Efficacy of interventions to improve motor performance in children with developmental coordination disorder: a combined systematic review and meta-analysis

Authors


Dr Bouwien Smits-Engelsman at Department of Kinesiology, KU Leuven, Tervuursevest 101, 3001 Heverlee, Leuven, Belgium. E-mail: bouwiensmits@hotmail.com

Abstract

Aim  The aim of this study was to review systematically evidence about the efficacy of motor interventions for children with developmental coordination disorder (DCD), and to quantify treatment effects using meta-analysis.

Method  Included were all studies published between 1995 and 2011 that described a systematic review, (randomized) clinical trial, or crossover design about the effect of motor intervention in children with DCD. Studies were compared on four components: design, methodological quality, intervention components, and efficacy. Twenty-six studies met the inclusion criteria for the review. Interventions were coded under four types: (1) task-oriented intervention, (2) traditional physical therapy and occupational therapy, (3) process-oriented therapies, and (4) chemical supplements. For the meta-analysis, effect sizes were available for 20 studies and their magnitude (weighted Cohen’s d [dw]) was compared across training types.

Results  The overall effect size across all intervention studies was dw=0.56. A comparison between classes of intervention showed strong effects for task-oriented intervention (dw=0.89) and physical and occupational therapies (dw=0.83), whereas that for process-oriented intervention was weak (dw=0.12). Of the chemical supplements, treatment with methylphenidate was researched in three studies (dw=0.79) and supplementation of fatty acids plus vitamin E in one study (no effect). The post hoc comparison between treatment types showed that the effect size of the task-oriented approach was significantly higher than the process-oriented intervention (p=0.01) and comparison (p=0.006). No significant difference in the magnitude of effect size between traditional physical and occupational therapy approaches and any of the other interventions emerged.

Interpretation  In general, intervention is shown to produce benefit for the motor performance of children with DCD, over and above no intervention. However, approaches from a task-oriented perspective yield stronger effects. Process-oriented approaches are not recommended for improving motor performance in DCD, whereas the evidence for chemical supplements for children with DCD is currently insufficient for a recommendation.

Abbreviations
CO-OP

Cognitive orientation to daily occupational performance

DCD

Developmental coordination disorder

NTT

Neuromotor task training

What this paper adds

  •  Children with DCD benefit from some form of motor intervention.
  •  Task-oriented interventions and traditional motor-based skill training yield significant effects.
  •  Task-oriented interventions improve motor performance more than process-oriented interventions.
  •  Few intervention studies capture changes in the level of participation.
  •  Higher-quality studies are needed on optimal intensity and treatment duration.

Children with developmental coordination disorder (DCD) are identifiable by the difficulties they have in performing fine and gross motor tasks, which affect their performance in the classroom and in activities of daily living.1 According to the DSM-IV,2 in DCD the level of motor coordination is below that expected given the child’s chronological age and intelligence, and may lead to problems in activities of daily living and/or academic performance.3 The estimated prevalence of children with DCD is between 6 and 13% of all school-aged children,4 with some reports finding that males experience a higher incidence than females.5 Forty per cent of the children diagnosed as having delayed motor development before starting school continue to have this problem 10 years later.6 These figures indicate that DCD is not a condition that exists only in earlier childhood.

Several comorbid problems are common in children with DCD including a substantial overlap with attention-deficit–hyperactivity disorder (ADHD), dyslexia, and autistic spectrum disorders.7,8 Children with DCD show higher rates of social difficulties, low self-esteem, and associated behavioural problems during childhood and adolescence.9 In particular, children with combined DCD and ADHD show poorer outcomes when evaluated in early adulthood, in terms of academic achievement and psychosocial adjustment.10 Indeed, the persistent nature of DCD in around one-half of individuals first diagnosed in childhood emphasizes the importance of targeted intervention.11–13

Therapeutic approaches are drawn from occupational therapy,14–20 physiotherapy,21–30 medicine31–33 (e.g. methylphenidate), diet34 (e.g. fatty acids plus vitamin E supplementation,), and education21,35,36 (teachers, parents, physical education). Although different terms can be found in the literature to describe interventions for DCD, approaches can be grouped under three main types: process-oriented, task-oriented, and conventional physical and occupational therapy. Chemical treatments (e.g. fatty acids) represent a newer, fourth type.

Process-oriented approaches target the components or body functions needed to perform activities. Sensory integration, kinaesthetic training, perceptual training, or combinations are examples of these bottom-up approaches. For DCD, the hypothesis is that the improvement of body functions such as sensory integration, kinaesthesia, muscle strength, core stability, visual–motor perception, and so on leads to better skill performance.37,38

By comparison, task-oriented approaches tend to focus on motor performance, i.e. on learning particular motor skills, with attention given to specific aspects of task performance that are causing the child difficulty. Leading examples are neuromotor task training (NTT),26,35,39,40 the cognitive orientation to daily occupational performance (CO-OP) approach,37,38,41 and imagery training.29 In reality, these task-oriented approaches are all based on a combination of current motor control or motor learning and ecological principles, with the relative contribution of these frameworks varying from one method to another.42

NTT can be considered a true hybrid: from motor learning theory, it draws strongly on the notion that task structure and scheduling are fundamental to the way skills are assembled over repeated learning trials and sessions; from the ecological approach, it considers how task and environmental constraints can be manipulated to provide some leverage for the individual child whose motor impairment does not enable direct and seamless movement in a particular workspace.20,35–38

CO-OP can be considered the most pure example of a top-down approach. It focuses particularly on the use of cognitive strategies to facilitate skill acquisition, and uses a collaborative, problem-solving approach adapted from cognitive-behavioural therapy, particularly the work of Meichenbaum. The child is encouraged to form a mental model of how to attack a movement task; they are led to generate a movement goal, plan its implementation, and reflect on how their performance was or was not successful (goal, plan, do, check).37,38,42

From a traditional physical therapy or occupational therapy training perspective, individuals with DCD are trained in the most important fundamental gross motor and fine motor skills (hopping, jumping, throwing, and catching; cutting, drawing, writing), and in the basic motor abilities that are thought to be prerequisite for skills (e.g. trunk stability for certain fine-motor skills). These approaches combine underlying process-oriented approaches with direct skill training; the underlying assumptions are that motor skills are developed in a sort of hierarchical fashion. Basic abilities (such as postural control, in-hand manipulation, visual–perceptual skills) need to be refined in conjunction with teaching complex motor skills.

Although many different intervention strategies have been used and studied, it remains unclear which best improve motor performance or activities in children with DCD and alleviate the associated problems. To determine the most effective intervention strategy, we searched the literature on interventions in children with DCD published between 1995 and December 2011. The longer an intervention method has existed, the larger the possibility that it has been investigated for its treatment effects. This is certainly true for sensory integration and kinaesthetic training.5,26 Studies investigating sensory integration and kinaesthetic training occurred mainly between the early 1970s and 1996, with very few since 1996. Several studies published after 1996 have investigated task-oriented approaches like CO-OP and NTT, both of which are relatively new. No meta-analysis has been conducted so far that includes this work.

The main aim of this meta-analysis was to synthesize the literature from 1995 to December 2011, a period during which DCD gained greater recognition as a diagnosis in its own right. In 1994 the London Consensus group43 produced a more detailed description of DCD compared with the DSM-III-R (1987)44 and DSM-IV (1994).2 This was an important milestone in the research into DCD. It meant the beginning of a uniform approach to children with motor impairment, not otherwise specified by a medical diagnosis. From 2004 to 2006, several professionals from different countries met to obtain a new agreement about the diagnosis, research, and intervention characteristics and comorbidities of DCD, resulting in the Leeds Consensus statement in 2006.45 From these meetings, several issues emerged: there was no consistency in the interpretation and application of diagnostic tests, and no uniformity in the application of the various evidence-based intervention methods. The Leeds Consensus set the following guidelines that intervention approaches should meet: (1) activities should be functional, based on goals that are relevant to daily living and meaningful to the child; (2) they should enhance generalization and application in the context of everyday life; and (3) interventions must be evidence-based and grounded in theories that are applicable to understanding children with DCD.38,45

Thereafter, various (inter-)national professionals inside and outside Europe, some of whom participated in the seminars in Leeds, decided to write a directive for diagnosis, prognosis, and intervention for children with DCD. This resulted in the European Academy for Childhood Disability recommendations for diagnosis and treatment of DCD, issued in January 2012.46 In support of these recommendations for intervention, a systematic review on treatment efficacy was initiated. It was the aim of this review to rate both the scientific quality of the studies and compare the magnitude of differences across treatment types, highlighting the most effective.

Method

Criteria for inclusion and exclusion

In advance of the actual literature search, selection criteria were set for the abstracts. Abstracts of meta-analysis, systematic reviews, randomized controlled trials (RCTs), clinical trials, and narrative reviews were included. The papers had to be written in English, German, or Dutch and published between 1995 and 2011.

The populations included in the studies were children of any age, diagnosed with DCD or with possible DCD. DCD was defined either according to the criteria of the DSM-IV2 or as motor impairment not otherwise specified by a medical diagnosis, but examined with standardized motor tests confirming motor impairment. Studies reporting on children with poor motor performance, which was defined as 1SD below the mean, were also included. Studies investigating a sample of children with a syndrome (like ADHD or dyslexia), were included in the descriptive part of the study but excluded from the meta-analysis.

The included studies had to have motor outcome measures with standardized and internationally accepted assessments, for example, the Movement Assessment Battery for Children, the Concise Assessment Method for Children’s Handwriting, or the Test of Gross Motor Development-2.

Literature search

The literature search spanned the period January 1995 to December 2011. The following databases were consulted: Medline, the Cochrane Library, PubMed, CINAHL, PsychInfo, PsychLit, OTDBase, OTseeker, PEDRO, ERIC, Embase, and HealthSTAR. The search terms were agreed upon at a meeting of the international working group for the EACD recommendations for DCD in 2008.

The following search terms were used (connected with ‘OR’): motor skills disorder, developmental coordination disorder, clumsiness, clumsy, clumsy child syndrome, clumsy child, in-coordination, dys-coordination, minimal brain dysfunction, minor neurological dysfunction/disorder, motor delay, perceptual-motor impairment, motor coordination difficulties/problems, motor learning difficulties/problems, mild motor problems, non-verbal learning disability/disorder/dysfunction, sensorimotor difficulties, sensory integrative dysfunction, physical awkwardness, physically awkward, psychomotor disorders, motor control and perception, apraxias, developmental dyspraxia, perceptual motor dysfunction, minimal cerebral dysfunction.

These search terms above where combined (with ‘AND’) with the general terms (connected with OR) ‘physical therapy’, ‘physiotherapy’, ‘occupational therapy’, ‘intervention’, ‘treatment’, and the more specific treatment approaches (connected with OR) sensory integration, Neuro-Developmental Treatment (NDT): Neuromotor Task Training (NTT), Cognitive Orientation to daily Occupational Performance (CO-OP), Perceptual Motor Training (PMT), motor imagery training (MIT), sensory integration training (SIT), task-specific training, cognitive training, timing control, kinaesthetic training.

The following exclusions (‘NOT’) were used to obtain more pure DCD studies: cerebral palsy, stroke, traumatic brain injury, leukodystrophia or muscular disorders. Additionally, references of all selected articles were checked for further papers suitable for inclusion.

Each abstract resulting from the search was individually screened for inclusion by at least two of four independent experts (BS-E, A-CvdK, RM-vdM, EV-vdB). When consensus existed, the full-text articles were evaluated.

Each full-text article was again rated by two independent experts who needed to reach a consensus to include the paper in the study. If consensus was not reached, a third expert was consulted.

Next, for each of the included studies, the experts had to agree on the level of evidence rated according to levels of evidence (Oxford Centre for Evidence-Based Medicine,47 or in case of an RCT, the PEDro scale;48Tables I and II). If consensus was not reached by the two experts, a third was consulted. Authors did not rank their own studies. Moreover, all conclusions and recommendations were reviewed by the international DCD guideline group in several meetings.46

Table I. Levels of evidence (modified according to Oxford Centre for Evidence-Based Medicine), March 200947
Level of evidenceGradeOxford levelOxford definition (intervention studies)
  1. QUADAS, Quality Assessment Tool for Diagnostic Accuracy Studies.

1 (high)Evidence from a meta-analysis of randomized controlled trialsIaEvidence from a meta-analysis or systematic review of randomized controlled trials (with homogeneity).
Evidence from at least one randomized controlled trial (intervention study) or well-controlled trial with well-described sample selection (diagnostic study); confirmatory data analysis. QUADAS rating >10.IbEvidence from at least one randomized controlled trial.
2 (moderate)Evidence from at least one controlled study without randomization or QUADAS rating >7.IIaEvidence from systematic review of cohort studies (with homogeneity); or evidence from at least one controlled study without randomization.
Evidence from at least one other type of quasi-experimental study.IIbIndividual cohort study (including low-quality randomized studies, e.g. <80% follow-up). Evidence from at least one other type of quasi-experimental study.
3 (low)Evidence from observational studies. QUADAS rating >4.IIIEvidence from case–control studies; or evidence from observational studies
4 (very low)Evidence from expert committee reports or experts.IV/VEvidence from expert committee reports or experts.
Table II. Definitions of the PEDro scale48
  1. The PEDro scale is a 10-point scale with dichotomous (1/0, referring to yes/no) responses that account for key quality aspects of an experimental study. A score of 7 or higher indicates a good, reliable study, and a score of 5 or 6 as acceptable. No, no points; yes, one point for questions 2–11. Total score ≥7, good; 5–6, moderate; <5, weak.

1. Eligibility criteria were specified (not rated).NoYes
2. Subjects were randomly allocated to groups (in a crossover study, subjects were randomly allocated another in which treatments were received).NoYes
3. Allocation was concealed.NoYes
4. The groups were similar at baseline regarding the most important prognostic indicators. Yes
5. There was blinding of all subjects.NoYes
6. There was blinding of all therapists who administered the therapy.NoYes
7. There was blinding of all assessors who measured at least one key outcome.NoYes
8. Measures of at least one key outcome were obtained from more than 85% of the subjects initially allocated to groups.NoYes
9. All subjects for whom outcome measures were available received the treatment or control condition as allocated or, whether this was not the case, data for at least one key outcome was analysed by ‘intention to treat’.NoYes
10. The results of between-group statistical comparisons are reported for at least one key outcome.NoYes
11. The study provides both point measures and measures of variability for at least one key outcome.NoYes

Data analysis

All abstracted study characteristics are shown in Table SI (online supporting information). These include population description, number of participants, age and relevant baseline testing results, type of intervention, description of the intervention, frequency, intensity and duration, outcome measures, description of results of the populations, short description of the conclusion and limitations, main conclusions.

Included articles were assessed by members of project group for methodological quality using the instruments mentioned, and results were added to Table SI results. To classify their efficacy, the results were characterized by either a ‘+’ for significant improvement in the experimental group(s) or ‘0’ for no significant change.

Estimates of effect size were calculated for each comparison of pre- and post-test results of a given treatment. The preferred estimate was Cohen’s d.49 To calculate estimates of effect size, means and standard deviations for pre- and post-test results were used whenever they were reported.

The inclusion of multiple dependent measures (and associated statistics) was allowed for each study. Mean effect sizes are not unduly affected by non-independence in most instances,50 and many authors choose not to weight studies according to the number of effect sizes. Pseudo-independence of effect size was assumed here. Outcome measures were therefore analysed at the level of each finding; there were 54 effect sizes, distributed among four treatment types, plus one for comparison groups, pre- and post test.

A random-effects approach to data analysis was used, based on the method of Lipsey and Wilson.51 Here it is assumed that variation between effect sizes is due not only to random sampling error but also to other factors within studies (e.g. methodological differences). The random-effects model is preferable when heterogeneity of effect sizes is present, indexed by Q statistics. Although less powerful than a fixed-effects approach, it does allow greater generalization. Effect sizes were weighted by the inverse of their variance. For each treatment type, a mean weighted Cohen’s d (dw) value was calculated using the SPSS (SPSS Inc, Chicago IL, USA).

Homogeneity of effect sizes was investigated using the Q statistic.52,53 Heterogeneity indicated that the variability in the effect sizes might be due to outliers or a moderator variable. A group of effect sizes was deemed homogeneous when the value of Q obtained was not significant at p<0.01. Outliers were also considered, if necessary.

The magnitudes of the mean effect size estimates (d) were then interpreted according to the convention of Cohen: d=0.30 (small effect size), d=0.50 (moderate effect size), and d=0.80 (large effect size).

Results

The total number of studies matching the search terms was 3708. Each abstract was then reviewed by the members of the project group (BS-E, A-CvdK, RM-vdM, EV-vdB) and assessed to determine inclusion or exclusion in the review. Eighty-five full-text studies were evaluated independently by two individual experts. Of these, 72 papers were deemed as candidates for this review. From these 72 studies, 44 were excluded after reading the full paper because the outcomes of the study did not include motor skills, were descriptive, did not describe any intervention method, or were a single-case study. The final complement of eligible studies in the systematic review was 26 (Fig. 1). For the meta-analysis proper, there were 20 eligible studies.

Figure 1.

 Flowchart of the literature search.

Level and quality of evidence

Twenty-six studies met the inclusion criteria. The dates of publication ranged from 1995 to 2011. Of those 26 articles, one was a systematic review, one was a meta-analysis, and the remainder were primary studies. The primary studies were scored with the PEDro scale to evaluate all RCTs and clinical trials. The overall quality was variable, with a mean of 6.0 (range 4–10).

The methodological quality of these studies was scored to determine the strength of the evidence according to the guidelines of the Oxford Centre for Evidence-based Medicine - Levels of Evidence.47

A total of 912 children with DCD or motor impairment classified with a test score of at least 1SD below the mean, with no other medical diagnoses, participated in the studies included in this review. The average sample size was 44 participants. Seventeen studies had fewer than 44 participants,2,5,14,16,17,20,22,24,26–28,30,31,33,35,36,65one had over a 100.34 The duration of the intervention ranged from 2 weeks to 6 months, and the amount of instruction ranged from 4 to 26 hours.

In the 26 studies, the following terms for the intervention were most often found: (1) under task-oriented intervention, CO-OP, NTT, and motor imagery were mentioned; (2) under process-oriented intervention, sensory integrative training and kinaesthetic training were mentioned; (3) under traditional therapy (physical therapy and occupational therapy), perceptual motor training and contemporary treatment approach, contemporary physical therapy based on fundamental motor training and occupational therapy, psychomotor training, and therapist-guided parent or teacher intervention were mentioned.

The Movement Assessment Battery for Children was the measure most frequently used for diagnostic purposes (12 studies), followed by the Concise Assessment Method for Children’s Handwriting (4), the Test of Gross Motor Development (second edition) (2), the Kinesthetic Sensitivity Test (2), and the Test Of Motor Impairment (2). Diagnostic tests that featured for a specific intervention were the following: The Tennessee Self-Concept Scale (1), the Child Anxiety Scale, the Perceived Motor Competence scale (1), the Developmental Test of Visual-Motor Integration (1), the Southern California Sensory Integration Test (1), the Finger Identification test (1) and the Evaluation Tool of Children’s Handwriting (1), the Bruininks–Oseretsky Test of Motor Proficiency (1), the Canadian Occupational Performance Measure (2), the Performance Quality Rating Scale (1), the Vineland Adaptive Behavior Scale (VABS) (1), the Self-Perception Profile for Children (1), the Harter’s Scale of Perceived Competence/Harter and Pike’s Pictorial Scale of Perceived Competence (1), the Motor Teaching Principles Taxonomy (1), the Visual Analogue Scales (1), the Rhythm Integrated Test (1), and the Sensory Integrated Praxis Test (1).

The outcome measures are all standardized, general assessments for gross- and/or fine-motor function. All studies included in our systematic review are presented in Table SI.

Twenty-two of the 24 primary studies were effective in improving motor performance. Of these, most were longer than 10 weeks, and the frequency of intervention was mostly once a week, with home exercises in three studies.24,28,55 In ten studies treatment was given in a group setting.15,18,20,21,23–25,28,30,54

Based on the data in Table SI, there is strong evidence that children with DCD benefit from task-oriented approaches (see also Pless and Carlsson55). In addition, there was sufficient evidence that motor-training-based interventions, as used in traditional physical or occupational therapies, were generally effective for children with DCD.14,17,19,22–26,28–30,36 Third, evidence for the efficacy of process-oriented approaches (e.g. sensory integrative training and kinaesthetic training) was conflicting.

Meta-analysis of intervention studies

Characteristics of the treatment and comparison samples are shown in Table SI. The total number of effect sizes for all studies combined was 54. Q statistics revealed homogeneity for each treatment category (p>0.01). The mean effect size across all treatment types was dw=0.56, indicating that treatment generally had a moderate effect. When the different treatment types were compared, the effect sizes were significantly different (F4,49=5.04, p=0.002); the task-oriented approach had the highest mean value (dw=0.89; 95% CI 0.64–1.14), followed by the traditional motor-training-based therapy (d=0.83; 95% CI 0.46–1.20), process-oriented intervention (dw=0.12; 95% CI −0.10–0.35), and comparison (dw=0.23; 95% CI 0.04–0.43). Chemical treatments contributed only four effect sizes, but yielded a moderate to large value (dw=0.79; 95% CI 0.15–1.42). The post-hoc Bonferroni comparison between treatment types showed that the effect size of the task-oriented approach was significantly higher than the process-oriented approach (p=0.01) and comparison (p=0.006). No significant difference in the magnitude of effect size between traditional physical and occupational therapy approaches and any of the other interventions emerged. The effect size of the methylphenidate was 0.81 and not significantly different from any of the other treatments.

Discussion

The results of meta-analysis of recent research support the view that task-oriented (dw=0.89) and traditional motor-training-based therapies (dw.83) show strong treatment effects in children with DCD, whereas that for process-oriented therapy is weak (dw=0.12). The implications of these numbers and of the other findings of the reviewed studies for theory and practice are discussed.

Task-oriented approach

For the task-oriented approaches, individual and group programmes were both effective ways of teaching motor skills in DCD. Task-oriented approaches work on teaching essential activities of daily living and, thereby, stimulate participation at home, school, in leisure, and sports.22,26,30,36–38,55 Whether the training effects are attributable mainly to flow-on effects for participation is an issue for continued research. On current understanding, it is recommended that NTT be used for children with DCD, with training yielding positive (task-specific) change on measures of gross- and fine-motor skill.21,22,26,27 The efficacy of CO-OP was also evident but work here requires more robust metrics. Children with DCD are shown to generate more effective strategies than those receiving other approaches consisting most commonly of sensory-integration and non-problem-related general fine- and gross-motor activities.14,16 Importantly, children with better verbal ability tend to make better gains in approaches where language plays an important mediating role; their ability to interpret cues and prompts and to implement responses to them may be a predictive factor here.14 In short, we recommend that task-oriented intervention methods like NTT and CO-OP should be prescribed with some confidence to children with DCD who are in need of intervention to improve their motor performance.

Motor imagery training

Caution is needed when considering the efficacy of motor imagery training because, first, there is only one study available (Wilson et al.29). In addition, of the total referred sample, 36% had a test score within the low normal range on the Movement Assessment Battery for Children. Therefore, conclusions about motor imagery training for severe DCD should be interpreted with care. However, the average change of scores on the Movement Assessment Battery for Children showed that perceptual motor training and motor imagery training improved motor performance to a similar degree.29 More high-quality research is needed to clarify the conditions under which motor imagery training is best implemented for children with DCD.

Traditional physical therapy and occupational therapy

There is sufficient evidence that traditional physical or occupational therapies (either individual or group) are also effective in teaching motor skills to children with DCD.14,17,19,22–26,28–30,36 Individual therapy was studied in two good-quality RCTs. In the RCT by Wilson et al.,29 it was evident that perceptual motor training results in improved motor performance (d=0.76). Here, 13 of the 17 children in the perceptual motor training group exceeded the upper limit of the 95% CI for comparisons. Moreover, Peens et al.23 showed that a combination of motor- and psychological-based intervention improved motor proficiency and self-concept in DCD. The findings from our meta-analysis showed that children with DCD benefit the most from task-oriented approaches; we recommend that perceptual motor training adopts a task-specific approach. For example, this integrated approach is effective in improving the motor performance of children with combined DCD/ADHD. Fifty per cent of these children obtained normal scores on the Movement Assessment Battery for Children immediately after training, while 35% reached normal limits after 4 weeks.28

Process-oriented training

Although the sensory integration method is the most commonly investigated approach, evidence is mixed. It is qualified by Hillier54 as being an effective treatment method in children with DCD. However, in the meta-analysis of Pless and Carlsson,55 sensory integration was not as effective as functional skill-specific interventions, and in the older analysis of Vargas and Camilli19 results for sensory integration did not show a generally positive effect for studies published between 1983 and 1993. This conflicting report warrants deeper consideration.

The meta-analysis by Vargas and Camilli19 evaluated 16 studies published between 1972 and 1994. Overall, the average effect size of sensory integration was d=0.29. Larger effect sizes were found in psycho-educational (d=0.39) and motor categories (d=0.40). Sensory integration methods were effective overall, but were not shown to differ from alternative ones. In studies comparing sensory integration with no treatment, a significant effect for sensory integration was replicated in earlier studies (1972–1982), but not in more recent studies (1983–1993). What must also be kept in mind is that sensory integration methods are relatively long in duration by normal standards.19

In the systematic review by Hillier,54 six studies reported positive effects for sensory integration,57–61 which was taken as support for this approach. Included among these were two very old studies by Wilson et al.61 and Humphries et al.57–59 Kaplan et al.60 investigated two papers in which all groups receiving intervention improved, but no one method was more effective than another. Leemrijse et al.,52 in a crossover study, showed a positive effect for sensory integration, although the sample was small (n=6). Finally, the meta-analysis of Pless and Carlsson55 showed that specific skill interventions in children with DCD had stronger treatment effects than sensory integration. The highest effect was noted for skill-specific interventions (1.46), the lowest for sensory integration (d=0.21).

Watemberg et al. concluded in his RCT that sensory integration is an effective treatment for DCD/ADHD; however, it was given in combination with other interventions and used in a task-specific manner.28 In addition, the RCT of Sudsawad et al.18 showed that although kinaesthetic training did improve kinaesthetic acuity, it did not assist handwriting or motor performance in children aged between 6 and 7 years.18 Taken together and in light of our review, there is no recent evidence of well-designed studies in support of sensory integration or kinaesthetic training.

Parent- and teacher-guided intervention

The complement of studies showed that parents and teachers were able to provide effective intervention for children with DCD provided this was tuned to their specific movement problems, and monitored by a professional. Involvement of parents and teachers is likely to ensure that the learned skills will continue to be used after the formal intervention.21,37,56 We need more studies to clarify how best to facilitate the involvement of parents and teachers, for example the use of user-friendly online resources that model good training practices.62–64

Methylphenidate

Methylphenidate had a positive effect on both behavioural ADHD symptoms and fine-motor performance (i.e. handwriting) in children with combined DCD and ADHD. Additional motor therapy was indicated for around one-half of the combined group. We do not know whether methylphenidate can be used to improve motor performance in DCD without comorbidities. Moreover, there are some important issues with longer-term side effects.31–33 Finally, although fatty acids with vitamin E can improve academic skills like reading and spelling as well as behaviour, there is no evidence for a positive effect on motor performance.34

Handwriting

Schoolchildren having handwriting problems are often presented for treatment; this is between 25 and 50% of all the referred children in the Netherlands.65 Prewriting exercises (and fine motor training, if required) has been shown to improve handwriting.27 Three different studies using a task-oriented approach based on NTT all showed significant improvement in handwriting.21,26,27 More specifically, task-oriented self-instruction involving knowledge of results had a positive effect on the quality of handwriting in children; importantly, this occurred both in regular and special schools.21

Limitations and implications for future research

After evaluating 3708 abstracts and 77 full-text papers, we were able to locate 24 primary studies addressing the research questions posed in this paper. Of these studies, 12 were categorized as ‘physical therapy and occupational therapy’, 10 were categorized as ‘task-oriented’, and three as methylphenidate. Of the 24 primary studies, seven used a research design that produced definite evidence (level Ib). However, most of the studies (17) used a research design (seven at level IIa and 10 at level IIb) that did not control for all other factors that could have produced the observed results, and are considered to have given moderate levels of evidence. This should be taken into account when interpreting the results. However, the advantage of a systematic review is that the consistency of the results adds to the grading of evidence. Moreover, analysis of the combined data set in the meta-analysis adds information not available in the isolated studies.

The scheduling of treatment was variable, from once a week14,16,22,24–27,29,31 to every (school-)day.15,17,31,32,34 As well, the amount of instruction ranged from 4 to 26 hours. The meta-analysis had too few effect sizes (n=54) to dissect the effects of frequency and duration. Moreover, few studies had a sufficient sample size to enable designs that would allow comparison between different forms of intervention.15–17,23,28,29 Most studies compared one intervention with no intervention.14,18,21,22,24–27,30–32,34,43

It was also alarming that many studies did not describe the intervention in sufficient detail to help researchers and clinicians alike to understand fully what it was about the treatment that made it successful. In other words, it was not always obvious that treatments using the same term were identical in nature (see Table SI). Most papers provided only a global description of the intervention, for example ball skills and balance activities.14–16,18,22–24,26–28,30,43

Additionally, there is a need to profile those children who do respond to treatment, as opposed to those who do not, including factors like age and comorbidity. Finally, only three studies17,25,27 investigated follow-up effects of 3 months or longer. Hence, no statements can be made about the longer-term effect of the treatment approaches. Next to more rigid scientific designs, this is a prime issue for future research. Importantly, the costs of intervention for DCD are enormous. For instance, in Germany, the public insurance system is paying about 400 million euros a year for occupational therapy for children with sensorimotor problems (mainly children with DCD, of whom the largest part are referred for fine motor and handwriting problems). A next step would be to conduct cost-effectiveness analysis, once these evidence-based recommendations are implemented.

Recommendations for practice

Our literature review largely substantiates the Leeds Consensus for intervention.45 These guidelines stipulate the following: treatment activities should be task-oriented, functional, and relevant to daily living; be child-centred in the process of therapy; involve significant others including parents and teachers; and be evidence-based and grounded in theories that are applicable to understanding children with DCD.

Although training effects are commonly shown at a task-specific level, it is not always apparent how they transfer across related and disparate areas of function. This refers to the classic issue of horizontal versus vertical transfer. Because the purpose of treatment is to achieve participation in meaningful life areas as independently as possible and with high quality of life, measures that capture the level of participation should also be used to evaluate treatment effects. Involvement of parents, teachers, and physical education is needed to maximize transfer into daily life and to ensure longer-term progress.21,36 More studies are needed to research the most effective way that parents and teachers can be engaged in instruction.38

Conclusion

We conclude that, in general, task-oriented (e.g. NTT and CO-OP) and motor-training-based intervention (physical therapy, occupational therapy) both show strong treatment effects (dw=0.89 and dw=0.83 respectively), whereas that for process-oriented therapy is weak (dw=0.12). Moreover, therapies should ideally have some task-oriented elements to them to promote transfer. Of the more top-down approaches, NTT could be the therapy of choice for younger children and those with lower verbal or learning capabilities. Children with well-developed verbal skills may well be the ones who benefit most from CO-OP. For handwriting problems, a task-oriented approach based on NTT is the best option. It is too early to state with confidence that motor imagery training is effective for treating DCD.29 Process-oriented approaches (e.g. sensory integration and kinaesthetic training) show only a weak effect, the same level as no-treatment comparison, and are not recommended for improving motor performance in DCD. Moreover, they do not follow current thinking on motor control or motor learning.

Finally, chemical supplements (like fatty acids and vitamin E) did not improve motor performance. However, daily doses of methylphenidate may be useful in children with comorbid ADHD and DCD, particularly for fine-motor issues like handwriting.

Ancillary