The Effect of Play-Based Occupational Therapy on Playfulness and Social Play of Children with Autism Spectrum Disorder: A Systematic Review

ABSTRACT

Since play is the main occupation of children, occupational therapists often use play as an approach to improve playfulness. However, for children with autism spectrum disorder (ASD), there is some uncertainty as to whether playfulness and social play can be improved through play-based interventions. To determine the effect of play-based occupational therapy on playfulness and social play of children with ASD, PubMed, EBSCOhost, OT seeker, Scopus, Sabinet, Cochrane, and ProQuest were searched. Rayyan was used to sort studies and extract data, following the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines. The methodological quality of the 12 eligible studies was reviewed using the Joanna Briggs Institute (JBI) checklist tool and data were synthesized using meta-analysis techniques. This review found moderate to strong evidence for 11 of the 12 studies, that play programmes, with a combination of guided and free play, individual, and group therapy sessions, effectively promoted playfulness and social play.

KEYWORDS:

• Play-based Interventions
• occupational therapy
• paediatrics
• social play
• playfulness
• autism spectrum disorders

Introduction

Play, an essential aspect of childhood development, can foster social interaction, sensorimotor planning, and the development of language-related skills (Jensen et al., 2021; Pyle et al., 2017; Weisberg et al., 2016; Yogman et al., 2018; Zosh et al., 2018). Play is spontaneous, intrinsically motivating, and autonomous. Active involvement and a focus on the process not the end product, are key aspects of play (Yogman et al., 2018; Zosh et al., 2018). Play, according to occupational therapists, is considered the main occupation of the child, since it is the vehicle for learning and contributes so significantly to optimal development (Kars & Akı, 2023; Kuhaneck et al., 2020). Occupational therapists use play as an end, such as when improving playfulness or social play. Play could also be a means to achieve functional or performance-based goals, such as when enhancing activities of daily living. For play to occur, the elements of playfulness are necessary, since playfulness is the way of becoming and remaining involved in play. Playfulness includes intrinsic motivation, which drives the player to desire the initiation, expansion, and continuation of play. The desire to engage in play can also be influenced by an extrinsic source, such as intrinsically and emotionally rewarding social play interactions (Bauminger-Zviely et al., 2020; DiDomenico & Ryan, 2017). Social play is the co-operative play with peers (Bauminger-Zviely et al., 2020). Occupational therapists often use social play scenarios to work on the social interaction skills of children with Autism Spectrum Disorders (ASD).

Children with ASD may experience difficulty using the elements of playfulness to become involved in, maintain or adapt social play (Bass & Mulick, 2007). ASD is a neurodevelopmental disorder, characterized by inflexible thoughts, sensory challenges, repetitive behaviors, restricted interests, and limited theory of mind. About one in 36 children has been identified with ASD in the United States, according to estimates from Centers for Disease, Control and Prevention (CDC’s) Autism and Developmental Disabilities Monitoring (ADDM) Network (Maenner et al., 2023). ASD is reported to occur in all racial, ethnic, and socioeconomic groups and is almost four times more common amongst males than females. Given the above characteristics, children with ASD may not demonstrate certain elements of playfulness (Bass & Mulick, 2007; Maenner et al., 2023; Mukherjee, 2017).

The Relationship Between Playfulness, Social Play, and Play Skills

Play exists in various forms, such as guided and free play, both of which are crucial to development (Pyle et al., 2017; Yogman et al., 2018). Both forms of play can be used by occupational therapists to enhance social-emotional development, problem-solving, social, and play skills (Jensen et al., 2021; Yogman et al., 2018; Zosh et al., 2018). Free play is self-directed without the active involvement of adults (Yogman et al., 2018), whereas guided play lies between free play and direct instruction (Weisberg et al., 2016; Zosh et al., 2018). All forms of play, however, require the elements of playfulness. Playfulness operates as a method of becoming involved in play and prioritizes engagement over reality, convention, and a specific outcome (Masek & Stenros, 2021). The six themes of “focused,” “open-ness,” “framing,” non-consequential,’ “non-real reality” and “unconventional” outline how playfulness operates (Masek & Stenros, 2021).

The playfulness element of intrinsic motivation is crucial to driving all six of these themes, since it drives an individual to initiate, maintain, adapt, and reengage in play (Masek & Stenros, 2021). Social play is an important form of play, since during social play, individuals incorporate playfulness themes such as framing to expand engagement for others by changing the play context (Bauminger-Zviely et al., 2020; Masek & Stenros, 2021). Social play and playfulness could promote the skills which are developed through play engagement. When the ability to skillfully engage in tasks improves, so does self-esteem, therefore possibly influencing occupational performance and satisfaction (Lynch & Ray-Kaeser, 2017) (see Figure B1).

The Themes of Playfulness and Theory of Intrinsic Motivation

The theory of intrinsic motivation suggests that the four factors of “competence,” “relatedness,” “autonomy” and “interest” can promote intrinsic motivation. The desire to reengage in activities occurs when the individual perceives their skills or themselves to be competent enough to engage (Borah, 2021; DiDomenico & Ryan, 2017; Zheng et al., 2023). Playfulness’ element of intrinsic motivation could also be promoted through emotionally rewarding engagement in social, guided, or free play (Bauminger-Zviely et al., 2020; DiDomenico & Ryan, 2017; Zosh et al., 2018). The purpose of this systematic review is to highlight both playfulness and social play, given the importance of playfulness’ element of intrinsic motivation and the way in which playfulness and social play encourage play engagement.

Play-Based Interventions for Learners with ASD

Previous reviews of play-based interventions for children with autism have focused on the improvement of play skills (Kent et al., 2020; Kossyvaki & Papoudi, 2016), communication (Deniz et al., 2022) and play scripts (Luckett et al., 2007). A systematic review by Lukett and colleagues, conducted in 2007, included 13 play interventions all of which were evaluated using single case study designs (Luckett et al., 2007). These interventions focused predominantly on behavior modification and the reduction of self-stimulatory behavior (Luckett et al., 2007). The review highlighted that although behavioral approaches are effective in reducing self-stimulatory and teaching splinter skills in terms of play, the approaches did not influence overall playfulness or social play (Luckett et al., 2007).

Kossyvaki and Papoudi (2016) through a review of 14 studies, similarly, highlighted how play-based interventions focus on developing play skills for children with ASD within the school context. Most of the interventions were delivered by teachers. The authors concluded that developmental-based approaches are more effectively used to teach pretend play and social skills in learners with ASD. Conversely, behavioral approaches could be used for teaching play scripts (Kossyvaki & Papoudi, 2016). Although most of the studies reported positive effects, Kossyvaki and Papoudi (2016) questioned the size of the effect, the generalizability of the results, and longer-term outcomes for play in children with autism. It was concluded that interventions, which ignore the meaning of play are also likely to ignore playfulness and social play as motivating factors. A more recent review of 19 Randomized Control Trials (RCTs) investigated play-based interventions to improve the play skills of children with ASD (Kent et al., 2020). Interventions were delivered by a range of interventionists including clinicians, teachers, caregivers, and peers.

The review reported a significant, although small effect of interventions on play skills (Hedges’ g = 0.439) following meta-analysis of 11 studies. One possible reason for the small effect of these interventions is that the underlying motivation to play was not addressed before trying to improve play skills. Kuhaneck et al. (2020) review of 20 studies of varying levels of evidence, focused specifically on the role of occupational therapy to improve play skills (Kuhaneck et al., 2020). The review included interventions which were focused on the improvement of play skills without addressing underlying motivators such as playfulness’ intrinsic motivation and social play. The study provided evidence that play skills can be improved through a variety of occupational therapy methods, not limited to play-based approaches (Kuhaneck et al., 2020). Kuhaneck and colleagues’ review provided preliminary support for certain strategies such as imitation of the child, modeling, and modification of the play environment which occupational therapists could use to improve play (Kuhaneck et al., 2020).

Deniz et al. (2022) published a protocol for a systematic review which will focus on the improvement of social communication skills for learners with ASD, through parent-mediated play interventions (Deniz et al., 2022). Similarly, O’Keeffe and McNally (2023) evaluated nine studies in their systematic review of play‑based interventions targeting the social communication skills of children with ASD in educational contexts. Both Deniz et al. (2022) and O’Keeffe and McNally’s (2023) publications highlighted that social play, communication skills, and the effectiveness of play interventions for enhancing these aspects, are under-researched (Deniz et al., 2022; O’Keeffe & McNally, 2023). Playfulness’ element of intrinsic motivation and social play are important precursors to the engagement in play, and consequently, the development of play skills (Bauminger-Zviely et al., 2020; Lynch & Ray-Kaeser, 2017; Masek & Stenros, 2021; Yogman et al., 2018).

However, the authors of this systematic review were unable to identify any reviews investigating the effect of play-based occupational therapy on both playfulness and social play as motivators for play engagement. The research question answered by this review is therefore: What is the effect of play-based Occupational Therapy interventions on the playfulness and social play of learners with autism spectrum disorder? The objectives were 1) to determine to what extent play-based interventions are effective in enhancing playfulness and social play. 2) Determine which specific programmes, frameworks or processes are used as a basis for the play-based interventions that target playfulness and social play, and 3) Determine whether certain characteristics (setting, duration, mode of delivery, inclusion of peers, caregivers, or educators) influence the effect of the interventions.

Method

A systematic review, using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) framework was conducted. Although forest plots were generated to analyze the effect of individual interventions, a meta-analysis was not conducted. The results could not be pooled due to heterogeneity (Page et al., 2021). The methodological quality of the included studies was assessed using the study-design-specific checklists compiled by Joanna Briggs Institute (JBI) (Porritt et al., 2014).

Eligibility Criteria

Studies included in the systematic review met the following criteria: Participants were limited to children (two to 18 years) diagnosed with autism spectrum disorder. Dual diagnoses could include any intellectual or learning disorder, with the exception of physical disabilities.

Interventions: Only play-based occupational-therapist-led interventions were included. Interventions could be group- or individual-therapy interventions, conducted in any setting. The intervention had to be delivered to the learners, not to caregivers, teachers, or therapists. Comparisons: Comparators were not limited in terms of intervention, thus could be any strategy, framework or technique used to promote playfulness or social play. Outcomes: Any outcomes related to playfulness, social play, or social competencies, like sharing and turn-taking, were included. Studies, which solely focused on improving reciprocal or expressive communication, or language-use and pragmatics, were excluded. Study type: All levels of evidence (I-V) according to the guidelines for evidence-based practice literature in healthcare, were assessed for inclusion in the systematic review (Tomlin & Borgetto, 2011). However, qualitative studies and book chapters were excluded. Systematic and scoping reviews were included when completing the process of pearling. Data Bases were searched without time, language, or publication status restrictions.

Search Strategy

Studies were identified through the search of electronic databases: EBSCOhost (CINAHL, ERIC, Africa-Wide Information, Academic Search Premier), Cochrane (CENTRAL), PubMed, Sabinet, Scopus, and OT Seeker. Keywords included: disorder (autism spectrum disorder; ASD) and occupational therapy, and play (playfulness or play-based interventions) or social play and social competencies. Searches were structured to exclude behavioral, animal assisted, or music therapy approaches from the titles, keywords, and abstracts of articles. MESH terms were used when searching PubMed. Grey literature was searched using Google Scholar, and unpublished theses were searched for using ProQuest and Sabinet. Pearling was conducted using citation lists, studies included in systematic and scoping reviews, and the reference lists of eligible included studies (see Figure B2).

Data Management and Extraction

The principal investigator (PI) conducted searches according the search strategy above, using specific search strings and keywords (see Figure B2). The PI imported the references from the search results into Rayyan. Rayyan is an online data extraction tool, which allows a reviewer and blinded reviewer to categorize, sort data and include or exclude articles. The PI and a second blinded reviewer first excluded articles based on title and keywords, then based on the abstract of the articles. Thereafter, the remaining articles were read and excluded based on the content of the article, using the inclusion criteria outlined above. Data extraction was done according to the JBI (Stone et al., 2023) and TIDieR checklists (Hoffmann et al., 2016). Four authors (Dionne & Martini, 2011; Fabrizi, 2015; Henning et al., 2016; Kent et al., 2021) of the included articles were contacted. Data from searches were managed by storing the information in table format within excel sheets.

Methodological Quality and Risk of Bias

The methodological quality of the included studies was assessed using the study-design-specific checklists compiled by JBI (Stone et al., 2023). JBI critical appraisal tools address both the risk of bias and methodological quality of published papers. JBI critical appraisal tools assist in assessing the trustworthiness, results, and relevance of studies (Stone et al., 2023). A higher score or percentage, such as 100%, indicates lower risk of bias. Therefore, percentages ranging from 0–40% were considered high risk, 40–70% moderate, and 70–100%, was considered low risk of bias. The study and methodological quality were summarized in a table format, along with other study characteristics. Levels of evidence of the studies included in the systematic review, and quality were listed in table format, according to the guidelines for evidence-based practice and literature within healthcare (Tomlin & Borgetto, 2011).

Data Analysis

Stata 17 (College Station, Texas 77,845 USA) was used for data analysis. We used the reported summary statistics for each intervention group to compute the mean difference for a continuous outcome, with the corresponding 95% confidence interval. If summary statistics were missing, we used the reported test statistic, z-statistic, F-statistic, or p-value to compute the standard error. Due to substantial heterogeneity, pooling results across studies was not appropriate. Individual study results were reported separately. Substantial heterogeneity was defined as I² statistic > 50%. Individual study results were presented on forest plots.

Results

A total of 1768 studies were identified and screened using the search terms (see Figure B2). The PI and co-reviewer (NBC) excluded 1358 studies by title. Thereafter, the remaining 410 abstracts were read for inclusion. Of the 410 abstracts, 351 were excluded for the following reasons: 240 studies were qualitative, 50 were not play-based, and 61 did not include play outcomes. A total of 59 full text articles were assessed for eligibility by NBC and the PI/GR.

Twelve (12) articles were included. Forty-Seven (47) of the fifty-nine (59) articles, were excluded for the following reasons: three included children with physical disabilities or other diagnoses within the sample group of ASD children; six studies did not use a play-based intervention; two studies were home-programme-orientated interventions; 10 interventions were not occupational therapist-led. These interventions may have been play-based but did not involve occupational therapists in any aspect of the intervention; 15 studies did not report playfulness outcomes. Six studies used participants who fell outside of 2–18 years, for example, 0–2 years, or 18 and above, and five were qualitative studies (see Figure B3).

Study and Participant Characteristics

Across the 12 studies, 11 of the interventions were conducted within high-income countries using the World Bank criteria (Hamadeh et al., 2022). Only one of the 12 studies was conducted in India, which is classified as a lower and middle-income country (LMIC) (Anu et al., 2019; Hamadeh et al., 2022) (see Table A1). A total of 299 children with ASD were included across the 12 included studies. The ratio of boys to girls was higher in all the sample groups, with five studies including only boys (Chen et al., 2022; Dionne & Martini, 2011; Henning et al., 2016; Kent et al., 2021; Owens et al., 2008). Only three of the 12 papers reported on the race and ethnicity of participants (Davidson & Stagnitti, 2021; Doernberg et al., 2021; Fabrizi, 2015) (see Table A1). With regards to confirming the diagnosis and level of support required by the participants, six studies used the Diagnostic and Statistical manual of Mental disorders, version 5 (DSM-V) (Anu et al., 2019; Chen et al., 2022; Davidson & Stagnitti, 2021; Henning et al., 2016; Kent et al., 2021; Marwick et al., 2022) (see Table A1).

In terms of the level of support required by the participants with ASD, six studies used participants who required low levels of support (Chen et al., 2022; Doernberg et al., 2021; Henning et al., 2016; Kent et al., 2021; Legoff & Sherman, 2006; Owens et al., 2008). These studies specified that participants had to have an IQ of 70, or an IQ of 85 and above to be included or had to be enrolled in mainstream schools (see Table A1). One study included participants of moderate to low support (Wolfberg et al., 2015). Three studies used participants who required a moderate to high level of support (Anu et al., 2019; Davidson & Stagnitti, 2021; Marwick et al., 2022) and two studies involved children who needed high levels of support (Dionne & Martini, 2011; Fabrizi, 2015) (see Table A1).

Study Intervention Particulars: Frameworks and Strategies

Social play and consequently social skills improvement was the aim of seven of the 12 included studies (Chen et al., 2022; Dionne & Martini, 2011; Henning et al., 2016; Legoff & Sherman, 2006; Owens et al., 2008; Wolfberg et al., 2015). Four of these seven studies implemented play-based interventions which used structured, adult-directed play to achieve social play and social skills outcomes (Legoff & Sherman, 2006; Owens et al., 2008; Wolfberg et al., 2015; Chen et al., 2022 (see Table A2). Improvement of social play and thus social skills was the aim of the three other social play studies (Dionne & Martini, 2011; Henning et al., 2016; Kent et al., 2021). These social play interventions used home play with guidance from caregivers during play, role play, and imitation of play sequences, in combination with adult-directed play (Dionne & Martini, 2011; Henning et al., 2016; Kent et al., 2021) (see Table A2).

Improvement of the elements of playfulness, including intrinsic motivation, and consequently, play skills, was the aim of five out of the 12 included studies (Anu et al., 2019; Davidson & Stagnitti, 2021; Doernberg et al., 2021; Fabrizi, 2015; Marwick et al., 2022). Two of these studies used the Learn to Play Programme as a framework (Anu et al., 2019; Davidson & Stagnitti, 2021), the third study used an aqua playfulness programme (Fabrizi, 2015) and the fourth intervention used a pretend play programme (Doernberg et al., 2021). The fifth study used the Playboxes Joint Play Approach, which incorporated joint play, using an individualized playbox for each participant filled with play scripts, novel, and familiar play objects (Marwick et al., 2022) (see Table A2). All the playfulness interventions used a guided and free play, home play with caregivers’ guidance, imitation of play sequences and role play (Anu et al., 2019; Davidson & Stagnitti, 2021; Doernberg et al., 2021; Fabrizi, 2015; Marwick et al., 2022) (see Table A2).

Study Intervention Particulars: Materials, Setting, and Duration

Duration of the interventions across the 12 studies ranged from one to 36 months, the duration of intervention sessions ranged from 20 to 90 min, and the number of sessions per week ranged from one to four sessions. The most common duration of a therapy session was 60 min. A school context was the intervention setting for five of the 12 studies (Doernberg et al., 2021; Legoff & Sherman, 2006; Marwick et al., 2022; Owens et al., 2008; Wolfberg et al., 2015). One of the play-based interventions used a community pool as the intervention setting (Fabrizi, 2015) (see Table A2). In terms of materials, all seven interventions which aimed to improve social play incorporated the use of familiar play objects, social scripts, and visuals during social play sessions (Chen et al., 2022; Dionne & Martini, 2011; Henning et al., 2016; Kent et al., 2021; Legoff & Sherman, 2006; Owens et al., 2008; Wolfberg et al., 2015). In contrast, the five interventions which aimed to improve the elements of playfulness, incorporated use of familiar play objects, visuals, and video-modeling into sessions (Anu et al., 2019; Davidson & Stagnitti, 2021; Doernberg et al., 2021; Fabrizi, 2015; Marwick et al., 2022).

Study Intervention Particulars: Social Play Measures

Three out of these seven social play interventions used the Vineland Adaptive Behavior Scale (VABS) (Chen et al., 2022; Legoff & Sherman, 2006; Owens et al., 2008). Two studies combined VABS with the Gillium Autism Rating Scale (GARS) to measure post-intervention social skills improvements (Legoff & Sherman, 2006; Owens et al., 2008). It should be noted that these three studies used low support participants with ASD and concluded that VABS and GARS could be more appropriate for verbal participants (Chen et al., 2022; Legoff & Sherman, 2006; Owens et al., 2008).

Out of the five playfulness interventions, two used the Child Initiated Pretend Play Assessment (ChIPPA) to measure post-intervention playfulness elements (Anu et al., 2019; Davidson & Stagnitti, 2021). Fabrizi (2015) used the Test of Playfulness (ToP) and Doernberg et al. (2021) used the Affect in Play Scale (APS) (Doernberg et al., 2021; Fabrizi, 2015). Marwick et al. (2022) used the Test of Pretend Play (ToPP) to measure playfulness (Marwick et al., 2022) (see Table A2).

Study Methodological Quality: Risk of Bias

The level of risk of bias across the 12 included studies ranged from moderate to low risk of bias with scores of 62.5% to 100%, according to the JBI critical appraisal tools. Of the seven studies which aimed to promote social play and social skills, two presented with moderate risk of bias, with scores of 62.5% (Chen et al., 2022; Dionne & Martini, 2011). The other five of these seven social play interventions, presented with low risk of bias scores of 77.7% (Henning et al., 2016), 84.6% (Kent et al., 2021), 88.8% (Wolfberg et al., 2015), 92.3% (Owens et al., 2008) and 100% (Legoff & Sherman, 2006). Of the five playfulness interventions, two presented with moderate risk of bias, with scores of 66.6% (Davidson & Stagnitti, 2021; Marwick et al., 2022). The other three playfulness intervention studies presented with low risk of bias with scores of 77.7% (Fabrizi, 2015) and 100% (Anu et al., 2019; Doernberg et al., 2021) (see Table A3).

Study Methodological Quality: Intervention Reporting

The quality of intervention reporting according to the components of the Template for Intervention Description and Replication checklist (TIDieR) (Hoffmann et al., 2016), ranged from 70% to 100% across the 12 interventions. Only Davidson and Stagnitti (2021) and Marwick et al. (2022) and colleagues reported on 100% of the intervention components (Davidson & Stagnitti, 2021; Marwick et al., 2022) (see Table A4).

Six out of the 12 studies reported on feasibility (Anu et al., 2019; Chen et al., 2022; Davidson & Stagnitti, 2021; Kent et al., 2021; Legoff & Sherman, 2006; Marwick et al., 2022) and only four out of the 12 studies reported on the intervention modifications (Davidson & Stagnitti, 2021; Dionne & Martini, 2011; Doernberg et al., 2021; Marwick et al., 2022) (see Table A4).

Effect of Interventions: Overall Effect and Long-Term Outcomes

Given the variety of effect size calculations, amount of data reported, and outcome measures used by the 12 included studies, it was not possible to calculate one overarching effect size, therefore the effect of the studies is reported individually. The individual effect on playfulness or social play for each of the 12 studies ranged from small to large. Eight of the 12 interventions produced large, significant effect sizes (Anu et al., 2019; Chen et al., 2022; Davidson & Stagnitti, 2021; Fabrizi, 2015; Legoff & Sherman, 2006; Marwick et al., 2022; Owens et al., 2008; Wolfberg et al., 2015). Three of the play-based interventions resulted in a moderate effect (Dionne & Martini, 2011; Doernberg et al., 2021; Kent et al., 2021) and one of the studies resulted in minimal improvements in social play (Henning et al., 2016) (see Table A5). With regard to the long-term outcomes of the play-based interventions, four of the 12 studies conducted follow-up assessments (Chen et al., 2022; Fabrizi, 2015; Henning et al., 2016; Kent et al., 2021). The play-based intervention follow-up periods ranged from one- to four-months post-intervention (Chen et al., 2022; Fabrizi, 2015; Henning et al., 2016; Kent et al., 2021). Chen et al. (2021), (Fabrizi (2015) and Kent et al. (2021) studies reported maintenance of results at follow-up (Chen et al., 2022; Fabrizi, 2015; Kent et al., 2021) and one study reported a decrease in social play (Henning et al., 2016). The overall mean adherence of participants ranged from 66% to 100% (see Table A5).

Effect of the Interventions: Social Play

Four of the seven social play interventions significantly improved social play and social skills for participants with ASD. Through structured, play-based social skills programmes, Chen et al. (2022), Legoff and Sherman (2006), Owens et al. (2008) and Wolfberg et al. (2015) enhanced social competence to a large, clinically significant extent, with effect sizes of (r = 0.54) (Chen et al., 2022); (F = 20.96) (Legoff & Sherman, 2006); (χ2 = 5.85; f = 2) (Owens et al., 2008) and (t = 11.41; d = 1.66) (Wolfberg et al., 2015). The post-intervention p values of these four interventions were less than 0.05 and the 95% confidence intervals did not include the null values (See Figures B4–B7) (Chen et al., 2022; Legoff & Sherman, 2006; Owens et al., 2008; Wolfberg et al., 2015). The key finding of these interventions were that adult-directed, group-based, structured social skills training improved social competence, therefore improving the motivation to engage in social play (Chen et al., 2022; Legoff & Sherman, 2006; Owens et al., 2008) (see Table A5).

On the contrary, two of the seven social play interventions which resulted in moderate improvements, found that social competence can be improved through guided, social, and home play with the involvement of caregivers who can guide the play (Dionne & Martini, 2011; Kent et al., 2021). Kent et al. (2021) play-based intervention moderately improved social play (p = 0.015; d = 0.78). It is important to note that although the intervention focused on improving social play, the study also reported a significant effect on the participants’ post-intervention playfulness levels (p = 0.016; d = 0.51) (see Figure B8) (Kent et al., 2021). The only play-based intervention which had minimal effects on social play was that of Henning and colleagues (Henning et al., 2016). The study reported that only two out of the five participants’ social play skills improved (PND for 2/5 participants = 88%–100%; PND for 3/5 participants = 55%–66%). Due to the limited data provided, the study results could not be included in the forest plots (Henning et al., 2016).

Effect of the Interventions: Playfulness

Of the five interventions which focused on playfulness, four improved playfulness to a large, significant extent (Anu et al., 2019; Davidson & Stagnitti, 2021; Fabrizi, 2015; Marwick et al., 2022) (see Figure B8). The two interventions which used the Learn to Play programme, resulted in significant playfulness improvements for the experimental groups (p < 0.005; 95% CI: 0.000–0.069) (Anu et al., 2019) and (p < 0.005; 95% CI: 0.000–0.043) (Davidson & Stagnitti, 2021). Similarly, Fabrizi’s (2015) play-based aqua group significantly improved all four elements of playfulness for the participants with ASD (ƞ_p² = 0.85; p = 0.014; p < 0.05) (Fabrizi, 2015). Marwick et al. (2022) intervention’s effect on playfulness could not be displayed on the forest plots due to limited data. However, the Playboxes Joint Play Approach resulted in significant improvements in playfulness elements such as suspension of reality and intrinsic motivation. This was represented by increased post-intervention age-equivalent scores for all participants with ASD, ranging from +8 months to +30 months (Marwick et al., 2022). Doernberg et al. (2021) adapted pretend play programme moderately improved playfulness’ elements of framing (ƞ_p² = 0.32; p = 0.004; p < 0.01) and suspension of reality (ƞ_p² = 0.31; p = 0.003; p < 0.01) (Doernberg et al., 2021) (see Figure B8). The overarching key finding of all five playfulness interventions was that targeting the element of intrinsic motivation was the most effective way to influence overall playfulness and consequently, play engagement (Anu et al., 2019; Davidson & Stagnitti, 2021; Doernberg et al., 2021; Fabrizi, 2015; Marwick et al., 2022).

Discussion

This systematic review evaluated the effect of play-based interventions on the playfulness and social play of learners with ASD. The review found moderate to strong evidence for increase in playfulness and social play across 11 of the 12 studies. An overarching key finding was that integral to effective play-based interventions are the factors which promote playfulness’ element of intrinsic motivation, derived from the theory of intrinsic motivation (Borah, 2021; DiDomenico & Ryan, 2017; Zheng et al., 2023). This systematic review also highlighted the integral relationship between social play and the elements of playfulness.

Four of the interventions which targeted social play focused significantly on the outcome of improved social competence to influence social play engagement (Chen et al., 2022; Legoff & Sherman, 2006; Owens et al., 2008; Wolfberg et al., 2015). The LEGO® Play, Integrated Play Groups (IPG) approach and social skills training group used social scripts, visuals, and adult-directed play situations to target the promoting factors of competence and relatedness. These four studies produced large, clinically significant results improving social skills such as sharing, framing, turn-taking, and communication (Chen et al., 2022; Legoff & Sherman, 2006; Owens et al., 2008; Wolfberg et al., 2015).

The interventions emphasized that social competence and social play are intertwined especially in terms of the element of framing (Chen et al., 2022; Wolfberg et al., 2015). Framing is one of the elements of playfulness (Masek & Stenros, 2021). However, it is also a crucial social play skill as it involves adapting the perceived play context to expand the engagement (Chen et al., 2022; Masek & Stenros, 2021). The Floortime® Play (Dionne & Martini, 2011) and Ultimate guide to Learning Play and friendship (Kent et al., 2021) interventions also targeted social competence, demonstrating that improved social skills could increase social play.

However, these play-based approaches involved guided, not adult-directed play (Dionne & Martini, 2011; Kent et al., 2021). The adult-directed, structured programmes produced large effect sizes with regards to social skills (Chen et al., 2022; Legoff & Sherman, 2006; Owens et al., 2008; Wolfberg et al., 2015). However, only Kent’s et al. (2021) social play intervention showed a relationship between improved social play skills and playfulness (post-intervention playfulness: p = 0.016; d = 0.51; post-intervention social play: p = 0.015; d = 0.78). Kent’s et al. (2021) intervention improved social competencies while also improving playfulness’ elements such as framing and the intrinsic motivation to engage in social play (Kent et al., 2021). The study also presented with high methodological quality with risk of bias and intervention reporting scores of 84.6% and 90%, respectively (Kent et al., 2021).

Dionne and Martini’s (2011) study also moderately improved social play skills, especially in terms of initiating and maintaining the play interactions (Dionne & Martini, 2011). It is crucial to note that Floortime® Play was the only study out of the seven social play interventions which involved a participant with ASD who required high levels of support (Dionne & Martini, 2011). In contrast, the participants involved in four of the five studies which targeted playfulness, all required moderate to high levels of support (Anu et al., 2019; Davidson & Stagnitti, 2021; Fabrizi, 2015; Marwick et al., 2022).

Another differentiating aspect was that all five of the playfulness interventions used a combination of free, guided, and social play to target the promoting factors of intrinsic motivation (Borah, 2021; Zheng et al., 2023). The five playfulness interventions also used caregiver-guided home play, visuals and video-modeling to incorporate intrinsic motivation factors such as interest and relatedness (Anu et al., 2019; Davidson & Stagnitti, 2021; Doernberg et al., 2021; Fabrizi, 2015; Marwick et al., 2022).

Fabrizi’s (2015) aqua group improved playfulness by structuring the intervention according to playfulness elements, such as intrinsic motivation (Fabrizi, 2015). Themes of playfulness “non-consequentiality” and “unconventionality” were targeted as well (Masek & Stenros, 2021). Participants felt safe and regulated enough to play without expected outcomes and without a sense of consequence, which was indicated by the post-intervention environmental suitability (TOES) score (χ2 = 7.54; p = 0.023). Additionally, the study had a low risk of bias score of 77.7% and an intervention reporting score of 80% (Fabrizi, 2015). Fabrizi demonstrated the importance of structuring a play-based intervention around playfulness elements and intrinsic motivation, shown by a significant improvement in post-intervention playfulness (F (2,6) = 9.55; ηp2 = 0.85, p = 0.014) and maintenance of results up to 4 months after the intervention (Fabrizi, 2015).

Despite the suitability of Fabrizi’s intervention setting, across the 12 studies, the study contexts did not seem to affect the results of the interventions. The Learn to Play programmes which were conducted in a clinic and school setting both produced significantly large effect sizes (p < 0.005, 95% CI: 0.000–0.069) (Anu et al., 2019) and (p < 0.005, 95% CI: 0.000–0.043) (Davidson & Stagnitti, 2021). These interventions improved the play competencies of imagination, sequencing the play and object substitution. Desire to engage in play increased as the participants’ perception of their play abilities increased (Anu et al., 2019; Davidson & Stagnitti, 2021).

The fourth effective playfulness intervention, and perhaps the most feasible regarding implementation into on-going practice and the class environment, was the Playboxes Joint Play Approach (Marwick et al., 2022). This was the only study out of the 12 play-based, which highlighted the importance of combining familiar play objects individualized to the interests of the participants with novel play objects and unfamiliar play sequences (Marwick et al., 2022).

With increases in every participant’s age equivalent Test of Pretend Play (ToPP) score, ranging from increases of eight to 30 months across the group, the Playboxes Approach significantly improved play engagement and playfulness (Marwick et al., 2022). The moderate risk of bias score of 66.6% should be considered when generalizing or implementing this intervention, however, the intervention reporting score was 100%. The Playboxes approach was shown to be suitable to incorporate into ongoing practice. It is flexible in terms of setting, requires limited resources, can be tailored to the individual, involves weekly short sessions and is suitable for learners with ASD of all levels of support (Marwick et al., 2022).

Integral to these four successful playfulness interventions was the use of caregiver-guided home play, visuals, and video-modeling (Anu et al., 2019; Davidson & Stagnitti, 2021; Fabrizi, 2015; Marwick et al., 2022). Caregiver-guided home play ensured that the intrinsic motivation factor of relatedness was part of the play interactions (Borah, 2021; Masek & Stenros, 2021; Zheng et al., 2023). Previous studies have commented on the important and motivating role of caregiver-guided activities when improving occupational performance, social interaction and play at home for children with ASD (Althoff et al., 2019; Deniz et al., 2022; Solomon et al., 2014). The success of using visuals and video-modeling to motivate learners with ASD to participate in activities and learn new skills is also heavily supported by previous literature (Deniz et al., 2022; Kent et al., 2020; Kossyvaki & Papoudi, 2016; Luckett et al., 2007).

Doernberg’s pretend play approach (Doernberg et al., 2021) was the only playfulness intervention which improved playfulness elements to a moderate extent. Similarly, Henning’s et al. (2016) intervention minimally improved social play (Henning et al., 2016).

Although the study presented with low risk of bias (77.7%), Henning’s et al. (2016) intervention only enhanced social play for two out of the five participants with ASD (Henning et al., 2016). Doernberg et al. (2021) and Henning’s et al. (2016) studies are worth discussing, since at five- and seven-weeks duration, respectively, both consisted of brief intervention periods (Doernberg et al., 2021; Henning et al., 2016). The intervention periods of the other 10 studies were two or more months in length. Previous studies have shown that it may be difficult for children with ASD to learn social play skills and integrate elements of playfulness into play (Gretschel et al., 2022; Jordan, 2003; Kent et al., 2020; Kuhaneck et al., 2020). More practice opportunities and repetition over a longer period of time, not necessarily more intensive therapy sessions, may be required to improve playfulness and social play (Linstead et al., 2017).

Another intervention characteristic worth considering is the outcome measures used to assess playfulness and social play. Legoff and Sherman (2006), Chen et al. (2022) and Owens et al. (2008) concluded that adult-directed, structured play, and behavior scales such as the Vineland Adaptive behavior Scale (VABS) should be used with caution, since learners with ASD may find it difficult to interact during standardized assessment processes. Those with moderate to high support needs may respond more willingly to assessment situations where there is no verbal interaction or social play with the assessors or other playmates (Beukelman & Light, 2020; Kasari et al., 2013).

Since caregiver-guided home play was an integral part of the eight playfulness studies, caregiver-rated play scales were important to include in the assessment process. The four studies with moderate to high support participants, which aimed to improve playfulness, used observational play assessments such as the ToP in combination with caregiver-rated play scales such as the Test of Environmental Supportiveness (TOES) and Pretend Play Enjoyment Developmental Checklist (PPE-DC) (Anu et al., 2019; Davidson & Stagnitti, 2021; Fabrizi, 2015; Marwick et al., 2022). It should be highlighted that these caregiver-rated checklists only pertained to certain play skills or play engagement, not necessarily playfulness and social play. However, an outcome measure which could be caregiver-rated and appropriate for measuring both social play and playfulness, is a modified Canadian Occupational Performance Measure (M-COPM) (Beheshti et al., 2022; Rodger et al., 2004). The M-COPM allows caregivers to rate social play, playfulness, and occupational performance goals on behalf of the participants with ASD (Beheshti et al., 2022; Rodger et al., 2004).

Underlying Mechanisms of Successful Play-Based Interventions

Several underlying mechanisms can lead to successful play-based interventions. Focusing on the four factors which can promote playfulness’ element of intrinsic motivation, can be a way to improve play engagement. To improve playfulness and social play for children with ASD, it is recommended that occupational therapists follow six main principles.

Firstly, to promote motivation to play, a combination of free, guided, and social play should be used. Secondly, visuals and videos should be used for modeling play and enhancing intrinsic motivation to engage. The promoting factor of interest or curiosity will be targeted when appealing, relevant visuals and videos are used during play. The third principle is about using home play with guidance from caregivers, which involves the playfulness theme of “open-ness” or the flexibility and willingness to engage in various play contexts. The factors of interest and competence can be targeted through the fourth principle, which recommends combining familiar and novel play objects and scripts (Marwick et al., 2022).

The fifth principle is that interventions which are 2 months or more in length seem to be more effective than shorter play-based intervention periods (Linstead et al., 2017). Lastly, outcome measures such as caregiver-rated or observational play scales, could be more appropriate for minimally verbal learners with ASD who require higher levels of support (Beukelman & Light, 2020; Kasari et al., 2013) (see Figure B9).

Strengths and Limitations

The Risk of Bias in Systematic reviews (ROBIS) tool was used to highlight limitations and strengths of this systematic review (Whiting et al., 2016). Efforts were made to minimize selection of studies and the second reviewer assisted in reducing bias. The use of the JBI checklists may have introduced bias, simply because the risk of bias score is correlated with a broad percentage range, not a specific calculation. The heterogeneity of the studies affected the pooling of results and overall effect size for the studies could not be calculated. Given the heterogeneity, the synthesis was presented as robustly as possible through forest plots. Although not a limitation of this review, two significant gaps in occupational therapy research were found. Only one play-based intervention was conducted in an LMIC (Anu et al., 2019) and only two interventions included high support learners with ASD (Dionne & Martini, 2011; Fabrizi, 2015).

Conclusion

Play-based interventions have been shown to improve playfulness and social play in learners with ASD. Strategies to enhance playfulness include incorporating home, guided, social, and free play to promote the motivation factors of autonomy, competence interest, and relatedness. Motivation to play could also be enhanced through visuals and familiar toys in combination with novel play objects. Play-based interventions, which are two or more months in length, could be more successful than shorter intervention periods. Caregiver-rated play scales combined with observation-based playfulness assessments, could be more appropriate for minimally verbal learners with ASD who require higher levels of support. The systematic review also found that play-based interventions which involve moderate to high support learners with ASD from LMICs are scarce. Additionally, the use of play programmes such as the Playboxes Approach could result in successful intervention in low resource environments since it can be adapted for various settings and the individual needs of the participants.

Acknowledgements

Acknowledgement is provided for the following individuals who assisted with the review process: Ingrid Van der Westhuizen (Faculty Librarian, Information Technology, Stellenbosch University).

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.

Appendix A:

Tables

Table A1. Study particulars.

References	Location	Diagnosis, age range	Demographics: gender; level of support; ethnicity/race	Sample size, group descriptors
Anu et al. (2019)	India	ASD (diagnosis: DSM-V)	−35 males, 7 females in control	42 participants in total (n = 42) −experimental group (n = 21) −control group (n = 21) −no information about prior treatment
		3–7 years	−2-word verbalization, productive interaction
		mean: 4.3 years	−medium to high support
Chen et al. (2022)	Taiwan	ASD (diagnosis: DSM-V)	−7 males	7 participants in total (n = 7) −4 received rehabilitation prior to intervention −only experimental group, no control
		6–7 years	−1st Grade, IQ of 85 or above
		mean: 78.1 months	−low support
Davidson and Stagnitti (2021)	Australia	ASD (diagnosis: DSM-V)	−5 males; 1 female	6 participants in total (n = 6) −one parent per child participated in the study (forming a dyad) −only experimental group, no control
		2.5–5.9 years	−developmental level: 12–18 months
		Mean: 3.8 years	−one meaningful word/focused attention
			−medium to high support
Dionne and Martini (2011)	United Sates of America	ASD (diagnosis: CARS)	−1 male	1 participant in total (n = 1) −received prior occupational therapy
		3.6-year-old boy	−single word utterances, use of Picture Exchange Communication System (PECS®) to
		diagnosed at 2.5 years.	communicate −high support
Doernberg et al. (2021)	Australia	ASD-HF (diagnosis: ADOS)	−22 males; 3 females	25 participants in total (n = 25) −experimental group (n = 18) −control group (n = 7) −no information about prior treatment
		6–9 years	−IQ 85 and above; mainstream schooling
			−low support
			−10 Caucasian; 2 African American; 1 Asian; 3 Other; 9 not reported
Fabrizi (2015)	United States of America	ASD (diagnosis:	−6 males; 4 females	10 participants with ASD (n = 10) −ASD participants paired with caregiver (dyad) −only experimental group, no control
		ADOS) 2–3 years	−high support
			−4 Hispanic; 3 Caucasian; 2 African American; 1 Asian
Henning et al. (2016)	Australia	ASD (diagnosis: DSM-V)	−5 males	5 participants in total (n = 5)
		4–11 years	−IQ of 70 and above for those with ASD	−ASD participants paired with typically developing peers (dyad)
		Mean: 7.8 years	−low support	−no information about prior treatment
			−4 Caucasian; 1 Asian in ASD group	−only experimental group, no control
Kent et al. (2020)	Australia	ASD (diagnosis: DSM-V)	−57 males	57 participants in total (n = 57)
		6–12 years	−only mainstream schooling included	−experimental group (n = 30)
		Included ADHD and anxiety;	−low support	−control group (n = 27)
		Excluded ID
Legoff and Sherman (2006)	United States of America	ASD (DSM-IV)	−96 males; 21 females	117 participants in total (n = 117)
		Mean ages-	−all were enrolled ay mainstream schools	−experimental group (n = 60)
		Experimental group: 9.3 years	−low support	−control group (n = 57)
		Control: 10.1 years
Marwick et al. (2022)	United Kingdom	ASD (diagnosis: DSM-V)	−6 males; 1 female	7 participants in total (n = 7)
		5–8 years	−medium to high support	−no information regarding prior intervention
		2 participants: moderate learning	−all enrolled in learners with special education needs (LSEN) schools	−only experimental group, no control
		difficulties
Owens et al. (2008)	United Kingdom	ASD (diagnosis: DSM-IV)	−31 males	31 participants in total (n = 31)
		6–11 years	−IQ of 70 and above	−experimental group (n = 16)
			−speaking in appropriate phrases	−control group (n = 16)
			−low support	−prior rehabilitation
Wolfberg et al. (2015)	United States of America	ASD (diagnosis: ADOS)	−41 males; 7 females	48 participants in total (n = 48)
		5–10 years	−ranged from pre-verbal to fluent speech	−experimental group (n = 9)
			−medium to low support	−control group (n = 39)
				−no information regarding prior intervention

Table A2. Intervention particulars.

Reference	Study Aims	Intervention Framework	Strategies used	Materials used	Duration and setting	Social Play and Playfulness Outcome Measures
Anu et al. (2019)	Improve social competence through pretend play	Learn to Play programme	Group & individual therapy Home play Imitation of play sequences Guided & free play	Familiar play objects Visuals Video-modeling	6-month intervention 80–100 sessions 1 30-minute individual session & 1 30-minute group session per week Clinic, school	−ChIPPA −Caregiver-rated goals −SIPDC
Chen et al. (2022)	Determine the efficacy of a play-based social skills programme	Play-based social skills training programme	Group therapy Imitation of play sequences Social skills games Homework assignments Role Play situations Adult-directed play	Familiar play objects Visuals Social skills scripts	3-month intervention 10–12 sessions 1 60-minute group session per week Clinic, school	−Video Coding −GAS goals −VABS-3
Davidson and Stagnitti (2021)	Analyze the effectiveness of the Learn to Play Process	Learn to Play programme	Group & individual therapy Home play Imitation of play sequences Guided & free play	Familiar play objects Visuals Video-modeling	6-month intervention 24 sessions 1 30-minute individual session & 1 30-minute group session per week School	−For 3 years or over: ChIPPA; −younger child: assessed using (PPE-DC) −videos of sessions analyzed using PPE-DC) −caregiver-rated goals
Dionne and Martini (2011)	To determine effectiveness of Floor Time® Play	Floor Time® Play programme	Individual therapy Home play Imitation of play sequences Guided play	Familiar play objects Visuals Social skills scripts	2-month intervention 35 sessions 4 45-minute individual sessions per week Clinic	−CARS −Sensory Profile −FAES −CoC −Video analysis
Doernberg et al.(2021)	To determine the efficacy of an adapted pretend play programme (for high-functioning ASD)	Adaption of a preexisting pretend play programme which was originally developed for typically developing children	Individual therapy Home play Imitation of play sequences Guided & free play	Familiar play objects Visuals	5-week intervention 5 sessions 1 20-minute individual session per week School	−The Affect in Play Scale (APS) −KAIR
Fabrizi (2015)	Analyze the effectiveness of an aquatic play group to improve playfulness and social play of children with ASD	The Test of Playfulness (ToP) items were used to create a model for play-based groups in the aquatic environment	Group therapy Home play Imitation of play sequences Guided & free play	Familiar aquatic toys Visuals Video-modeling	2-month intervention 9 sessions 1 40-minute group session per week Community pool	−ToP −TOES: caregiver rated
Henning et al. (2016)	Determine the feasibility and effectiveness of a play-based intervention to address the social play skills	Adapted design of a previous play-based intervention for improving the social play skills of children with ADHD	Individual therapy Home play Imitation of play sequences Guided play	Familiar play objects Visuals Social skills scripts	7-week intervention 8 sessions 1 60-minute individual session per week Clinic, School	−ToP −CCBRS-P −CCC-2
Kent et al. (2020)	Improve social play of children with ASD through peer-mediated intervention	The Ultimate Guide to Play, Learning and friendship programme (PLF)	Individual therapy Home play Imitation of play sequences Role Play situations Guided play	Familiar play objects Visuals Video-modeling Social skills scripts	3-month intervention 10 sessions 2 30-minute individual sessions per week Clinic, school	−ToP −caregiver-rated goals −PRQ -Piers-Harris
Legoff and Sherman (2006)	Determine the effectiveness of LEGO® Play to improve social play of children with ASD (high functioning)	LEGO® Play programme	Group Therapy Imitation of play sequences Modelling Adult-directed play	Familiar LEGO® blocks Visuals Social skills scripts	36-month retrospective Intervention; number of sessions not provided 1 90-minute group session per week School	−VABS-3 −GARS-SI
Marwick et al. (2022)	Encouraging playfulness, social and joint play with use of playboxes	The Playboxes Joint Play Approach	Group & individual therapy Home play Imitation of play sequences Peer involvement Fantasy and Role Play Guided & free play	Familiar play objects Novel play objects Visuals Video-modeling Fantasy play scripts	3-month intervention 10 sessions 1 40-minute individual session & 1 40-minute group session per week School	−ToPP −SPT
Owens et al. (2008)	Determine the effectiveness of LEGO® Play to improve social play of children with ASD (high functioning)	LEGO® Play programme	Group Therapy Imitation of play sequences Adult-directed play	Familiar LEGO® blocks Visuals Social skills scripts	5-month intervention; number of sessions not provided 1 60-minute group session per week School	−VABS-3 −GARS-SI
Wolfberg et al. (2015)	Determine the effect of the Integrated Play Groups (IPG) Approach on the social play of children with ASD	Integrated Play Groups Approach	Group Therapy Imitation of play sequences Scaffolding Modelling Adut-directed play	Familiar play objects Visuals Social skills scripts	3-month intervention; number of sessions not provided 2 60-minute group sessions per week School	−Video analysis −IPG Approach Assessments (found in IPG guidelines)

Key for intervention particulars table

ChIPPA: Child Initiated Pretend Play Assessment; FAES: Functional Emotional Assessment Scale

CoC: Closed Circles of Communication; PPE-DC: Pretend Play Enjoyment Developmental Checklist

K-CPT2: Conners Kiddie Continuous Performance Test-Second Edition; CCBRS-P: Conners’ Comprehensive Behaviour Rating Scales–Parent

CCC-2: Children’s Communication Checklist; WISC–III: Wechsler Intelligence Scale for Children–Third Edition

KAI-R: The Kusche Affective Inventory Revised (KAIR); HCSBS: Home and Community Social Behaviour Scales

WPPSI–R: Wechsler Preschool and Primary Scale of Intelligence–Revised; CDDC: Communication Deall Developmental Checklist

SIPDC: Symbolic and imaginative play developmental checklist; VABS: Vineland Adaptive Behaviour Scales

GARS-SI: Gilliam Autism Rating Scale; ToP: Test of Playfulness

TOES: Test of Environmental Supportiveness; ToPP: Test of Pretend Play

GAS: Goal Attainment Scales; SPT: Symbolic Play Test

PRQ: Parent Relationship Questionnaire

Table A3. Risk of Bias.

		JBI Checklist results, according to Yes (Y) and No (N) answers
Reference	Study Design	1	2	3	4	5	6	7	8	9	10	11	12	13	Analysis of Risk of Bias (ROB) with reasons	Evidence level
Anu et al. (2019)	Quasi Experimental	Y	Y	Y	Y	Y	Y	Y	Y	Y					LROB: 9/9 (100%) −very low risk, all limitations mentioned in study	II
Chen et al. (2022)	Quasi Experimental	Y	Y	UN	N	Y	Y	N/A	Y	Y					MROB: 5/8 (62.5%) −no control; unclear about prior treatment	III
Davidson and Stagnitti (2021)	Quasi Experimental	Y	Y	Y	N	N	UN	Y	Y	Y					MROB: 6/9 (66.6%) −no control; no multiple measures of outcome	III
Dionne and Martini (2011)	Cross Sectional	N	Y	Y	Y	N	N	Y	Y						MROB: 5/8 (62.5%) −criteria unclear; no follow up;	IV
Doernberg et al. (2021)	Quasi Experimental	Y	Y	Y	Y	Y	Y	Y	Y	Y					LROB: 9/9 (100%) −very low risk; all limitations mentioned by study	II
Fabrizi (2015)	Quasi Experimental	Y	Y	UN	N	Y	Y	Y	Y	Y					LROB: 7/9 (77.7%) −unclear about previous treatment; no control	III
Henning et al. (2016)	Quasi Experimental	Y	Y	UN	N	Y	Y	Y	Y	Y					LROB: 7/9 (77.7%) −unclear about previous treatment; no control	III
Kent et al. (2021)	RCT	Y	UN	Y	UN	Y	Y	Y	Y	Y	Y	Y	Y	Y	LROB: 11/13 (84.6%) −providers of treatment not blinded	I
Legoff and Sherman (2006)	Quasi Experimental	Y	Y	Y	Y	Y	Y	Y	Y	Y					LROB: 9/9 (100%) −very low risk, all limitations mentioned in study	II
Marwick et al. (2022)	Quasi Experimental	Y	Y	Y	N	Y	UN	Y	Y	UN					MROB: 6/9 (66.6%) −no control group; unclear about follow up	II
Owens et al. (2008)	RCT	Y	Y	Y	Y	N	Y	Y	Y	Y	Y	Y	Y	Y	LROB: 12/13 (92.3%) −providers of treatment not blinded to treatment	I
Wolfberg et al. (2015)	Quasi Experimental	Y	Y	Y	N	Y	Y	Y	Y	Y					LROB: 8/9 (88.8%) −no control group	II

Key for JBI Checklist evaluation questions in relation to study design

	Quasi-Experimental study design	Randomised Control Trial study design	Cross Sectional study design
1	Is it clear in the study what is the “cause” and the “effect”?	Was true randomization used for assignment of participants to treatment groups?	Were the criteria for inclusion in the sample clearly defined?
2	Were the participants included in any comparisons similar?	Was allocation to treatment groups concealed?	Were the study subjects and the setting described in detail?
3	Were the participants included in any comparisons receiving similar treatment?	Were treatment groups similar at the baseline?	Was the exposure measured in a valid and reliable way?
4	Was there a control group?	Were participants blind to treatment assignment?	Were objective, standard criteria used for measurement of the condition?
5	Were there multiple measurements of the outcome both pre- and post-intervention?	Were those delivering treatment blind to treatment assignment?	Were confounding factors identified?
6	Was follow up complete/were differences between groups in terms follow up analysed?	Were outcomes assessors blind to treatment assignment?	Were strategies to deal with confounding factors stated?
7	Were the outcomes of participants included in any comparisons measured in the same way?	Were treatment groups treated identically other than the intervention of interest?	Were the outcomes measured in a valid and reliable way?
8	Were outcomes measured in a reliable way?	Was follow up complete/were differences between groups in terms follow up analysed?	Was appropriate statistical analysis used?
9	Was appropriate statistical analysis used?	Were participants analyzed in the groups to which they were randomized?
10		Were outcomes measured in the same way for treatment groups?
11		Were outcomes measured in a reliable way?
12		Was appropriate statistical analysis used?
13		Was the trial design appropriate, and any deviations from the standard RCT design (individual randomization, parallel groups) accounted for?

LROB: Low Risk of Bias.

MROB: Moderate Risk of Bias.

HROB: High Risk of Bias.

N/A: Not Applicable.

UN: Unclear.

Y: Yes.

N: No.

Table A4. Quality of Intervention reporting (Tidier Checklist).

Study, Year	Name	Rationale	Materials	Providers	Delivery	Location	Period	Modifications	Feasibility: planned	Feasibility: Actual	Score	Percentage
Anu et al. (2019)	1	1	1	1	1	1	1	0	1	1	9	90%
Chen et al. (2022)	1	1	0	1	1	1	1	0	1	1	8	80%
Davidson and Stagnitti (2021)	1	1	1	1	1	1	1	1	1	1	10	100%
Dionne and Martini (2011)	1	1	0	1	1	1	1	1	0	0	8	70%
Doernberg et al. (2021)	1	1	1	1	1	1	1	1	1	0	9	90%
Fabrizi (2015)	1	1	1	1	1	1	1	0	1	0	8	80%
Henning et al. (2016)	1	1	0	1	1	1	1	0	1	0	7	70%
Kent et al. (2021)	1	1	1	1	1	1	1	0	1	1	9	90%
Legoff and Sherman (2006)	1	1	1	1	1	1	1	0	1	1	9	90%
Marwick et al. (2022)	1	1	1	1	1	1	1	1	1	1	10	100%
Owens et al. (2008)	1	1	1	1	1	1	1	0	1	0	8	80%
Wolfberg et al. (2015)	1	1	1	1	1	1	1	0	1	0	8	80%

Table A5. Effect of the interventions.

Reference	Key findings	Data correlating to key findings and mean adherence of participants	Overall effect of the intervention on social play and playfulness (post-intervention)	Long term outcomes for social play and playfulness (follow-up)
Anu et al. (2019)	Learn to Play is more effective than general social-skills training in improving social play, play skills and playfulness. Play-based occupational therapy can enhance the social play and playfulness of children with ASD.	19/21 participants in the experimental group showed indicators of typical play post-intervention. Significant improvement (p < 0.005, 95% CI: 0.000–0.069) in pretend and social play in experimental group. Mean adherence = 100%	Overall large effect Effect of pretend play in the experimental group (p < 0.005, 95% CI: 0.000–0.133) was more than that in the control group who participated in a general social skills programme.	No follow up assessments
Chen et al. (2022)	Intervention protocol to improve social play and social skills of children with ASD was developed. Play-based occupational therapy can enhance the social play and social skills of children with ASD. Video-analysis and GAS goals could be more effective in assessing long term effects of social play in children with ASD.	GAS goals relating to social play and social skills improved from a score of −2 to +2 post-intervention. VABS-3 scores showed improvement for two subdomains: written (r = 0.64) and receptive language (r = 0.59). These are large effect sizes but do not reflect social play. Mean adherence = 83%	Overall large effect Effect size for social play according to GAS goals and video analysis (r = 0.54) and effect size for positive affect during play (r = 0.63); (p < 0.05)	Follow up 1-month post-intervention Post-intervention scores were maintained but did not increase. Suggested that VABS-3 may not be sensitive enough to smaller changes in social play.
Davidson and Stagnitti (2021)	Learn to Play improved playfulness and play skills in children with ASD. Key techniques used by the therapist during play, such as repetition were reduced as the child’s playfulness improved.	Frequency of therapists’ repetition pre-intervention (n = 12). Frequency of therapists’ repetition post-intervention (n = 3). Frequency of participants’ play initiation pre-intervention (n = 11). Frequency of participants’ play initiation post-intervention (n = 16). Mean adherence = 84%	Overall large effect Social play and playfulness (play initiation, role play, substitution, positive affect) factors improved in the experimental group post-intervention (p < 0.005, 95% CI: 0.000–0.043).	No follow up assessments
Dionne and Martini (2011)	Floortime Play improved playfulness and social play in a child with ASD. Intensive (more frequent) therapy sessions over a shorter period improved social play more than less frequent sessions over a longer period.	Mean number of closed circles of communication (COCs) pre-intervention (n = 10). Mean number of CoCs post-intervention (n = 15.7). Mean adherence = 100%	Overall moderate effect Significant difference between CoC phase B and CoC in phase A. 2–5 successive sessions (22 in total) fell outside SD line= moderate significance.	No follow up assessments
Owens et al. (2008)	LEGO® Play improved social skills of children with high functioning ASD more so than the general language and social skills programme. LEGO® Play was found to only be suitable for verbal children with ASD.	11/16 children in the LEGO® Play group gave an enjoyment (positive affect) score of 10/10 on a ‘Parent satisfaction and Child enjoyment’ scale. 5/15 children in the Social Use of Language Programme (SULP) group gave 10/10 on this scale. No significant difference in the mean of these scores between the groups. Mean adherence not provided	Overall large effect Gillium Autism Rating Scale-Social Interaction subdomain (GARS-SI) pre-intervention scores (χ2 = 0.844, df = 2, p = 0.66) Gillium Autism Rating Scale-Social Interaction subdomain (GARS-SI) post-intervention scores (χ2 = 5.85, df = 2, p = 0.05)	No follow up assessments
Wolfberg et al. (2015)	Social and pretend play skills of participants improved due to the IPG intervention. However, playfulness factors were not influenced.	Social play domain score revealed significant main effects post-intervention (F = 7.66; p <0.01) Mean adherence not provided	Overall large effect Symbolic Play post-intervention (t = 6.59; p < 0.001; d = 0.96) Social Play post-intervention (t = 11.41; p<0.001; d = 1.66)	No follow up assessments

Key for effect of the interventions table

r: Pearson’s r. Value closer to -1 or 1indicates higher effect size. A positive value means both variables either increase or decrease together (Sullivan & Feinn, 2012).

d: Cohen’s d. 0.2 = small; 0.5 = medium; 0.8 or greater = large effect size (Sullivan & Feinn, 2012).

F: Cohen’s f. Estimates the proportion of variance in a sample. 0.10 = small; 0.25 = medium; 0.40 = large (Chen & Chen, 2010).

p: Relates to the null hypothesis of a study. Small p value = more likely to reject null hypothesis (Sullivan & Feinn, 2012).

t: t Statistic. To determine if there is a statistically significant difference between the means of two variables and whether to reject the null hypothesis (Chen & Chen, 2010).

PND: Percentage of non-overlapping data. PND < 50% = no observed effect; 50–70% = questionable; > 70% = effective (Henning et al., 2016).

ƞ_p² : Partial eta squared. Effect size used in studies using a repeated measures design. 0.01 = small; 0.06 = medium; 0.14 = large effect size (Richardson, 2011).

χ² : Chi-squared is to determine if a difference between observed data and expected data is due to chance, or due to a relationship between the variables being studied. Larger

chi-squared value is associated with smaller p value (McHugh, 2013).

CI: Confidence Interval. Indication of a population’s parameters. An estimate of how certain or uncertain the estimated difference between variables is (O’Brien & Yi, 2016).

df: Degrees of freedom. The number of independent pieces of information used to calculate a statistic of the sample. Small sample size = only a few independent pieces of information = small degree of freedom (Chen & Chen, 2010).

Appendix B:

Figures

Figure B1. Diagram of the relationship between playfulness, play and social play.

Figure B2. Search Strings.*ti, ab, kw when searching Cochrane, refers to ‘title’, ‘abstract’ and ‘keyword’. For example, the articles’ title,abstract and keywords had to contain the word ‘autism’.

Figure B3. PRISMA flow diagram.

Figure B4. Forest plot of a study which improved social play.

Figure B5. Forest plots of a study which improved social play.

Figure B6. Forest Plot of a study which improved social play.

Figure B7. Forest Plot of a study which improved social play.

Figure B8. Forest plot of studies which improved playfulness and social play.

Figure B9. Principles of a successful play-based intervention for children with ASD.