other hand, teachers are one of the groups of professionals that spend most time with children and, therefore, are in a position to identify difficulties in children with ASD earlier, especially difficulty in social play, which is more easily detected in a school setting [55]. This is why training and involving teachers in play-based interventions is important for the inclusion of children with ASD in the school environment, one of the most important settings in a child's life [56,57].
Children and adolescents with ASD also present deficits in social cognition (SC) [58]. Animal studies try to explain SC deficits in people with ASD by anatomical differences in some brain areas related to social competence [59]. SC is defined as a variety of cognitive skills necessary to recognize and use socially relevant information to respond adequately in social situations [60]. In this sense, some authors have described that play is essential for increasing the development of social skills in general [61], thanks to improvements in aspects of SC such as joint attention, social referencing and mentalization due to the development of problem-solving abilities [62,63]. Thus, play-based interventions can be an optimal treatment option to increase SC and should be complemented with a specific and adequate evaluation of SC [64].
Finally, we would like to point out that the present scoping review can be considered as a complementary document to the systematic review carried out by Dijkstra-de Neijs et al. in 2021 [21]. In this systematic review, the authors evaluate the effectiveness of playbased interventions for children and adolescents with ASD, while in our scoping review we describe the characteristics of play-based interventions for children and adolescents with ASD. There is a discrepancy in the number of articles reviewed between the two reviews, possibly because Dijkstra-de Neijs et al. used the Google Scholar database in their literature searches and we did not. This systematic review showed that play-based interventions mainly improve primary dimensions of ASD, specifically communication and social interaction. To these results, we can add that play-based interventions to improve social interaction seem to be characterized by the use of new technologies (Kinect Pico's Adventure and the Secret Agent Society computer game) and delivered once a week, while play-based interventions to improve communication seem to be child-directed free play with therapists and peers, delivered once a week. However, it should be noted that these characteristics have been based on a small number of articles and more research is needed to draw conclusions.
The present scoping review has some limitations. First, as in most review articles, we cannot rule out publication bias, which leads to under-representation of null results of interventions in published articles. However, in the present review we have included several articles in which we found null results. The second limitation is the selection bias, also present in most review articles. We may have been increased it by excluding articles published in different languages other than Spanish or English. However, we have included articles written in English, the language in which the majority of the articles are written. This bias may also have been magnified by only including articles with full text available because we may have overlooked some potentially relevant articles for this scoping review. Third, it was difficult for us to establish the search strategy because the classification of ASD has changed in recent years, resulting in a lack of agreement about which disorders it includes. Thus, we decided to include all the following ASD disorders shown in the American Psychiatric Association's (APA) DSM-IV, 2002 [65] in our search strategy: classic autistic disorder, pervasive developmental disorders, Rett syndrome, childhood disintegrative disorder and Asperger syndrome. As a result, studies analyzing other related disorders present in the ICD-10 may have been overlooked. Fourth, only experimental studies were included, so there may be some biases associated with this type of design, such as results based on a small sample size or a low representative sample. Fifth, the quality of the articles included was not assessed and, although this is not a mandatory requirement in scoping reviews, this may have led to the inclusion of low-quality articles. However, we have included a description of the main limitations of each included article in our results section and we have also collected and shown different issues closely related to the studies' quality, such as funding and conflict of interest, in Table 4. With all this information, we wanted to make readers aware of the different quality indicators of each article so they could be careful when interpreting the results shown in this scoping review.
We can highlight some strengths such as the fact that no other reviews have been found that deal with the same study objective as ours. However, the greatest strength of any scoping review is the identification of knowledge gaps [66] which require new research and more scientific evidence. In this regard, we can highlight that: (1) Europe is very poorly represented in studies on play-based interventions for children and adolescents with ASD; (2) the structure of play-based interventions is unclear and therefore difficult to replicate.

Conclusions
Play-based interventions can be classified into those using new technologies, symbolic play or official techniques, but the most used and current are those using new technologies. These interventions were mainly delivered by parents or teachers of children and adolescents with ASD. They lasted approximately 11 weeks and comprised one weekly session, although this varied between studies. Further studies are needed to create play-based intervention protocols which can be implemented in clinical practice with children and adolescents with ASD, thus promoting evidence-based interventions in this field.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.