Early Motor Function of Children With Autism Spectrum Disorder: A Systematic Review

In this systematic review, we follow the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines,²⁴  and the study protocol was registered with PROSPERO (registration number: CRD42019131708).

Articles were included in the systematic review if (1) the cohort of interest was children with NDD, as identified in the ICD-11² ; (2) they included a comparator group, including typically developing (TD) children or normative data; (3) children were tested for motor function between 0 and 24 months of age; (4) motor functions were assessed by using standardized assessments; and (5) they were published in English peer-reviewed journals. Articles were excluded if the cohort included children at risk for NDD without a later formal diagnosis.

The article search was performed systematically in the Cumulative Index to Nursing and Allied Health Literature, Embase, Medline, PsycINFO, and Scopus electronic databases. A combination of key terms, as well as free-text words, was included in the systematic search: “infant,” “toddler,” “intellectual developmental disorder,” “developmental speech sound disorder,” “autism spectrum disorder,” “developmental learning disorder,” “developmental coordination disorder,” “attention deficit hyperactivity disorder,” “stereotyped movement disorder,” “fetal alcohol spectrum disorder,” “sensory integration disorder,” “motor impairment,” and “motor delay.” Search results were dated from the earliest record to the first week of June 2019. Reference lists from relevant articles were hand searched for eligible studies. The Medline search strategy is displayed in Supplemental Table 3.

One reviewer (Y.H.L.) screened all the search results, and 3 independent reviewers (M.L., A.F.-J., and J.D.) each screened one-third of the search results by applying the eligibility criteria on the title and abstract of identified articles and then conducting a full-text screening. Disagreements were resolved through discussion to achieve a final consensus on included articles.

Four reviewers (Y.H.L., M.L., J.D., and A.F.-J.) independently extracted data from the eligible studies using standardized forms. Data included country of study, study design, participant characteristics, motor function assessment tool, outcome measure, results, and quality appraisal of the study. The methodologic quality of the studies was assessed by using the Standard Quality Assessment Criteria for Evaluating Primary Research Papers from a Variety of Fields checklist²⁵  (Supplemental Table 4). Study quality was classified on the basis of the calculated score percentages²⁶ : strong (>80%), good (70%–80%), adequate (50%–69%), or limited (<50%). Studies rated as limited quality were excluded from analysis.

Narrative synthesis was used to synthesize data across all studies. Motor functions were categorized on the basis of the domains of the outcome measures: fine motor, gross motor, generalized motor, and general movement functions. Generalized motor function was derived from outcome measures that provided composite scores for fine and gross motor functions. In addition, a meta-analysis was used to synthesize data from studies involving children with later ASD because both mean and SD values of motor function outcomes were reported in most of those studies, whereas they were not reported in studies involving other children with later NDD. Data used in the meta-analysis were analyzed by age group (0–6, 7–12, 13–18, and 19–24 months). When SD values were not available, they were estimated from available confidence interval (CI) data, SEs, or other methods recommended by the Cochrane Collaboration.²⁷  Data represented in figures and graphical representations were extracted by using the WebPlotDigitizer software.²³  Standardized mean difference and 95% CI values for each motor outcome were calculated and presented in forest plots by using the Review Manager software version 5.3.²⁸  Random-effects models were used to calculate the pooled estimates and their CI. Statistical significance was assumed when the P value was <.05. The I² value was used to interpret the degree of heterogeneity.²⁷  The magnitude of the effect size, the standardized mean difference, was interpreted as follows: small, 0.20 to <0.50; medium, 0.50 to <0.80; and large, ≥0.80.²⁹ 

In the search, we identified 7689 studies from across 5 databases. After eligibility screening, 25 studies were included in the systematic review, and 13 of these were included in the meta-analysis (Fig 1). In total, 1028 children had later NDD, and 74 861 were TD children. In 2 of the included studies, normative data were used as a comparator.

The main characteristics of the included studies are summarized in Table 1. Participants ranged between 0 and 24 months old at the time of motor function assessment. Later diagnoses of NDD included ADHD (2 studies), ASD (21 studies), DCD (1 study), FASD (2 studies), and pervasive developmental disorder not otherwise specified (PDD-NOS) (1 study). The motor function assessment tools used included the Ages and Stages Questionnaire, Second Edition³⁰  (ASQ-2), Bayley Scales of Infant Development, Second Edition³¹  (BSID-II), Bayley Short Form Research Edition³²  (BSFR), Denver Developmental Screening Test³³  (DDST), General Movements Assessment³⁴  (GMA), Griffiths Mental Developmental Scales–Extended Revised³⁵  (GMDS-ER), Kyoto Scale of Psychological Development³⁶  (KSPD), Mullen Scales of Early Learning³⁷  (MSEL), Peabody Developmental Motor Scales, Second Edition³⁸  (PDMS-2), and Vineland Adaptive Behavior Scales, Second Edition³⁹  (VABS-II). Of these, the MSEL was the most frequently used motor function assessment tool.

In Table 2, we describe the motor outcomes reported in each study across the 4 age ranges. In one study,⁴⁰  both the MSEL and VABS-II were used to assess fine and gross motor functions; data from both motor assessment tools were displayed in forest plots, but only data from the MSEL were used in the calculation of the effect size.

Classification of the quality of the included studies revealed that 12 studies were of strong quality, 4 were of good quality, and 9 were of adequate quality (Supplemental Table 5). Three studies identified as being of limited quality were excluded from analysis.

Fine motor function of children with later ASD was examined in 8 studies by using the DDST, MSEL, and PDMS-2. In 5 studies,^41–45  the authors reported no difference in fine motor function between the later ASD and TD groups (mean age: 6 months). In contrast, the authors of 3 studies^44,46,47  found that children with later ASD (mean age: 6 months) demonstrated poorer fine motor function in relation to TD children.

A meta-analysis of 6 studies involving children with later ASD and TD children revealed an overall small effect size of 0.40 (95% CI −0.57 to −0.23; P < .001) (Fig 2A), indicating poorer fine motor function in the later ASD group. There was evidence of insignificant heterogeneity of the effect size (I² = 0%).

Gross motor function of children with later ADHD and ASD was examined in 7 studies by using the DDST, KSPD, MSEL, and PDMS-2. In 4 studies,^43,44,46,47  the authors reported no difference in gross motor function between the later ASD and TD groups (mean age: 6 months). In contrast, the authors of 3 studies^42,48,49  reported that children with later ADHD and ASD (ADHD, 1 study; ASD, 2 studies; age range: 3–6 months) showed poorer gross motor function compared with TD children.

A meta-analysis of 5 studies involving children with later ASD and TD children revealed an overall small effect size of 0.24 (95% CI −0.46 to −0.03; P = .03) (Fig 2B), indicating poorer gross motor function in the later ASD group. There was evidence of insignificant heterogeneity of the effect size (I² = 20%).

Generalized motor function of children with later ASD and FASD was examined in 2 studies by using the BSID-II and VABS-II. In both studies,^42,50  the authors reported that children with later ASD and FASD (ASD, 1 study; FASD, 1 study; mean age: 6 months) showed poorer generalized motor function when compared with TD group. A meta-analysis was not performed because of the availability of only one study in which generalized motor function in children with later ASD was reported.

General movements of children with later ADHD, ASD, DCD, and PDD-NOS was examined in 3 studies by using the GMA. In one prospective study,⁵¹  the authors assessed GMA in 5 children with 3 different NDD (ADHD, n = 1; DCD, n = 3; PDD-NOS, n = 1) and 23 TD children (age range: 3–5 months) and found that the general movement optimality scores of children with later-diagnosed ADHD, DCD, and PDD-NOS were in the normal range of those of TD children. It is important to note that the findings by Yuge et al⁵¹  are limited by a small sample size and included insufficient quantitative data to compare general movement function between different groups with later-diagnosed NDD and the TD group. In contrast, the authors of one retrospective study⁵²  compared the general movement optimality scores between 20 children with later ASD (age range: 1–3 months) and 20 TD children and found that those with later ASD demonstrated lower optimality scores, indicating poorer general movement function. The authors of another retrospective study⁵³  found that 87.5% (7 of 8) of children with later ASD showed abnormal general movements within the first 6 months of life. A meta-analysis of the studies involving children with later ASD was not possible because SDs could not be obtained.

Fine motor function of children with later ADHD and ASD was examined in 8 studies by using the DDST, MSEL, and VABS-II. In 2 studies,^48,54  the authors reported no difference in fine motor function between the later ADHD and ASD groups (ADHD, 1 study; ASD, 1 study; age range: 9–12 months) and the TD group. In contrast, the authors of 5 studies^{40–42,45,55}  found that children with later ASD (age range: 7–12 months) demonstrated poorer fine motor function in relation to TD children. Statistical difference between the later ASD and TD groups was not reported in one study.⁵⁶ 

A meta-analysis of 6 studies involving children with later ASD and TD children revealed an overall moderate effect size of 0.79 (95% CI −1.06 to −0.52; P < .001) (Fig 3A), indicating poorer fine motor function in the later ASD group. There was evidence of moderate heterogeneity of the effect size (I² = 44%).

Gross motor function of children with later ADHD, ASD, and FASD was examined in 6 studies by using the DDST, GMDS-ER, MSEL, and VABS-II. In one study,⁵⁷  the authors reported no difference in gross motor function between the later FASD (age range: 7–12 months) and TD groups. In contrast, the authors of 5 studies^{40,42,48,55,57}  found that children with later ADHD, ASD, and FASD (ADHD, 1 study; ASD, 3 studies; FASD, 1 study; age range: 7–12 months) demonstrated poorer gross motor function in relation to TD children.

A meta-analysis of 3 studies involving children with later ASD and TD children revealed an overall moderate effect size of 0.56 (95% CI −0.84 to −0.29; P < .001) (Fig 3B), indicating poorer gross motor function in the later ASD group. There was evidence of insignificant heterogeneity of the effect size (I² = 0%).

Generalized motor function of children with later ASD and FASD was examined in 3 studies by using the BSID-II, BSFR, and VABS-II. In one study,⁵⁸  the authors reported no difference in generalized motor function between the later ASD (mean age: 9 months) and TD groups. In contrast, the authors of 2 studies^42,50  found that children with later ASD and FASD (ASD, 1 study; FASD, 1 study; mean age: 12 months) demonstrated poorer generalized motor function in relation to TD children.

A meta-analysis of 3 studies involving children with later ASD and TD children revealed an overall nonsignificant effect size of 0.50 (95% CI −1.25 to 0.24; P = .18) (Fig 3C), indicating comparable generalized motor function between the later ASD and TD groups. There was evidence of considerable heterogeneity of the effect size (I² = 90%).

Fine motor function of children with later ASD was examined in 8 studies by using the ASQ, DDST, MSEL, and VABS-II. In 2 studies,^48,54  the authors reported no difference in fine motor function between the later ASD and TD groups (mean age: 18 months). In contrast, the authors of 5 studies^{40,41,43,59,60}  found that children with later ASD (age range: 14–18 months) demonstrated poorer fine motor function in relation to TD children. Statistical difference between the later ASD and TD groups was not reported in one study.⁶¹ 

A meta-analysis of 6 studies involving children with later ASD and TD children revealed an overall moderate effect size of 0.72 (95% CI −0.95 to −0.50; P < .001) (Fig 4A), indicating poorer fine motor function in the later ASD group. There was evidence of substantial heterogeneity of the effect size (I² = 62%).

Gross motor function of children with later ADHD and ASD was examined in 5 studies by using the ASQ, DDST, KSPD, MSEL, and VABS-II. In one study,⁴⁰  the authors reported no difference in gross motor function between the later ASD and TD groups (mean age: 14 months). In contrast, the authors of 4 studies^43,48,49,60  found that children with later ADHD and ASD (ADHD, 1 study; ASD, 3 studies; age range: 14–18 months) demonstrated poorer gross motor function in relation to TD children.

A meta-analysis of 3 studies involving children with later ASD and TD children revealed an overall large effect size of 1.11 (95% CI −1.98 to −0.24; P = .01) (Fig 4B), indicating poorer gross motor function in the later ASD group. There was evidence of considerable heterogeneity of the effect size (I² = 96%).

Fine motor function of children with later ASD was examined in 12 studies by using the MSEL and VABS-II. In 3 studies,^40,54,59  the authors reported no difference in fine motor function between the later ASD and TD groups (mean age: 24 months). In contrast, the authors of 6 studies^{41–43,55,62,63}  found that children with later ASD (mean age: 24 months) demonstrated poorer fine motor function in relation to TD children or normative data. Statistical differences between the later ASD and TD groups were not reported in 3 studies.^45,61,64 

A meta-analysis of 7 studies involving children with later ASD and TD children revealed an overall large effect size of 1.14 (95% CI −1.48 to −0.81; P < .001) (Fig 5A), indicating poorer fine motor function in the later ASD group. There was evidence of substantial heterogeneity of the effect size (I² = 71%).

Gross motor function of children with later ASD was examined in 6 studies by using the MSEL and VABS-II. In 5 studies,^{42,43,55,62,63}  the authors found that children with later ASD (mean age: 24 months) demonstrated poorer gross motor function in relation to TD children or normative data. Statistical difference between the later ASD and TD groups was not reported in one study.⁶⁴ 

A meta-analysis of 4 studies involving children with later ASD and TD children revealed an overall large effect size of 1.47 (95% CI −1.74 to −1.21; P < .001) (Fig 5B), indicating poorer gross motor function in the later ASD group. There was evidence of insignificant heterogeneity of the effect size (I² = 0%).

Generalized motor function of children with later ASD was examined in 2 studies by using the BSFR and VABS-II. In both studies,^42,58  the authors reported that children with later ASD demonstrated poorer generalized motor function in relation to TD children.

A meta-analysis of 2 studies involving children with later ASD and TD children revealed an overall large effect size of 1.33 (95% CI −1.69 to −0.96; P < .001) (Fig 5C), indicating poorer generalized motor function in the later ASD group. There was evidence of substantial heterogeneity of the effect size (I² = 61%).

Our aim for this systematic review was to synthesize evidence of early motor functioning in children later diagnosed with NDD. Although the included studies documented early motor impairments among children later diagnosed with ADHD, ASD, DCD, FASD, and PDD-NOS, 84% (21 of 25) of the included studies in this review involved young children with later ASD. This limits our capacity to make inferences about the transdiagnostic nature of early motor impairments as an early marker of NDD risk. Accordingly, we focus the discussion on the findings of children with later ASD. Additional longitudinal studies with early motor assessments that are linked to later developmental functioning and diagnostic outcomes are essential to enable comparisons of early motor difficulties across different NDD groups.

Consistent with previous findings of motor impairments in individuals with NDD,^65–68  we found that children with later ASD exhibited early impairments in fine, gross, and generalized motor functions. Our meta-analysis indicated that children with later ASD showed increasingly poorer motor function, compared with TD children, with increasing age. During early infancy (0–6 months), those with later ASD displayed small differences in fine and gross motor functions relative to TD children. By 19 to 24 months, large effect sizes were found across fine, gross, and generalized motor functions. These findings suggest that motor impairments can be detected in children from as early as 6 months, and thus these early motor impairments could be useful to identify children at risk for ASD. However, the predictive validity of early motor impairment should not be used alone and should be considered together with other markers of neurodevelopmental vulnerability, such as deviations in early speech and language development.⁶⁹ 

An important consideration in developing strategies for the early identification of at-risk children is the selection of appropriate standardized motor function assessment tools for children between 0 and 24 months. In the present review, we identified 10 different types of standardized assessments that were used to measure motor function. The use of a variety of assessment tools to measure motor function is not uncommon. Spittle et al⁷⁰  found that across the 9 assessment tools used to measure motor development in preterm infants, all were appropriate for use, but each tool had a different purpose (ie, discriminative, predictive, or evaluative purposes). In the studies included in the current review, the MSEL was most frequently used. This is a practitioner-administered assessment that can be used on children from birth to 68 months.⁷¹  Although the MSEL has been reported to have high interrater reliability (range: 0.91–0.99) and good construct and criterion validity in a normative population,⁷²  there is limited information regarding its sensitivity and specificity. More research on the psychometrics of the MSEL is needed to determine its appropriateness as a gold standard assessment tool for the measurement of motor function in at-risk children between 0 and 24 months.

Only 3 of the 4 motor domains examined in the present review revealed early motor impairments in at-risk children. A large proportion of fine, gross, and generalized motor functions in young children with later ASD were found to be different from that of TD children; however, there were mixed findings in the general movement function. It is uncertain how the general movement function can provide meaningful information to detect motor impairments in at-risk children because only 3 included studies in the present review used the GMA to evaluate the general movement function. However, previous work has found support for the notion that GMA is a transdiagnostic measure of neurodevelopmental vulnerability. For example, among preterm infants, abnormal writhing general movements are associated later cognitive vulnerability.⁷³  Further research on the utility of the general movement function in the detection of early motor impairments in children later diagnosed with NDD is warranted.

In addition to the meta-analysis only comprising studies of children later diagnosed with ASD, this review has several limitations. First, a strict selection criterion was used to include studies in which motor function was only examined in children with later-diagnosed NDD. This led to many studies being excluded, for example, studies in which motor function was examined in at-risk children with no later diagnosis. In future work, researchers may wish to consider the association between early motor development and dimensional measures of later functioning, for example, cognitive development, or other transdiagnostic markers of neurodevelopmental vulnerability, such as dysregulated irritability. Second, the current study contained only data from children with a later diagnosis of ADHD, ASD, DCD, FASD, and PDD-NOS; therefore, results may not be applicable to children with other types of NDD. Third, a wide range of heterogeneity was found in the meta-analyses; effect sizes with substantial and considerable heterogeneity should be interpreted with caution.

There is scope for research to further understand how early detection of motor impairments in children can be used as an early marker of ASD risk. In addition, the understanding of early motor impairment as a transdiagnostic marker of later NDD is currently limited by a paucity of studies on motor function in children later diagnosed with intellectual disability, language and speech disorder, ADHD, DCD, and FASD. Further studies in which motor function is examined in children across a wide range of NDD can be used to develop a comprehensive transdiagnostic profile of motor function in NDD. Given the heterogeneity in the clinical presentation of children with NDD, it is likely that motor profiles will be similarly heterogenous; however, transdiagnostic comparisons provide a basis for understanding latent classes of functioning that have important clinical implications.⁷⁴  In future work, researchers should also aim to determine best practice screening strategies for the early detection of motor impairments in children at risk for ASD and other NDD. This includes identifying appropriate assessment tools, standardized cutoffs for the classification of motor impairment, the recommended age for assessment, screening procedures, and cost.⁷⁵ 

With the present systematic review, we provide evidence of early motor impairments in young children later diagnosed with ASD. With this review, we also provide mixed evidence of shared features of motor impairments across different types of NDD. Further work is needed to understand the clinical utility of motor impairment detection as a transdiagnostic early marker of NDD risk.

ADHD

attention-deficit/hyperactivity disorder

ASD

autism spectrum disorder

ASQ-2

Ages and Stages Questionnaire, Second Edition

BSID-II

Bayley Scales of Infant Development, Second Edition

BSFR

Bayley Short Form Research Edition

CI

confidence interval

DCD

developmental coordination disorder

DDST

Denver Developmental Screening Test

FASD

fetal alcohol spectrum disorder

GMA

General Movements Assessment

GMDS-ER

Griffiths Mental Developmental Scales–Extended Revised

ICD-11

International Classification of Diseases, 11th Revision

KSPD

Kyoto Scale of Psychological Development

MSEL

Mullen Scales of Early Learning

NDD

neurodevelopmental disorders

PDD-NOS

pervasive developmental disorder not otherwise specified

PDMS-2

Peabody Developmental Motor Scales, Second Edition

TD

typically developing

VABS-II

Vineland Adaptive Behavior Scales, Second Edition