REVIEW ARTICLE

The Effect of Physical Activity Interventions on Youth with Autism

Spectrum Disorder: A Meta-Analysis
Sean Healy , Adam Nacario, Rock E. Braithwaite, and Chris Hopper

The purpose of this meta-analysis was to examine the effect of physical activity interventions on youth diagnosed
with autism spectrum disorder. Standard meta-analytical procedures determining inclusion criteria, literature searches
in electronic databases, coding procedures, and statistical methods were used to identify and synthesize articles
retained for analysis. Hedge’s g (1988) was utilized to interpret effect sizes and quantify research findings. Moderator
and outcome variables were assessed using coding procedures. A total of 29 studies with 30 independent samples
(N 5 1009) were utilized in this analysis. Results from meta-analyses indicated an overall moderate effect (g 5 0.62).
Several outcomes indicated moderate-to-large effects (g  0.5); specifically, moderate to large positive effects were
revealed for participants exposed to interventions targeting the development of manipulative skills, locomotor skills,
skill-related fitness, social functioning, and muscular strength and endurance. Moderator analyses were conducted to
explain variance between groups; environment was the only subgrouping variable (intervention characteristics) to
produce a significant difference (QB 5 5.67, P < 0.05) between moderators. While no significant differences were found
between other moderators, several trends were apparent within groups in which experimental groups outperformed
control groups. Autism Res 2018, 0: 000–000. V C 2018 International Society for Autism Research, Wiley Periodicals,

Inc.

Lay Summary: Results of the meta-analysis—a method for synthesizing research—showed physical activity interven-
tions to have a moderate or large effect on a variety of outcomes, including for the development of manipulative
skills, locomotor skills, skill-related fitness, social functioning, and muscular strength and endurance. The authors
conclude that physical activity’s standing as an evidence-based strategy for youth with ASD is reinforced.

Keywords: Exercise; ASD; evidence-based strategy; sport; youth

Introduction affect health outcomes. Physical activity allows the

human body to develop and strengthen across the five
Individuals with Autism Spectrum Disorder (ASD)—chil- components of physical fitness; muscular strength,
dren and adults—are at an even greater risk of having muscular endurance, cardiorespiratory endurance, flexi-
co-occurring medical and psychiatric illnesses (Bauman, bility, and body composition (Caspersen, Powell, &
2010; Bradley & Bolton, 2006; Croen, et al., 2015), Christenson, 1985). Working to improve or maintain
including obesity and cardiovascular disease (McCoy, these factors within everyday life helps prevent life-
Jakicic, & Gibbs, 2016) compared to the general popula- threatening diseases and conditions, such as cardiovas-
tion. For example, Curtin et al. (2010) reported obesity cular disease, diabetes, hypertension, and obesity (CDC,
rates of 30.4% among the sample with ASD, in compari- 2016).
son to 23.6% of the group without ASD, a finding Individuals with ASD have been shown to be less
reflected in other research (Curtin et al., 2005; Memari active than their TD counterparts (Healy, Haegele, Gre-
et al., 2012; Phillips et al., 2014). Research has also nier & Garcia, 2017; McCoy et al., 2016; Cai & Korn-
revealed that individuals with ASD have lower physical span, 2012). Parents have also indicated that their
fitness scores (cardiovascular endurance, upper body children with ASD participated in significantly fewer
and abdominal muscular strength and endurance, and types of physical activities than their TD peers, as well
lower body flexibility) when compared to their typically as spent less time annually participating in these physi-
developing (TD) peers (Pan et al., 2016). Physical activ- cal activities compared to their TD peers (Bandini et al.,
ity participation is one modifiable risk factor that can 2013). Researchers have found age to be a determinant

From the Department of Behavioral Health and Nutrition, University of Delaware, Newark, Delaware (S.H.); Department of Kinesiology and Recrea-
tion Administration, Humboldt State University, Arcata, California (A.N., R.E.B., C.H.)
Received July 13, 2017; accepted for publication March 19, 2018
Address for correspondence and reprints: Sean Healy, Department of Behavioral Health and Nutrition, University of Delaware, 26 North College
Avenue, Newark, DE 19716. E-mail: healys@udel.edu
Published online 00 Month 2018 in Wiley Online Library (wileyonlinelibrary.com)
DOI: 10.1002/aur.1955
C 2018 International Society for Autism Research, Wiley Periodicals, Inc.
V

INSAR Autism Research 00: 00–00, 2018 1

factor in a child with ASD’s activity level, and older characteristics. This restricts our understanding of how
children with ASD are significantly more inactive than a broad range of moderators may account for the differ-
their younger peers (MacDonald, Esposito, & Ulrich, ences in the effect being meta-analyzed, thus limiting
2011; Memari et al., 2013). A multitude of barriers pre- how the meta-analysis informs practice.
venting increased participation exist (Must, Phillips, The primary purpose of this study is to determine the
Curtin, & Bandini, 2015; Healy et al., 2017; Obrusni- effect of physical activity interventions on young indi-
kova & Miccinello, 2012; Obrusnikova & Cavalier, viduals diagnosed with ASD. For the purpose of this
2011; Healy, Msetfi, & Gallagher, 2012): on a child/fam- study, interventions were included that were conducted
ily level, barriers reported by parents and children in physical activity settings: physical activities were
include the need for supervision, behavioral problems, defined as (a) activities resulting in energy expenditure
and motor skills deficits (Must et al., 2015; Healy, Msefi, that are planned, structured, and repetitive, and pur-
& Gallagher, 2012), a preference for sedentary behav- posely completed to target the development of skills
iors, particularly screen-based activities (Healy et al., used for exercise, including motor and sport skill devel-
2017; Obrusnikova & Cavalier, 2011), and parental time opment, or social skills, or (b) exercise interventions;
constraints (Obrusnikova & Miccinello, 2012). Social planned, structured, repetitive, and purposely com-
barriers also exist such as a lack of experts to include pleted to improve physical fitness (Caspersen, Powell, &
the child (Must et al., 2015), and a lack of peer exercise Christenson, 1985). The secondary purpose of this
partners (Obrusnikova & Miccinello, 2012). Finally, study was to analyze the specific characteristics of these
community barriers, including a lack of opportunities interventions (e.g. type of intervention, environment,
and high cost may also hinder participation (Must duration) in order to further understand why an inter-
et al., 2015). vention may or may not be effective. In analyzing these
Researchers have long sought to understand the effect
practices and methods of physical activity, it is the
of exercise on a plethora of outcomes for individuals
hope of authors to give a variety of practitioners (e.g.,
with ASD, such as motor skills, social skills, stereotypi-
physical educators, coaches, and recreation leaders)
cal behaviors, self-stimulating behaviors, and physical
tools to utilize in order to aid young individuals with
fitness—see Young and Furgai (2016) for a review of
ASD to achieve the benefits of exercise.
exercise interventions for individuals with ASD. Previ-
ous meta-analysis’ (Sam, Chow, & Tong, 2015; Sowa &
Meulenbroek, 2012) have sought to quantify the effect Method
of exercise, with the aim of providing recommenda- Search Strategy
tions to practitioners. Sam, Chow, and Tong analyzed
Following PRISMA guidelines (Moher, Liberati, Tetzlaff,
eight studies examining three outcome measures; physi-
Altman, & Prisma Group, 2009), search strategies for
cal fitness, exercise mastery, and social competence;
this study were developed around several keywords
moderate to large effects were noted for the latter two
determined by the authors. The main keywords utilized
outcomes, d 5 0.57 and d 5 0.58 respectively. Sowa and
Meulenbroek (2012) examined the effects of exercise in the article searches included the following: autism
interventions—categorized as either group or individ- spectrum disorder, autism, ASD, physical education, physical
ual–on the areas of motor, social, and communication activity, adapted physical education, adapted physical activ-
skills. Pooled results demonstrated an overall improve- ity, exercise, and evidence-based practices. Combinations
ment of 37.5% and found that both motor skills and of these keywords identifying the condition (ASD, etc.),
social skills individual programs yielded medium effect setting/context (Physical education, sport, and physical
sizes, outperforming group interventions (r 5 20.31 and activity), and design (intervention, etc.) were inserted
r 5 20.62 respectively). These meta-analyses have pro- into searches on several academic journal databases
vided a valuable insight into outcomes and moderators that include SPORTDiscus, ERIC, PsychINFO, PubMed/
of exercise in individuals with ASD. However, these Medline, Child Development and Adolescent Studies, and
studies have not provided a current (since 2014) syn- PsychARTICLES. To identify unpublished work—to
thesis of the literature; five studies have been published examine publication bias – thesis and dissertations were
since 2014 examining the effect of exercise on individu- searched for using ProQuest. In order to determine
als with ASD (e.g., Gabriels, 2015; Riggenback, 2015; whether articles saved from the initial searches were rel-
Putetti, 2016). Furthermore, published meta-analyses evant, a three-stage screening process was implemented.
on the topic do not provide a comprehensive analysis In stage 1, two authors conducted the initial searches
of outcomes and moderators of exercise interventions; by dividing search engines and utilizing the list of key-
for example, previous meta-analyses did conduct com- words developed for this analysis. If the title appeared
prehensive moderator analyses on intervention charac- relevant to the context of the study, the author saved
teristics, participant characteristics, and study the citation of the article to a citation program

2 Healy et al./Physical activity interventions INSAR

(EndNote X7) and after completing the initial screening including children without ASD, or specialized; includ-
all duplicates were removed. In stage 2, two authors ing only children with ASD), type of training held by
independently screened the abstracts of articles saved practitioners (adapted physical education, physical edu-
to the citation database and if the abstract did not pro- cation, medical, or other), outcomes measured (i.e., psy-
vide sufficient information related to the inclusion cri- chomotor, cognitive, or affective), severity of ASD
teria or appeared to be unavailable it was excluded (mild, moderate, severe) support (i.e., whether parental
from the study. In stage 3, two authors independently support was reported in study protocol or not), location
retrieved the remaining articles in full-text form for fur- (i.e., rural vs. urban area), environment (i.e., physical
ther screening; if the articles did not meet inclusion cri- activity, physical education, or sport setting). Participant
teria, they were excluded from the study. If a study did information included age range, gender, country, and
not provide sufficient data during review of full-texts sample size. Study features included the measurement
the lead author was contacted requesting missing infor- tool used (i.e., objective or self-report).
mation. A follow-up email was sent two weeks later and Two coders independently reviewed and reported
after one month the study/paper was excluded when codes for each of the studies meeting inclusion criteria;
no response was received. Disagreements between these codes were then examined by a third coder, who
authors during stage two or three were further discussed also looked at any discrepancies between the first two
with a third author until consensus was reached. coders. Coding results were compared and analyzed
using agreement rates and an inter-rater reliability coef-
Inclusion and Exclusion Criteria
ficient. Prior to the statistics being calculated, discrep-
Several inclusion criteria were implemented to deter- ancies between study codes were reviewed and classified
mine which articles the authors would save during the as factual or interpretative. Factual errors were consid-
initial screening. The inclusion criteria for this study ered transcription errors where the correct answer was
were as follows: (a) the study took place in a physical present in the study and either missed by the coder or
education (PE), physical activity (PA), or sport (S) set- inaccurately reported. Interpretative errors were consid-
ting; (b) the participants of the study were two to ered to be errors where study information was inferred
twenty-two (2–22) years of age; (c) the study imple- or not clear and required the coder to make an interpre-
mented a method or intervention (d) the study tation on the classification. All factual errors were cor-
included a quantifiable outcome measure; (e) the popu- rected; interpretative errors were reviewed by a third
lation of the study was determined to have ASD; (f) the author and a simple majority decision determined the
study was written in the English language; and (g) the appropriate code.
study was published after the year 1970. Outliers and Publication Bias
Definitions of Settings Outliers. Outliers are studies determined to be two
To define the settings extracted from included studies, standard deviations above or below the overall mean
a physical education (PE) setting was determined as effect of the meta-analysis. Studies deemed as outliers
activity taking place in an educational setting during had relative residual scores (z-scores) outside the
school hours. Sport (S) settings were described as tradi- ninety-fifth percentile of the mean effect score (z
tional team or individual, organized, sport-specific score  6 1.96). Outliers were present and a sensitivity
gameplay settings that occur outside of educational set- analysis (“one study removed”) was utilized to examine
tings. Physical Activity (PA) settings were defined as the impact of retention/removal of outliers (studies)
activity taking place outside of an educational setting and their influence on the overall effect score (Green-
or a sport-specific based setting (e.g., recreational activi- house & Ivengar, 1994). Outliers were retained if results
remained significant (P < 0.05) and within the 95% con-
ties such as walking, hiking, playing at a playground, or
fidence interval.
lab-based activity).
Coding and Data Extraction
Publication bias. In order to assess publication bias,
Coding and data extraction forms were developed using three separate procedures were used to limit the impact
established protocols (Brown, Upchurch, & Acton, of publication bias. A comprehensive search of the liter-
2003; Wilson & Lipsey, 2001). Study information was ature was conducted to locate published and unpub-
separated into three categories: Intervention, Participant, lished research. Literature included in the study is
and Study features. Intervention information extracted calculated in a funnel plot by the standard error (y-axis)
from each article included study design (i.e., experi- and effect size (x-axis) to determine if the plot is bal-
mental or quasi-experimental), duration (<10 weeks, anced. Funnel plots are either symmetrical or asymmet-
10–16 weeks, and >16 weeks), setting (i.e., inclusive; rical (Duval & Tweedie, 2000a,b) and if the funnel plot

INSAR Healy et al./Physical activity interventions 3

is found to be asymmetrical, a “Trim and Fill” proce- based on characteristics of study outcomes. For exam-
dure is used to determine and adjust an effect size by ple, body composition was measured by body mass
the number of studies needed to balance the plot. index (BMI), skinfolds, and waist circumference each
Finally, a ‘Fail-Safe N’ calculation was used to determine provided estimates of body fat. As a result, the authors
the number of missing studies needed to nullify signifi- attempted a thematic organization to reflect variable
cant results (Duval & Tweedie, 2000a; Pearson et al., constructs. These constructs included (a) body composi-
2014; Rosenthal, 1979). tion, (b) muscular strength/endurance, (c) cardiovascu-
Effect Size Calculations lar endurance, (d) locomotor skills, (e) manipulative
skills, (f) skill-related fitness, and (g) social functioning.
Comprehensive Meta-Analysis (CMA) version-2 software
was utilized to calculate effect size statistics (Borenstein,
Hedges, Higgins, & Rothstein, 2005). Hedges g was the Results
effect size metric selected as the number of studies
The main purpose of the current study was to deter-
(k < 20) in different analyses (moderator and outcomes)
mine the overall effectiveness across multiple outcomes
were smaller and used to correct for an overestimate
and moderators of physical activity interventions on
the effect size (Hedges, 1981). Data extracted from
young individuals with ASD. There was a total of 29
included studies uses mean (M), sample size (N), and
standard deviation (SD) as the primary methods for studies with 30 independent samples that included
effect size calculations. When these data were not avail- 1,009 participants meeting inclusion criteria. Figure 1
able, F-values, t-values, and/or P-values were extracted provides an overall search strategy and article screening
from each study (Rosenthal, 1994). A random-effects process, while Table 1 displays the coded methodologi-
approach was used to model error for the current meta- cal, participant, and study features for each study as
analysis (Borenstein, Hedges, Higgins, & Rothstein, well as each study’s overall treatment effect. Six studies
2009). A random effects model uses both sampling error included during the screening process provided insuffi-
and between study variance to estimate the effect size. cient data and when authors failed to respond to
Also, when several outcomes were extracted, the study requests all papers were excluded. Analysis of the coder
was the unit of analysis and used a procedure was used agreement determined reliability was acceptable
that averaged the outcomes for a single effect size calcu- (j 5 0.84) with 28 total disagreements that included 22
lation (Borenstein, Hedges, Higgins, & Rothstein, 2009). factual errors that were corrected and 6 interpretative
Cohen’s (1988) criteria for small (> 0.20), moderate (> errors that were analyzed by a third coder and cor-
0.50), and large (> 0.80) effect sizes was used to aid the rected. When interpreting the treatment effects,
interpretation of results. Cohen’s (1988) criteria were used for interpretation of
standardized mean differences and summarized effect
Heterogeneity of Variance sizes as small ( 0.20), medium ( 0.50), and large (
Due to expected differences in sampling error and 0.80). Positive effect sizes were interpreted as treatment
between study variance in this review, there was an groups (intervention groups) showing stronger results
assumption that there would be variability in the true than control groups or groups not included in the
effect sizes between studies. Between-study heterogene- interventions or programs. Negative treatment effects
ity was quantified using the Q-value, tau-squared (s2), indicated that the control group or non-intervention
and I-squared (I2) statistics. The Q-value partitions vari- group produced larger outcome results than the inter-
ance and is used to determine if excess variation of vention group.
between study differences exists, tau-squared provides Random Effects Model, Outlier Analyses, and Publication
an estimate of the variation of true effects between Bias
studies, and I-squared is an estimate of variance that
can be explained by moderators (Shadish & Haddock, The average treatment effect for all exercise interven-
2009). Moderator analyses were conducted to examine tion studies was moderate (g 5 0.62; SE 5 0.20; 95%
associations between physical activity interventions and C.I. 5 0.23, 1.01; P < 0.001) across all outcomes and rep-
outcomes (i.e., psychomotor, cognitive, and affective) resented about six-tenths a standard deviation advan-
and the influence of selected demographic and method- tage for treatment groups over control groups. Figure 2
ological characteristics. displays the relevant statistical analyses utilized when
Due to a diversity of outcome measures reported in evaluating the overall effect sizes. Moderator analyses
individual studies and the relatively small number of of characteristics coded for studies were conducted in
studies meeting inclusion criteria, the authors chose to order to further explain the between-study variation
combine outcomes into several different constructs based on a significant heterogeneous distribution

4 Healy et al./Physical activity interventions INSAR

Figure 1. Search Strategy and Article Screening Process. [Color figure can be viewed at wileyonlinelibrary.com]

(QT 5 249.24, P < 0.001; I2 5 88.77) that was indicative and no studies were required to yield non-significant
of between study variation. (P > 0.05) results within these outcomes.
Outcome Analyses
Outliers and Publication Bias
Several outcome analyses that were conducted pro-
One independent sample (Favazza et al., 2013) was duced both positive and negative effects, which ranged
found to be an outlier (z 5 2.54), thus an outlier analy- from g 5 20.18 to g 5 2.76. Outcomes that were positive
sis was conducted through evaluation of residual values for groups included muscular strength/endurance, loco-
and a “one-study removed” procedure was performed. motor skills, manipulative skills, skill-related fitness,
The single effect size was retained in the analysis as and social functioning. The largest positive effects were
results indicated a change (20.21), remaining signifi- found for manipulative skills (k 5 3, g 5 2.76, P < 0.001),
cant (P  0.05) and within the 95% confidence interval. locomotor skills (k 5 6, g 5 1.60, P < 0.001), skill-related
Publication bias was assessed across all constructs of fitness (k 5 12, g 5 1.07, P < 0.001), muscular strength/
outcomes referenced in Table 2 and reported with the endurance (k 5 7, g 5 0.78, P < 0.01), and social func-
‘Fail Safe N’ measurement. Across five outcomes, several tioning (k 5 6, g 5 0.57, P > 0.05). A negative effect size
studies were deemed necessary to produce non- was found for body composition (k 5 5, g 5 20.18,
significant results (Muscular strength/endurance N 5 32; P > 0.05).
Locomotor skills N 5 171; Manipulative skills N 5 162;
Moderator Analyses
Skill-related fitness N 5 271; Social functioning N 5 26).
However, the two outcomes of body composition and Heterogeneity statistics for the random effects model
cardiovascular endurance produced a Fail Safe N of 0, confirmed that there was a heterogeneous (QT 5 249.24,
suggesting that publication bias may have been violated P < 0.05) distribution and that a large level (I2 5 88.77)

INSAR Healy et al./Physical activity interventions 5

 6
Table 1. Study Characteristics Meeting Inclusion Criteria
Duration
Study Design (weeks) Setting Training Outcomes Level N Age (years) Sex Environ-ment Country School Support Loca-tion Mea-sure Effect (g)

Anderson-Hanley, QE <10 S NS C NR 22 12–14.9 B PA US M P NR O 20.07

2011
Arzoglou et al., E <10 SP NS P NR 5 NR PE GER NR NR NR O 0.41
2013
Bahrami et al., E 10–16 SP O A C 15 B PA IR COM NR NR O 20.83
2012
Chan, 2013 E <10 SP M A C 20 B PA C COM NR NR O 0.24
Borgi, 2016 E >16 I O COM NR 28 5–14.9 M PA I COM NR NR O 0.87
Chi-Hua, 2012 E 10–16 SP PE COM M 42 5–11.9 B PE TAI E NP U O 1.46
Dickinson, 2014 E >16 S PE P MOD 100 5–14.9 B PE UK COM NP NR O 0.31
Favazza, 2013 E <10 SP O P NR 233 5–11.9 B PE US E P NR O 3.11
Fragala-Pinkham, QE 10–16 S M P M 16 5–11.9 B PA US E NP NR O 0.07
2008
Fragala-Pinkham, QE 10–16 S M P M 12 5–14.9 B PA US COM P NR O 0.21
2011
Gabriels, 2015 E 10–16 SP M COM NR 116 5–15 B PA US COM NP NR O 0.47
Hilton, 2014 QE 10–16 S M COM NR 14 NR NR PA US COM NP NR O 20.21
Hinckson, 2013 QE 10–16 SP O P NR 17 12–15 B PE NZ COM P NR C 20.06
Koenig et al., N 10–16 SP PE A COM 24 B PE US E NP NR O 0.40
2012
Lanning, 2014 E 10–16 SP O COM NR 13 5–14.9 B PA US COM NP NR O 0.65
Lourenco, 2015 E >16 S O P M 17 5–11.9 B PA AUS E NP NR O 0.21
MacDonald, 2012 E NR S O P NR 42 5–15 B PA US COM NP NR O 0.69
Movahedi, 2013 E 10–16 SP O A COM 15 B PA IR COM NP NR O 21.11
Oriel, 2011 E <10 SP PE C S 9 5–11.9 B PE US E NP NR O 0.67
Pan, 2010 QE >16 S PE COM M 16 5–11.9 NR PA TAI E NP NR O 0.23

Healy et al./Physical activity interventions

Pan, 2011 QE >16 S PE P NR 15 5–11.9 B PA TAI E NP NR O 0.06
Pan, 2016 E 10–16 S PE COM NR 22 5–11.9 NR PA TAI E NP NR O 0.33
Pitetti, 2007 E >16 S O P NR 10 15 B PA US H NP NR O 3.44
Ringenbach, QE <10 SP NS COM COM 10 B PA US COM NP NR O 0.51
2015
Rosenblat, 2011 QE <10 SP M A COM 24 B PA US COM NP NR O 0.78
Schleien, 1990 E NR I APE A NR 34 5–11.9 B PE US E NP NR O 0.885
Weber, 1989 QE <10 SP APE P S 28 5–14.9 M PE US COM NP NR O 2.735
Weber, 1992 QE NR SP APE P NR 12 5–15 M PE US COM NP NR O 2.533
Wuang, 2010 E >16 S M COM NR 60 5–11.9 B PA TAI E NP U O 0.675

Note. Design: QE, quasi-experimental; E, experimental. Duration (Weeks): NR, not reported. Setting: S, study; I, inclusive; SP, specialized. Training: NS, not specified; PE, physical education; APE,
adapted physical education; M, medical; O, other. Outcomes: C, cognitive; P, psychomotor; A, affective; COM, combined. Level: NR, not reported; M, mild; MOD, moderate; S, severe. Gender: NR, not
reported; B, both; M, male. Environment: PA, physical activity; PE, physical education. Country: US, United States; I, Italy; TAI, Taiwan; UK, United Kingdom; NZ, New Zealand; AUS, Austria. School:
M, Middle; E, Elementary; H, High; COM, Combined. Support: P, Parent Support; NP, No Parent Support; NR, Not Reported. Location: U, Urban; NR, Not Reported. Measure: O, Objective; C, Combined.

INSAR
Figure 2. Forest plot for studies meeting inclusion criteria.

of between-study variation existed to justify conducting 10 weeks (k 5 8 g 5 1.06, Z 5 3.00, P < 0.05), conducted
sub-group analyses for coding characteristics. These in specialized settings (k 5 12, g 5 0.7,5 Z 5 2.38,
results indicate that between study variance was not P < 0.05), facilitated by an instructor with adapted phys-
random and could be explained as a result of the confi- ical education training (k 5 3, g 5 1.94, Z 5 2.31,
dence interval overlap. Subgroup results can be impre- P < 0.05), and interventions that focus on psychomotor
cise when there are not a critical number of studies outcomes (k 5 12, g 5 1.21, Z 5 3.24, P < 0.05) produced
(k  5) used in the analysis (Borenstein et al., 2009). The significant within group differences between interven-
authors have selected to report subgroup findings with tion and control groups/conditions. All intervention
imprecise estimates of effects for discussion purposes. categories producing significant within group compari-
Table 3 displays all relevant statistical results from mod- sons displayed a high degree of heterogeneity
erator analyses on intervention characteristics, partici- (QT < 0.05).
pant characteristics, and study characteristics.
Sample Characteristics
Intervention Characteristics
There were no subgroup variables for sample character-
Environment was the only intervention characteristic istics, however, several trends for sample characteristics
to produce significant differences between subgroups. were discovered within samples that included partici-
Physical education environments (k 5 10, g 51.20, pants diagnosed with a ‘severe’ degree of ASD (k 5 2,
Z 5 3.92, P  0.05) produced significant large effects for g 5 1.68, Z 5 2.06, P  0.05), samples including both
outcomes when compared to physical activity environ- males and females (k 5 16, g 5 0.74, Z 5 2.63, P  0.05),
ments (k 5 19, g 5 0.29, Z 5 1.31, P  0.05). No other samples at elementary grade levels (k 5 11, g 5 0.77,
intervention characteristics produced significant differ- Z 5 2.53, P  0.05), and samples from the US (k 5 16,
ences between subgroups, however, several trends were g 5 0.97, Z 5 3.49, P  0.05). There was also a large vari-
apparent due to within group comparisons. Studies ability within subgroups as indicated by Q and s2 values
employing experimental designs (k 5 17, g 5 0.67, with potential to explain variance between studies
Z 5 2.67, P < 0.05), employing interventions less than (I2 > 70).

INSAR Healy et al./Physical activity interventions 7

Table 2. Outcome Analysis
Effect size statistics Heterogeneity statistics
Null test Publication bias
2
Variable k g SE s 95% C.I. Z Q s2 I2 Fail Safe N

Body Composition 5 20.18 0.15 0.02 (-0.465, 0.103) 21.25 1.53 0 0 0

Muscular Strength/Endurance 7 0.78 0.29 0.08 (0.223, 1.344) 2.74* 16.98* 0.37 64.66 32
Cardiovascular Endurance 5 0.10 0.30 0.09 (-0.480, 0.686) 0.35 13.42* 0.29 70.20 0
Locomotor Skills 6 1.60 0.56 0.31 (0.516, 2.693) 2.89* 50.04* 1.59 90.01 171
Manipulative Skills 3 2.76 0.85 0.72 (1.099, 4.413) 3.26* 23.09* 1.92 91.34 162
Skill Related Fitness 12 1.07 0.52 0.27 (0.054, 2.082) 2.06* 234.07* 2.98 95.38 271
Social Functioning 6 0.57 0.30 0.09 (-0.023, 1.153) 1.88 26.55* 0.41 81.16 26

Note. K, number of effect sizes; g, effect size (Hedges g); SE, standard error; s2, variance. 95% C.I., confidence intervals (lower limit, upper
limit); Z, test of the null hypothesis; s2, between-study variance in random effects model; I2, total variance explained by moderators. *P  0.05.

Study Characteristics Delays of motor milestones and atypical fine and gross
motor patterns are consistently reported among this
No significant differences between subgroups were population (Lloyd, Macdonald, & Lord, 2013) and have
found for study characteristics. Given the limited num- been suggested as a core feature of ASD (Lee & Bo, 205).
ber of studies no subgroup comparisons can be made. To help alleviate these deficits, practitioners must be
Smaller subgroups within the study location (urban, k able to depend on the literature base for clearly-
53), study measures (combined reporting measures, defined, theoretically-sound intervention strategies.
k 5 1), and publication status (unpublished studies, Favazza et al. (2013) demonstrates well the structure of
k 5 0) prevent precise estimates of effect size. the intervention, instructional approaches, and training
used to effect motor skill improvements in the pre-
school aged children in a Young Athletes Program. In
Discussion
addition, Favazza et al. delineated the theoretical foun-
dation—Lerner’s and Clark’s theories of motor develop-
The purpose of this study was to assess the effect of
ment and Newell’s theory of motor acquisition—for the
physical activity interventions on youth with ASD.
context and instructional approach, and described fidel-
Results indicated an overall moderate-positive effect for
ity measures applied. These details are sparse among
participants exposed to physical activity interventions,
research on interventions focused on locomotor and/or
particularly for interventions targeting the development
manipulative outcomes—interventions were largely
of manipulative skills, locomotor skills, skill-related
atheoretical. Future research should seek to overcome
fitness, social functioning, and muscular strength/
such shortcomings of past research for the refinement
endurance.
of replicable interventions. Research should also con-
Manipulative and Locomotor Skills sider examining the sustainability of motor gains and
the consequence of changes in motor performance on
Reflecting the positive findings in previous meta- physical activity levels.
analyses (Sam, Chow, & Tong, 2015; Sowa & Meulen-
broek, 2012), in the current study interventions focused Skill-Related Fitness
on the development of manipulative and locomotor The category of skill-related fitness, encapsulated a vari-
skills were demonstrated to have a large positive effect ety of outcomes including balance, body coordination,
(g  0.80). A variety of intervention types were exam- visual motor control, mobility skills, and response
ined, including the Young Athletes program (YAP) speed; skills that have been previously associated with
(Favazza et al., 2013), trampoline training (Lourenço, physical activity participation among TD youth (e.g.,
Esteves, Corredeira, & Seabra, 2015), stimulated horse- speed and agility; Wrotniak, Epstein, Dorn, Jones, &
riding program (Wuang, Wang, Huang, & Su, 2010), Kondilis, 2006). This category of outcomes was shown
and task variation/constant task methods (Weber & to be greatly affected for experimental groups by the
Thorpe, 1989, 1992). Improvements in locomotor and implementation of physical activity interventions
manipulative skills for this population are particularly (g  0.80). Intervention modalities utilized included a
important as poor motor skills have been revealed as a computer-based activity program (Dickinson & Place,
significant barrier to physical activity participation 2014), exergaming (Hilton et al., 2014), trampolining
among youth with ASD (Must et al., 2015), and, con- (Laurenco et al., 2015; Giagazoglou 2013), physical
versely, a predictor of activity levels among TD children training (Pan, 2016), and a Simulated Developmental
(Barnett, Van Beurden, Morgan, Brooks, & Beard, 2009). Horse-Riding Program (SDHRP) (Wuang et al., 2010).

8 Healy et al./Physical activity interventions INSAR

Table 3. Moderator Analysis
Effect size statistics Heterogeneity statistics
Null test
2
k G SE s 95% C.I. Z Q s2 I2

Random Effects Model A 29 0.62 0.20 0.04 (0.227, 1.010) 3.10* 249.24* 0.96 88.77
B
Intervention Characteristics
Design 0.97B
Experimental 17 0.67 0.26 0.07 (0.160, 1.182) 2.57* 203.22* 1.24 92.13
Quasi 12 0.54 0.31 0.10 (20.070, 1.158) 1.74 36.13* 0.37 69.56
Duration 5.35B
<10 Weeks 8 1.06 0.35 0.13 (0.366, 1.753) 3.00* 131.55* 2.09 94.68
10 to 16 11 0.12 0.30 0.09 (20.474, 0.711) 0.40 31.62* 0.27 68.38
>16 Weeks 7 0.64 0.30 0.16 (20.139, 1.415) 1.60 10.75 0.14 44.16
Not Reported 3 1.23 0.59 0.36 (0.061, 2.405) 2.06* 4.74 0.25 57.80
Setting 1.06B
Inclusive 2 1.15 0.75 0.56 (20.313, 2.612) 1.54 1.31 0.04 23.90
Specialized 12 0.75 0.31 0.10 (0.131, 1.359) 2.38* 203.92* 2.26 94.61
Study Designed 15 0.44 0.28 0.08 (20.117, 0.996) 1.54 16.35 0.02 14.40
Training 3.82B
APE 3 1.94 0.67 0.45 (0.634, 3.245) 2.91* 13.76* 1.30 85.47
PE 7 0.49 0.48 0.23 (20.425, 1.446) 1.18 9.76 0.01 24.59
Multiple 2 0.51 0.76 0.58 (20.978, 2.005) 0.68 1.36 0.05 36.91
Medical 6 0.28 0.48 0.23 (20.657, 1.214) 0.58 21.08* 0.74 76.28
Not Specified 3 0.27 0.65 0.42 (21.007, 1.545) 0.41 1.36 0 0
Other 8 0.67 0.39 0.15 (20.329, 1.314) 1.71* 156.46* 2.17 88.77
Random Effects Model A 29 0.62 0.20 0.04 (0.227, 1.010) 3.10* 249.24* 0.96 88.77
Outcomes 3.55 B
Affective 6 0.08 0.42 0.18 (20.743, 0.911) 0.20 33.49* 0.50 85.07
Cognitive 2 0.28 0.75 0.56 (21.187, 1.745) 0.37 1.79 0.12 44.16
Psychomotor 12 1.21 0.32 0.10 (0.410, 1.663) 3.24* 165.94* 2.06 93.37
Combined 9 0.56 0.32 0.13 (20.146, 1.262) 1.33 7.93 0 0
Level 2.65 B
Combined 5 0.22 0.49 0.24 (20.739, 1.185) 0.45 12.33* 0.23 67.55
Mild 5 0.44 0.51 0.26 (20.563, 1.433) 0.85 6.76 0.14 40.80
Moderate 1 0.31 1.06 1.12 (21.772, 2.382) 0.28 1.378 0 0
Severe 2 1.68 0.82 066 (0.079, 3.274) 2.06* 8.60* 1.45 88.37
Not Reported 16 0.71 0.29 0.08 (0.150, 1.268) 2.49* 195.05* 1.89 92.31
Environment 5.67 B*
Physical Activity 19 0.29 0.23 0.05 (-0.147, 0.736) 1.31 42.14* 0.20 57.29
Physical Education 10 1.20 0.31 0.09 (0.599, 1.799) 3.92* 148.65* 1.52 93.95
Sample Characteristics B
Gender 0.79 B
Female & Male 16 0.74 0.27 0.07 (0.208, 1.263) 2.73* 175.44* 1.10 91.45
Male 9 0.59 0.36 0.13 (20.109, 1.296) 1.66 57.07* 0.88 85.98
Not Reported 4 0.19 0.55 0.31 (20.892, 1.274) 0.35 0.89 0 0
School 4.82 B
Elementary 11 0.77 0.30 0.09 (0.174, 1.367) 2.53* 133.91* 1.44 92.53
Middle 1 20.07 0.97 0.95 (21.979, 1.839) 20.07 0 0 0
High 1 3.44 1.41 2.02 (0.666, 6.212) 2.43* 0 0 0
Combined 14 0.51 0.27 0.07 (20.026, 1.041) 1.86 59.74* 0.43 78.24
Not Reported 2 0.09 0.76 0.58 (21.398, 1.581) 0.12 0.88 0 0
Random Effects Model A 22 0.81 0.238 0.057 (0.342, 1.276) 3.40* 200.55* 1.04 89.53
Support 0.42 B
No Parents 24 0.56 0.21 0.04 (0.152, 0.961) 2.70* 78.83* 0.32 70.83
Not Reported 1 0.87 0.99 0.98 (21.064, 2.812) 0.88 0 0 0
Parents 4 0.87 0.49 0.24 (20.084, 1.827) 1.79 116.17* 3.79 97.42
Country 6.79B
Austria 1 0.21 1.12 1.25 (21.978, 2.401) 0.19 0 0 0
China 1 0.24 1.06 1.11 (21.826, 2.309) 0.23 0 0 0
Greece 1 0.41 1.16 1.35 (21.865, 2.692) 0.36 0 0 0
Iran 2 20.97 0.76 0.58 (22.461, 0.520) 21.26 0.26 0 0
Italy 1 0.87 1.08 1.17 (21.248, 2.996) 0.81 0 0 0
New Zealand 1 20.06 1.07 1.15 (22.154, 2.041) 20.05 0 0 0
Taiwan 5 0.55 0.50 0.25 (20.418, 1.526) 1.12 5.74 0.09 30.33

INSAR Healy et al./Physical activity interventions 9

Table 3. Continued
Effect size statistics Heterogeneity statistics
Null test
2
k G SE s 95% C.I. Z Q s2 I2

United Kingdom 1 0.31 1.03 1.06 (21.709, 2.319) 0.30 0 0 0

United States 16 0.97 0.28 0.08 (0.423, 1.506) 3.49* 176.58* 1.20 91.51
B
Study Characteristics
Location 0.12 B
Not Reported 26 0.60 0.22 0.05 (0.173, 1.018) 2.76* 245.32* 1.06 89.81
Urban 3 0.83 0.63 0.39 (20.505, 2.639) 1.32 3.92 0.09 30.78
Measure 0.43 B
Combination 1 20.06 1.06 1.11 (22.124, 2.011) 20.05 0 0 0
Objective 28 0.65 0.20 0.04 (0.243, 1.047) 3.14* 244.72* 0.96 88.77

Note. K, number of effect sizes; g, effect size (Hedges g); SE, standard error; s2, variance; 95% C.I., confidence intervals (lower limit, upper
limit); Z, test of the null hypothesis; s2, between study variance in random effects model; I2, total variance explained by moderators; A, total Q-value
used to determine heterogeneity; B, between study Q-value used to determine significance (a 5 0.05). *P < 0.05.

Deficits in skill-related fitness—for example, related to physical activity to encourage appropriate play behavior
postural stability (Molloy, Dietrick, & Bhattacharya, (Schlein et al., 1990) and promote interactions with
2003), body coordination, and agility (Pan, 2012)—are peers, siblings and instructors (Chia-Hua, 2012); and,
prevalent among youth with ASD. To prevent skill- specifically related to equine interventions, the forma-
related fitness levels from impeding physical activity tion of relationships with instructors and horses, and
participation, and thus, the attainment of the related the ability of the animals to positively engage people
health and social benefits, early intervention is crucial. thus counteracting social withdrawal (Borgi, 2016; Lan-
Future research should continue to refine the structure, ning, 2014). Future research should continue to identify
and dose, of early interventions focused on developing the ‘active ingredient’ (O’Haire, 2013) of physical activ-
skill-related fitness, and seek to examine the impact on ity interventions that aim to develop the social domain
physical activity participation. Curriculum-based of youth with ASD. Fidelity measures should also be
instruction and assessment should be considered when collected. Randomized control trials, with a comparable
appropriate to ensure delivery of interventions that are control group, will be key to disentangling the factors
goal-orientated, structured, and progressive (Pan, 2011). contributing to positive outcomes. Intervention charac-
teristics that showed a trend for producing differences
Social Functioning
between subgroups in this meta-analysis, that may be
Due to the social deficits at the core of ASD, research- worthy of examination in future exercise interventions
ers, including in the field of adapted physical activity, to impact the social domain, include the training of the
have long tried to positively impact development in intervention facilitator; a trained facilitator, skilled in
this domain. In the current meta-analysis multiple stud- fostering an environment that promotes social interac-
ies were included that measured outcomes in the social tion, may be key for success for development of social
functioning category (which encompassed outcomes skills.
related to factors such as social communication, adap- Muscular Strength and Endurance
tive functioning, and appropriate play behaviors). Strat-
egies utilized included horseback riding (Borgi et al., Youth with ASD are demonstrated to have lower mus-
2016; Gabriels et al., 2016; Lanning, Baier, Ivey-Hatz, cular strength and endurance that their TD counter-
Krenek, & Tubbs, 2014), various types of group play parts (Pan, 2014; Tyler, MacDonald, & Menear, 2014).
(Schleien, Rynders, Mustonen, & Fox, 1990), running/ It was significant, therefore, that within the outcome of
jogging programs (Oriel, George, Peckus, & Semon, muscular strength and endurance, experimental groups
2011), and exergaming interventions (Anderson-Han- outperformed control groups to a large effect
ley, Tureck, & Schneiderman, 2011). The outcome anal- (g 5 0.818). Studies measuring these outcomes utilized a
ysis indicated that social functioning in young number of intervention modalities, such as Nintendo
individuals with ASD was moderately influenced by the Wii exergaming (Dickinson & Place, 2014), aquatic
implementation of physical activity interventions exercise programs (Fragala-Pinkham, Haley, & O’Neil,
(g 5 0.57); similar to previous meta-analyses (Sowa & 2008; Fragala-Pinkham, Haley, & O’Neil, 2011; Pan,
Meulenbroek, 2012). Various factors are postulated as 2011), and horse riding programs (Wuang et al., 2010).
contributing to the social development that occurs in The research in this area highlights some challenges,
physical activity settings, including the nature of and areas in need for future study; for example, Fragala-

10 Healy et al./Physical activity interventions INSAR

Pinkham et al. (2011) discuss that the 14-week program detail on the role of parents, in physical activity inter-
may not have been of sufficient duration to show ventions for youth with ASD. Parents may also have a
changes; perhaps the longer duration of other interven- role to increase sustainability and scalability of physical
tions that showed to positively affect muscular strength activity interventions for youth with ASD.
and/or endurance contribute to their success (e.g., 20- Finally, the moderator analysis indicated that a dura-
week intervention: Wuang et al, 2010; one-year inter- tion of up to 16 weeks had a significant influence on
vention: Dickinson, 2014). Future research should seek the performance of experimental groups (versus inter-
to identify the dosage required for significant, sustained ventions of more than 16 weeks). Further exploration is
improvement in muscular strength and endurance. Fur- required to fully understand why a shorter duration
thermore, Fragala-Pinkham et al. (2008) makes refer- may influence the effect of physical activity interven-
ence to the challenges of motivating the participants to tions among youth with ASD. Perhaps shorter interven-
increase resistance. Future research should clearly define tions encourage a higher degree of engagement in the
instructional and motivation methods use for the bene- intervention, higher fidelity of the intervention imple-
fit of practitioners and other researchers. mentation, or, perhaps, the tendency for shorter inter-
ventions to be more frequent and intense? Identifying
Moderator Effects
an optimum duration for physical activity interventions
Moderator analyses were conducted on intervention for provided youth with ASD with sustainable results is
characteristics, participant characteristics, and study necessary. In doing so, researchers must be cognizant
characteristics overcoming limitations of previous meta- that factors including the anticipated outcome (e.g.,
analyses on this topic (Sam, Chow, & Tong, 2015; Sowa the development of social skills may take longer than
& Meulenbroek, 2012). The moderator analysis indi- the development of motor skills), the nature of the
cated that specialized classes (i.e., classes consisting of intervention (for example exercise intensity), and sever-
students with ASD only) had a significant influence on ity levels of the participants will be influential. Further-
the performance of the experimental groups. Special- more, the examination of the effect of interventions of
ized classes are designed to help practitioners modify varying durations should include sustained results.
the environment to suit the specific needs of individual
students, and this ability to modify has been shown to Future research. This meta-analysis provides a com-
be an influential component to the success of students prehensive synthesis of the literature base on physical
with ASD within a physical education context (Hamil- activity interventions involving individuals with ASD.
ton, 2006). The moderator analysis also indicated a sig- The potential for physical activity interventions to
nificant influence of the intervention facilitator having effect change in a variety of domains has been rein-
training in APE on the overall effect of the intervention forced. To continue the trajectory of increasingly
(reported in three studies). It has been suggested (Jones impactful experimental investigation of interventions
et al., 2017) that due to complexities of ASD and the for youth with ASD, it is worthy to reflection on oppor-
prevalence of ASD comorbidities that may affect physi- tunities for improving the rigor of evidence underling
cal activity participation, specialized facilitators are physical activity as an intervention modality. Here, we
most suited to implementing interventions. In addi- will outline four suggestions for future research; (a)
tion, the moderator analysis indicated that studies not Interventions were largely theoretical in nature; the
utilizing parental support within the study protocols dearth of theory masks the causal factors underlying
had the greatest influence on the performance of exper- behavior change. Future intervention research should
imental groups. This is an interesting finding, as the lit- strive to clearly delineate how the intervention’s com-
erature suggests that involving parents as support ponents are underpinned by sound theory, to allow for
personnel in physical activity for young individuals reproducibility of research, and ultimately achieve real-
with ASD is a vital component to their child’s success; world impact and exportability of the findings. (b)
including parents allows them to be advocates for their Fidelity of the intervention implementation should be
children, increases collaborative efforts, and can assessed; to influence future research and practice,
decrease teaching challenges for practitioners (Obrusni- researcher need to demonstrate that the intervention
kova & Dillon, 2011; An, 2011). It is possible that the was implemented as planning; the intervention mecha-
significant findings within this moderator were due to nism was delivered as intended. Only then can one
the fact that only four of the 29 total studies reported infer, in confidence, the process-outcome linkage pre-
any parental involvement; perhaps many authors may sented in the research. (c) Randomized control trials,
have overlooked reporting the fact that they specifically with a comparable control group will be key to disen-
chose to not include parents in the study protocols. tangling the factors contributing to positive outcomes
Future research should seek to include, and provide in interventions. A true experimental design allows

INSAR Healy et al./Physical activity interventions 11

researchers to make a clear and sensible interpretation current study it further highlights the importance of
of whichever variables they are attempting to detect the need for additional high-quality research and pub-
within their research, whereas quasi-experimental lishing sufficient data to allow for such analyses.
designs are subject to internal validity as participants
may not be comparable at baseline (Quinn & Keough,
Conclusion
2002). For future researchers to make clearer interpreta-
tions about the effects of physical activity interventions
The physical activity interventions examined in this
on young individuals with ASD, it is recommended that
study have been shown to have an overall moderate
true experimental designs be employed in order to
effect (g 5 0.62). Specifically, moderate to large positive
ensure participants are at comparable levels prior to the
effects were revealed for participants exposed to inter-
implementation of an intervention. Furthermore, a
ventions targeting the development of manipulative
comparable control group, consisted of individuals with
skills, locomotor skills, skill-related fitness, social func-
ASD of similar severity should be used. (d) The modera-
tioning, and muscular strength and endurance. For
tor analysis was to attempt to account for differences in
future replication of studies, and refinement and imple-
the size of the effect that was meta-analyzed. Due to
mentation of effective physical activity interventions
the limitations of the current meta-analysis (e.g., the for youth with ASD, several recommendations for
heterogeneity of studies in the analysis) the moderator future research can be made; increased attention should
analysis should be interpreted with caution. Neverthe- be paid to utilizing true experimental designs, applica-
less, several interventions factors emerged as significant; tion and delineation of a sound theoretical foundation,
factors that should be detailed, and when possible iso- and rich details of intervention structure. Future
lated in future research, to understand their true mod- research should also examine sustainability of interven-
erating influence. These factors included the use of tion effects.
specialized settings, a trained intervention facilitator,
the inclusion of parent support, and a duration of less
than sixteen weeks. References
American Psychiatric Association (2000). Diagnostic and Statis-
Limitations. Synthesizing studies into one large, tical Manual of Mental Disorders DSM-IV-TR (Text Revi-
sion) Washington, D.C.: American Psychiatric Association.
comprehensive and critical statistical analysis, the cur-
An, J. (2011). Exploring the meaning of parental involvement
rent findings can help to clarify the current state of the
in physical education for students with developmental dis-
literature. However, relatively little research has been abilities (Doctoral dissertation, The Ohio State University).
conducted on the effects of physical activity interven- Anderson-Hanley, C., Tureck, K., & Schneiderman, R.L. (2011).
tions on individuals with ASD and much of what has Autism and exergaming: effects on repetitive behaviors and
been published lacks scientific rigor. The current meta- cognition. Psychology Research and Behavior Management,
analysis did not complete a methodical quality assess- 4, 129.
ment of studies included; this is recommended for Arzoglou, D., Tsimaras, V., Kotsikas, G., Fotiadou, E.,
future reviews of the literature. Furthermore, it is rec- Sidiropoulou, M., Proios, M., & Bassa, E. (2013). The effect
ommended that future researchers use Medical Subject of [alpha] tradinional dance training program on neuro-
muscular coordination of individuals with autism. Journal
Headings (MeSH) search terms. Problematic to the cur-
of Physical Education and Sport, 13, 563.
rent analysis was the selection of control groups as
Ayvazoglu, N.R., Kozub, F.M., Butera, G., & Murray, M.J.
many studies chose to use TD populations for compari- (2015). Determinants and challenges in physical activity
son. Furthermore, the frequency and variability of out- participation in families with children with high function-
comes (and measurement of outcomes) reported are ing autism spectrum disorders from a family systems per-
both inconsistent and inconclusive. Physical activity spective. Research in Developmental Disabilities, 47, 93–
interventions have been found to improve physical, 105.
cognitive, emotional, and social outcomes and in the Bahrami, F., Movahedi, A., Marandi, S.M., & Abedi, A. (2012).
current investigation limited information was reported Kata techniques training consistently decreases stereotypy
for cognitive, emotional, and social outcomes. As in children with autism spectrum disorder. Research in
Developmental Disabilities, 33, 1183–1193.
research continues to improve and more sophisticated
Bandini, L., Gleason, J., Curtin, C., Lividini, K., Anderson, S.,
methods and measures are developed results of a future
Cermak, S., Maslin, M., & Must, A. (2013). Comparison of
meta-analysis regarding the effects of physical activity physical activity between children with autism spectrum
on individuals with ASD may result in different find- disorders and typically developing children. Autism, 17,
ings. Another limitation of this study is the small num- 44–54.
ber of studies with sufficient data to be included in a Barnett, L.M., Van Beurden, E., Morgan, P.J., Brooks, L.O., &
meta-analysis. While this problem is not unique to the Beard, J.R. (2009). Childhood motor skill proficiency as a

12 Healy et al./Physical activity interventions INSAR

 predictor of adolescent physical activity. Journal of Adoles- Chu, C.H., & Pan, C.Y. (2012). The effect of peer-and sibling-
cent Health, 44, 252–259. assisted aquatic programon interaction behaviors and
Bauman, M.L. (2010). Medical comorbidities in autism: chal- aquatic skills of children with autism spectrum disorders
lenges to diagnosis and treatment. Neurotherapeutics, 7, and their peers/siblings. Research in Autism Spectrum Dis-
320–327. orders, 6, 1211–1223.
Beamer, J.A., & Yun, J. (2014). Physical educators beliefs and Cohen, J. (1988). Statistical power analysis for the behavioral
self-reported behaviors toward including students with sciences. Hilsdale. NJ: Lawrence Earlbaum Associates, 2.
autism spectrum disorder. Adapted Physical Activity Quar- Croen, L.A., Zerbo, O., Qian, Y., Massolo, M.L., Rich, S.,
terly, 31, 362–376. Sidney, S., & Kripke, C. (2015). The health status of adults
Berkeley, S., Zittel, L., Pitney, L., & Nichols, S. (2001). Locomo- on the autism spectrum. Autism, 19, 814–823.
tor and object control skills of children diagnosed with Curtin, C., Anderson, S.E., Must, A., & Bandini, L. (2010). The
autism. Adapted Physical Activity Quarterly, 18, 405–416. prevalence of obesity in children with autism: a secondary
Bieberich, A.A., & Morgan, S.B. (2004). Self-regulation and data analysis using nationally representative data from the
affective expression during play in children with autism or National Survey of Children’s Health. BMC Pediatrics, 10,
Down syndrome: A short-term longitudinal study. Journal 11.
of Autism and Developmental Disorders, 34, 439–448. Curtin, C., Bandini, L.G., Perrin, E.C., Tybor, D.J., & Must, A.
Bo, J., Lee, C.M., Colbert, A., & Shen, B. (2016). Do children (2005). Prevalence of overweight in children and adoles-
with autism spectrum disorders have motor learning diffi- cents with attention deficit hyperactivity disorder and
culties? Research in Autism Spectrum Disorders, 23, 50–62. autism spectrum disorders: a chart review. BioMed Central
Boddy, L., Downs, S., Knowles, Z., & Fairclough, S. (2015). Pediatrics, 5, 1.
Physical activity and play behaviours in children and young Dickinson, K., & Place, M. (2014). A randomised control trial
people with intellectual disabilities: A cross-sectional obser- of the impact of a computer-based activity programme
upon the fitness of children with autism. Autism Research
vational study. School Psychology International, 36, 154–
and Treatment, 2014,
171.
Dillon, S.R., Adams, D., Goudy, L., Bittner, M., & McNamara,
Borenstein, M., Hedges, L., Higgins, J., & Rothstein, H. (2005).
S. (2016). Evaluating exercise as evidence-based practice for
Comprehensive meta analysis: A computer program for
individuals with autism spectrum disorder. Frontiers in Pub-
research synthesis (version 2.0) [Computer software]. Engle-
lic Health, 4, 290. http://doi.org/10.3389/fpubh.2016.00290
wood, NJ: Biostat.
Duval, S., & Tweedie, R. (2000a). A nonparametric “trim and
Borenstein, M., Hedges, L., Higgins, J., & Rothstein, H. (2009).
fill” method accounting for publication bias in meta-analy-
Introduction to meta-analysis. Hoboken, NJ: John Wiley
sis. Journal of American Statistical Association, 95, 89–98.
Publications.
Duval, S., & Tweedie, R. (2000b). Trim and fill: a simple
Borgi, M., Loliva, D., Cerino, S., Chiarotti, F., Venerosi, A.,
funnel-plot-based method of testing for publication bias in
Braminis, M., et al. (2016). Effectiveness of a standarised
meta-analysis. Biometrics, 56, 455–463.
equine-assisted therapy program for children with autism
Favazza, P.C., Siperstein, G.N., Zeisel, S.A., Odom, S.L., Sideris,
spectrum disorder. Journal of Autism and Developmental
J.H., & Moskowitz, A.L. (2013). Young athletes program:
Disorders, 46, 1–9.
Impact on motor development. Adapted Physical Activity
Bradley, E., & Bolton, P. (2006). Episodic psychiatric disorders
Quarterly, 30, 235–253.
in teenagers with learning disabilities with and without Finkelstein, S.L., Nickel, A., Barnes, T., & Suma, E.A. (2010).
autism. The British Journal of Psychiatry, 189, 361–366. Astrojumper: Designing a virtual reality exergame to moti-
Brown, S.A., Upchurch, S.L., & Acton, G.J. (2003). A framework vate children with autism to exercise. In Virtual Reality
for developing a coding scheme for meta-analysis. Western Conference (VR), 2010 IEEE (pp. 267–268). IEEE.
Journal of Nursing Research, 25, 205–222. Fournier, K.A., Hass, C.J., Naik, S.K., Lodha, N., & Cauraugh,
Cai, S.X., & Kornspan, A.S. (2012). The use of exergaming with J.H. (2010). Motor coordination in autism spectrum disor-
developmentally disabled students, strategies. A Journal for ders: A synthesis and meta-analysis. Journal of Autism and
Physical and Sport Educators, 25, 15–18. Developmental Disorders, 40, 1227–1240.
Case, L., & Yun, J. (2015). Visual practices for children with Fragala-Pinkham, M.A., Haley, S.M., & O’neil, M.E. (2011).
autism spectrum disorders in physical activity. Palaestra, Group swimming and aquatic exercise programme for chil-
29, dren with autism spectrum disorders: a pilot study. Devel-
Caspersen, C.J., Powell, K.E., & Christenson, G.M. (1985). opmental Neurorehabilitation, 14, 230–241.
Physical activity, exercise, and physical fitness: definitions Fragala-Pinkham, M., Haley, S.M., & O’neil, M.E. (2008).
and distinctions for health-related research. Public Health Group aquatic aerobic exercise for children with disabilities.
Reports, 100, 126. Developmental Medicine & Child Neurology, 50, 822–827.
Centers for Disease Control and Prevention. (2016). Childhood Gabriels, R.L., Pan, Z., Dechant, B., Agnew, J.A., Brim, N., &
obesity facts. Available at: http://www.cdc.gov/obesity/data/ Mesibov, G. (2015). Randomized controlled trial of thera-
childhood.html. Accessed on October, 10. peutic horseback riding in children and adolescents with
Chan, A.S., Sze, S.L., Siu, N.Y., Lau, E.M., & Cheung, M.C. autism spectrum disorder. Journal of the American Acad-
(2013). A Chinese mind-body exercise improves self-control emy of Child & Adolescent Psychiatry, 54, 541–549.
of children with autism: A randomized controlled trial. Gillette, M.L.D., Borner, K.B., Nadler, C.B., Poppert, K.M.,
PLoS One, 8, e68184. Stough, C.O., Romine, R.S., & Davis, A.M. (2015).

INSAR Healy et al./Physical activity interventions 13

 Prevalence and health correlates of overweight and obesity Little, L.M., Sideris, J., Ausderau, K., & Baranek, G.T. (2014).
in children with autism spectrum disorder. Journal of Activity participation among children with autism spec-
Developmental & Behavioral Pediatrics, 36, 489–496. trum disorder. American Journal of Occupational Therapy,
Greenhouse, J.B., & Iyengar, S. (1994). Sensitivity analysis and 68, 177–185.
diagnostics. In H. Cooper & L. V. Hedges (Eds.), The hand- Lloyd, M., MacDonald, M., & Lord, C. (2013). Motor skills of
book of research synthesis (pp. 383–98). New York: Russell toddlers with autism spectrum disorders. Autism, 17, 133–
Sage Foundation. 146.
Gresham, F.M., Sugai, G., & Horner, R.H. (2001). Interpreting Lord, C., Cook, E.H., Leventhal, B.L., & Amaral, D.G. (2000).
outcomes of social skills training for students with high- Autism spectrum disorders. Neuron, 28, 355–363.
incidence disabilities. Teaching Exceptional Children, 67, Lord, C., Cook, E.H., Leventhal, B.L., & Amaral, D.G. (2013).
331––344. Autism spectrum disorders. Autism: The Science of Mental
Grether, J., Anderson, M., Croen, L., Smith, D., & Windham, Health, 28, 217.
G. (2009). Risk of autism and increasing maternal and Lourenço, C., Esteves, D., Corredeira, R., & Seabra, A. (2015).
paternal age in a large North American population. Ameri- Children with autism spectrum disorder and trampoline
can Journal of Epidemiology, 170, 1118–1126. training. Wulfenia Journal, 22, 342–351.
Hamilton-Pope, M., & Miller, S. (2006). Teaching physical edu- MacDonald, M., Esposito, P., & Ulrich, D. (2011). The physical
cation to children within the autism spectrum. Texas Asso- activity patterns of children with autism. BioMed Central
ciation for Health, Physical Education, Recreation, and Research Notes, 4, 1.
Dance Journal, 74, 12–14. MacDonald, M., Lord, C., & Ulrich, D.A. (2013). The relation-
Healy, S., Haegele, J.A., Grenier, M., & Garcia, J.M. (2017). ship of motor skills and social communicative skills in
Physical activity, screen-time behavior, and obesity among school-aged children with autism spectrum disorder.
13-year olds in Ireland with and without autism spectrum Adapted Physical Activity Quarterly, 30, 271–282.
disorder. Journal of Autism and Developmental Disorders, Ma€ıano, C. (2011). Prevalence and risk factors of overweight
1–9. and obesity among children and adolescents with intellec-
Healy, S., Msetfi, R., & Gallagher, S. (2013). ‘Happy and a bit tual disabilities. Obesity Reviews, 12, 189–197.
Nervous’: The experiences of children with autism in physi- Mark W, L., & Wilson, D.B. (2001). Practical meta-analysis
cal education. British Journal of Learning Disabilities, 41, (Vol. 49). Thousand Oaks, CA: Sage publications.
222–228. McCoy, S.M., Jakicic, J.M., & Gibbs, B.B. (2016). Comparison
Hilton, C.L., Cumpata, K., Klohr, C., Gaetke, S., Artner, A., of obesity, physical activity, and sedentary behaviors
Johnson, H., & Dobbs, S. (2014). Effects of exergaming on between adolescents with autism spectrum disorders and
executive function and motor skills in children with autism without. Journal of Autism and Developmental Disorders,
spectrum disorder: a pilot study. American Journal of Occu- 1–10.
pational Therapy, 68, 57–65. Memari, A.H., Ghaheri, B., Ziaee, V., Kordi, R., Hafizi, S., &
Hinckson, E.A., Dickinson, A., Water, T., Sands, M., & Moshayedi, P. (2013). Physical activity in children and ado-
Penman, L. (2013). Physical activity, dietary habits and lescents with autism assessed by triaxial accelerometry.
overall health in overweight and obese children and youth Pediatric Obesity, 8, 150–158.
with intellectual disability or autism. Research in Develop- Memari, A.H., Kordi, R., Ziaee, V., Mirfazeli, F.S., & Setoodeh,
mental Disabilities, 34, 1170–1178. M.S. (2012). Weight status in Iranian children with autism
Jahromi, L.B., Bryce, C.I., & Swanson, J. (2013). The impor- spectrum disorders: Investigation of underweight, over-
tance of self-regulation for the school and peer engagement weight and obesity. Research in Autism Spectrum Disorders,
of children with high-functioning autism. Research in 6, 234–239.
Autism Spectrum Disorders, 7, 235–246. Molloy, C.A., Dietrich, K.N., & Bhattacharya, A. (2003). Postural
Jones, R.A., Downing, K., Rinehart, N.J., Barnett, L.M., May, T., stability in children with autism spectrum disorder. Journal of
McGillivray, J.A., et al. (2017). Physical activity, sedentary Autism and Developmental Disorders, 33, 643–652.
behavior and their correlates in children with autism spec- Movahedi, A., Bahrami, F., Marandi, S.M., & Abedi, A. (2013).
trum disorder: A systematic review. PLoS One, 12, Improvement in social dysfunction of children with autism
e0172482. spectrum disorder following long term Kata techniques
Koenig, K.P., Buckley-Reen, A., & Garg, S. (2012). Efficacy of training. Research in Autism Spectrum Disorders, 7, 1054–
the get ready to learn yoga program among children with 1061.
autism spectrum disorders: A pretest–posttest control group Moher, D., Liberati, A., Tetzlaff, J., & Altman, D.G.; Prisma
design. American Journal of Occupational Therapy, 66, Group. (2009). Preferred reporting items for systematic
538–546. reviews and meta-analyses: The PRISMA statement. PLoS
Lang, R., Koegel, L.K., Ashbaugh, K., Regester, A., Ence, W., & Medicine, 6, e1000097.
Smith, W. (2010). Physical exercise and individuals with Must, A., Phillips, S., Curtin, C., & Bandini, L.G. (2015). Bar-
autism spectrum disorders: A systematic review. Research in riers to physical activity in children with autism spectrum
Autism Spectrum Disorders, 4, 565–576. disorders: Relationship to physical activity and screen time.
Lanning, B.A., Baier, M.E.M., Ivey-Hatz, J., Krenek, N., & Journal of Physical Activity and Health, 12, 529–534.
Tubbs, J.D. (2014). Effects of equine assisted activities on O’Haire, M.E. (2013). Animal-assisted intervention for autism
autism spectrum disorder. Journal of Autism and Develop- spectrum disorder: A systematic literature review. Journal of
mental Disorders, 44, 1897–1907. Autism and Developmental Disorders, 1–17.

14 Healy et al./Physical activity interventions INSAR

Obrusnikova, I., & Cavalier, A.R. (2011). Perceived barriers and with severe autism. Journal of Autism and Developmental
facilitators of participation in after-school physical activity Disorders, 37, 997–1006.
by children with autism spectrum disorders. Journal of Quinn, G.P., & Keough, M.J. (2002). Experimental design and
Developmental and Physical Disabilities, 23, 195–211. data analysis for biologists. Cambridge University Press.
Obrusnikova, I., & Dillon, S.R. (2011). Challenging situations Raudenbush, S.W., & Liu, X.F. (2001). Effects of study dura-
when teaching children with autism spectrum disorders in tion, frequency of observation, and sample size on power
general physical education. Adapted Physical Activity Quar- in studies of group differences in polynomial change. Psy-
terly, 28, 113–131. chological Methods, 6, 387.
Obrusnikova, I., & Miccinello, D.L. (2012). Parent perceptions Rimmer, J.H., Braddock, D., & Fujiura, G. (1993). Prevalence of
of factors influencing after-school physical activity of chil- obesity in adults with mental retardation: Implications for
dren with autism spectrum disorders. Adapted Physical health promotion and disease prevention. Mental Retarda-
Activity Quarterly, 29, 63–80. tion, 31, 105.
Oriel, K.N., George, C.L., Peckus, R., & Semon, A. (2011). The Ringenbach, S.D., Lichtsinn, K.C., & Holzapfel, S.D. (2015).
effects of aerobic exercise on academic engagement in Assisted cycling therapy (ACT) improves inhibition in ado-
young children with autism spectrum disorder. Pediatric lescents with autism spectrum disorder. Journal of Intellec-
Physical Therapy, 23, 187–193. tual and Developmental Disability, 40, 376–387.
Pan, C.Y. (2010). Effects of water exercise swimming program Rosenblatt, L.E., Gorantla, S., Torres, J.A., Yarmush, R.S., Rao,
on aquatic skills and social behaviors in children with S., Park, E.R., . . . Levine, J.B. (2011). Relaxation response–
autism spectrum disorders. Autism, 14, 9–28. based yoga improves functioning in young children with
Pan, C.Y. (2011). The efficacy of an aquatic program on physi- autism: A pilot study. The Journal of Alternative and Com-
cal fitness and aquatic skills in children with and without plementary Medicine, 17, 1029–1035.
autism spectrum disorders. Research in Autism Spectrum Rosenthal, R. (1979). The file drawer problem and tolerance for
Disorders, 5, 657–665. null results. Psychological Bulletin, 86, 638–641.
Pan, C.Y. (2014). Motor proficiency and physical fitness in Rosenthal, R. (1994). Statistically describing and combining
adolescent males with and without autism spectrum disor- studies. In H. Cooper, & L. Hedeges (Eds.), The handbook
ders. Autism, 18, 156–165. of research synthesis, 231–244. New York: Russell Sage
Pan, C.Y., Liu, C.W., Chung, I.C., & Hsu, P.J. (2015). Physical Foundation.
activity levels of adolescents with and without intellectual Sam, K.L., Chow, B.C., & Tong, K.K. (2015). Effectiveness of
disabilities during physical education and recess. Research Exercise-Based Interventions for Children with Autism: A
in Developmental Disabilities, 36, 579–586. Systematic Review and Meta-Analysis. Social Behavior, 10,
Pan, C.Y., Tsai, C.L., & Hsieh, K.W. (2011b). Physical activity 11.
correlates for children with autism spectrum disorders in Sandt, D.R., & Frey, G.C. (2005). Comparison of physical activ-
middle school physical education. Research Quarterly for ity levels between children with and without autistic spec-
Exercise and Sport, 82, 491–498. trum disorders. Adapted Physical Activity Quarterly, 22,
Pan, C.Y., Tsai, C.L., Chu, C.H., & Hsieh, K.W. (2011a). Physi- 146–159.
cal activity and self-determined motivation of adolescents Schleien, S.J., Miller, K.D., Walton, G., & Pruett, S. (2014). Par-
with and without autism spectrum disorders in inclusive ent perspectives of barriers to child participation in recrea-
physical education. Research in Autism Spectrum Disorders, tional activities. Therapeutic Recreation Journal, 48, 61.
5, 733–741. Schleien, S.J., Rynders, J.E., Mustonen, T., & Fox, A. (1990).
Pan, C., Tsai, C., Chu, C., Sung, M., et al. (2016). Objectively Effects of social play activities on the play behavior of chil-
measured physical activity and health-related physical fit- dren with autism. Journal of Leisure Research, 22, 317.
ness in secondary school-aged male students with autism Shadish, W.R., & Haddock, CK. (1994). Combining estimates
spectrum disorders. Physical Therapy, 96, 511. of effect size. H. Cooper, & L.V. Hedges (eds.). The hand-
Pearson, N., Braithwaite, R.E., Biddle, S.J.H., Sluijs, E.M.F., & book of research synthesis (pp. 261–281). New York, NY:
Atkin, A.J. (2014). Associations between sedentary behav- Russell Sage Foundation.
iour and physical activity in children and adolescents: A Simonoff, E., Pickles, A., Charman, T., Chandler, S., Loucas, T.,
meta-analysis. Obesity Reviews, 9–10. & Baird, G. (2008). Psychiatric disorders in children with
Petrus, C., Adamson, S.R., Block, L., Einarson, S.J., Sharifnejad, autism spectrum disorders: Prevalence, comorbidity, and
M., & Harris, S.R. (2008). Effects of exercise interventions associated factors in a population-derived sample. Journal
on stereotypic behaviours in children with autism spectrum of the American Academy of Child & Adolescent Psychia-
disorder. Physiotherapy Canada, 60, 134–145. try, 47, 921–929.
Phillips, K.L., Schieve, L.A., Visser, S., Boulet, S., Sharma, A.J., Sowa, M., & Meulenbroek, R. (2012). Effects of physical exer-
Kogan, M.D., . . . Yeargin-Allsopp, M. (2014). Prevalence and cise on autism spectrum disorders: a meta-analysis.
impact of unhealthy weight in a national sample of US ado- Research in Autism Spectrum Disorders, 6, 46–57.
lescents with autism and other learning and behavioral dis- Staples, K.L., & Reid, G. (2010). Fundamental movement skills
abilities. Maternal and Child Health Journal, 18, 1964. and autism spectrum disorders. Journal of Autism and
Pitetti, K.H., Rendoff, A.D., Grover, T., & Beets, M.W. (2007). Developmental Disorders, 40(2), 209–217.
The efficacy of a 9-month treadmill walking program on Tyler, K., MacDonald, M., & Menear, K. (2014). Physical activ-
the exercise capacity and weight reduction for adolescents ity and physical fitness of school-aged children and youth

INSAR Healy et al./Physical activity interventions 15

 with autism spectrum disorders. Autism Research and Treat- Weber, R.C., & Thorpe, J. (1992). Teaching children with
ment, 2014, 1–6. autism through task variation in physical education. Excep-
U.S. Department of Health and Human Services. (1996). A tional Children, 59, 77–86.
report of the surgeon general: Physical activity and health. Wrotniak, B.H., Epstein, L.H., Dorn, J.M., Jones, K.E., &
Springfield, VA: National Technical Information Service. Kondilis, V.A. (2006). The relationship between motor pro-
Ulrich, D.A. (2000). Test of gross motor development-2. Aus- ficiency and physical activity in children. Pediatrics, 118,
tin: Prod-Ed. e1758–e1765.
Warburton, D.E., Nicol, C.W., & Bredin, S.S. (2006). Health Wuang, Y.P., Wang, C.C., Huang, M.H., & Su, C.Y. (2010). The
benefits of physical activity: the evidence. Canadian Medi- effectiveness of simulated developmental horse-riding pro-
cal Association Journal, 174, 801–809. gram in children with autism. Adapted Physical Activity
Weber, R.C., & Thorpe, J. (1989). Comparison of task variation Quarterly, 27, 113–126.
and constant task methods for severely disabled in Young, S., & Furgai, K. (2016). Exercise effects in individuals
physical education. Adapted Physical Activity Quarterly, 6, with autism spectrum disorder: A short review. Autism
338–353. Open Access, 6, 2.

16 Healy et al./Physical activity interventions INSAR