Professional Documents
Culture Documents
Leanne M. Slater, BPhty (Hons); Susan L. Hillier, PhD; Lauren R. Civetta, BPhty (Hons), Grad Dip Pub Hlth
Centre for Allied Health Evidence, School of Health Sciences, University of South Australia, Adelaide, South Australia,
Australia
Purpose: Performance-based measures of gross motor skills are required for children with developmental
coordination disorder to quantify motor ability and objectify change. Information related to psychometrics,
clinical utility, feasibility, and client appropriateness and acceptability is needed so that clinicians and research-
ers are assured that they have chosen the most appropriate and robust tool. Methods: This review identified
performance-based measures of gross motor skills for this population, and the research evidence for their
clinimetric properties through a systematic literature search. Results: Seven measures met the inclusion criteria
and were appraised for their clinimetric properties. The Movement Assessment Battery for Children and the
Test for Gross Motor Development (second version) scored highest on appraisal. Conclusions: The 2 highest
scoring measures are recommended in the first instance for clinicians wishing to evaluate gross motor perfor-
mance in children with developmental coordination disorder. However, both measures require further testing
to increase confidence in their validity for this population. (Pediatr Phys Ther 2010;22:170 –179) Key words:
child, developmental coordination disorder, human movement system, motor skills disorders, outcome
measures, psychomotor performance, validation studies as topic
Pediatric Physical Therapy Measures of Gross Motor Skills for Children With DCD 171
Studies were included if they (1) assessed psychomet- Three groups of search terms were used; outcome
ric properties of any of the measures that met the afore- measurement, DCD or associated terms, and psychometric
mentioned inclusion criteria, (2) were published in En- properties (separated by Booleans “OR” and “AND”). A full
glish in a peer-reviewed journal between January 1980 and list of search terms have been included in Table 1. The
February 2008 (the date 1980 was chosen to exclude those search was conducted between February 4 and 8, 2008.
measures that do not reflect current thinking regarding
motor performance13), and (3) if participants were aged Critical Appraisal
between 4 and 16 years. Studies that used normative sam-
Tool to Identify and Evaluate Measures. The Measure-
ples were included when relevant, for example, to gather
ment Critical Appraisal Tool (MCAT) was used to identify
normative data or when the population was representa-
and then evaluate the outcome measures in separate processes
tional, and therefore included children with motor difficul-
(Table 2). This tool was modified to suit the needs of this
ties. Such studies are clearly identified in the results. Stud-
review from a version of the criteria developed by the United
ies were excluded if (1) the majority of participants had
Kingdom’s Clearing House on Health Outcomes.18
been diagnosed with a disorder other than DCD or equiv-
Tool to Appraise the Quality of Studies. The quality
alent DCD term (ie, cerebral palsy, Down syndrome, and
scoring of the studies evaluating each measure was carried
autism were excluded), (2) the performance-based assess-
out using a modified critical appraisal tool from the
ment was used as a criterion measure for a screening tool or
NHMRC Handbook series.19 This tool is referred to as the
nonmotor performance-based test, or (3) the studies were
Quality Critical Appraisal Tool (QCAT) (a copy of the tool
published before 1980 because it was thought that the in-
can be obtained from the authors).
formation would pertain to a generation different from the
current child population and a different philosophical ap-
proach to motor skills. Data Extraction and Synthesis
Descriptive data on the clinimetrics and psychomet-
Search Strategies rics of outcome measures were extracted from the included
A systematic search of computer databases was used studies by the first reviewer and entered into separate
to locate studies. The databases searched were MEDLINE, spreadsheets. Two reviewers then appraised all articles us-
AMED, Meditext, Ausport, Ausportmed, Humanities and ing both the MCAT and the QCAT tools. If disagreements
Social Science Collection, A⫹ Education, CINAHL, Aca- arose between scorers, then a third reviewer was available
demic Search Premier, SPORTdiscus, ERIC, Health Source: to facilitate consensus. This was not necessary.
Consumer Edition, Health Source: Nursing/Academic Edi- The outcome measures quality score (MCAT) was ob-
tion, PsycINFO, Current Contents Connect, Blackwell tained from summing scores for validity (3 points), reli-
Synergy, Cochrane, PEDro, OTseeker, EMBASE, and Sco- ability (3 points), responsiveness (1 point), precision (1
pus. Pearling of the reference lists of all retrieved studies point), client centeredness (2 points), and tester centered-
was also performed. ness (2 points). Half scores were given if part of an item was
TABLE 1
Search Terms Used
Pediatric Physical Therapy Measures of Gross Motor Skills for Children With DCD 173
TABLE 3
Risk of Bias Scoring
TABLE 4
Clinical Properties of Outcome Measures
duplicates, and the studies excluded during the search The quality of the studies (risk of bias) was evaluated
procedure are summarized in the Figure 1. Six manuals using the QCAT. The number of studies for each measure,
meeting the inclusion criteria were identified in the lit- QCAT score ranges, and mean total quality scores with
erature, and 5 were obtained from 2 experienced pedi- standard deviation for each measure are summarized in
atric physiotherapists. The sixth manual for the ZNA Table 5. All studies, except 1, were at a low to moderate risk
was not obtained; however, the authors were contacted for bias in their reporting. The 1 study scoring poorly was
and subsequently referred researchers to studies already a descriptive report of concurrent validity between the M-
acquired. The BGMA manual was written in 1979, and ABC and a balance test. The main methodologic problems
thus was excluded. identified that increased chances of bias in studies were
TABLE 5
Quality Scores From QCAT on Studies for Each Outcome Measure
Pediatric Physical Therapy Measures of Gross Motor Skills for Children With DCD 175
lack of information on tester training and/or blinding and Reliability and Responsiveness of Measures
subject recruitment methods. Additionally, many subjects The BGMA has good to excellent test-retest reliability
represented samples of convenience and not true represen- and interrater reliability (r ⫽ 0.97 and 0.99, respectively) in
tations of the DCD/child populations, thus reducing the small samples (n ⫽ 48 and n ⫽ 10, respectively).22 The BOT-2
generalizability of the results. also has demonstrated good to excellent (r ⬎ 0.80) test-retest
and interrater reliability correlations.23 Correlations for
Critical Appraisal of Psychometrics of Each the test-retest24,34,45– 47 and interrater reliability46 – 49 on
Identified Outcome Measure total scores of M-ABC are consistently good to excellent
(r ⫽ 0.72– 0.95) in children developing typically and
Using the MCAT, relevant data on the psychometrics children with MI. The internal consistency of the M-ABC
of the outcome measures were extracted from the 33 arti- was investigated in a single study that showed the items
cles and 5 manuals. Information on the psychometrics of measured different constructs.50 The M-ABC total scores
each outcome measure is outlined briefly here. The mea- are reportedly adequately responsive to change45,51 and
surement critical appraisal scores and types of validity, re- sensitive45,51; however, subtest and individual item scores
liability, and responsiveness established are summarized have only moderate to low responsiveness to change.51
for each outcome measure in Table 6. Information on the The PDMS-2 has demonstrated good internal consis-
population in which validity and reliability were estab- tency and excellent (r ⫽ 0.89) interrater reliability correla-
lished and user/tester centeredness of individual outcome tions.26,41 The TGMD-2 reportedly has excellent internal
measures are also summarized in Table 6. For further de- consistency and interrater reliability27(r ⫽ 0.98). However,
tails, contact the authors. Note that normative data were for interrater reliability, raters used previously scored proto-
provided adequately for all outcome measures. cols and thus only summed and converted scores. Test-retest
reliability correlations of the TGMD-2 were good to excellent27
Validity of Measures (r ⫽ 0.84–0.96). The ZNA test-retest reliability correlations for
timed performance scores were good to moderate28,52–54 (r ⫽
Construct validity of the BGMA was demonstrated 0.65) but only fair for scores based on accessory move-
through factor analysis, all but 1 item loaded onto 7 factors; ments28,52–54 (r ⫽ 0.45). Inter- and intrarater reliability correla-
however, the BGMA has only 8 tasks, suggesting each item tions overall were moderate to excellent for the ZNA; however,
measures a different construct.22 The BOT-2 has moderate some very low correlations (⬍0.1) were obtained on some items
to excellent concurrent validity correlations with the with accessory movement scores.28,52–54
PDMS-2 and BOT of Motor Proficiency and adequate con-
tent and construct validity demonstrated via item and fac-
tor analysis and goodness of fit tests, respectively.23 Con- DISCUSSION
current validity correlation coefficients on the M-ABC with The search strategies identified several performance-
other motor tests were moderate to excellent (r ⫽ 0.50 – based outcome measures that assess gross motor skills of chil-
1.00) in more than 50% of 13 studies, with 10 of the studies dren with DCD. The critical appraisal process has also indi-
of children with DCD/MI.24,29 – 40 The lowest correlation co- cated which tests are the most robust psychometrically and
efficients obtained between the M-ABC and other motor clinically. The studies identified that investigated outcome
tests occurred when the subjects had DCD and/or learning measures were nearly all at a low to moderate risk of bias, and
disabilities.33,35,38 Lower correlations also occurred when thus the results could be considered of acceptable quality.
the M-ABC scores were compared with very specific tests The M-ABC and TGMD-2 scored equally on overall
measuring multiple constructs of a particular skill.31,32,37,40 psychometric and clinical quality ratings. The M-ABC
The PDMS-2 and MAND also have moderate to excel- scored well due to demonstrations of adequate sensitivity
lent concurrent validity (r ⱖ 0.50) with other similar mo- and because it was the only measure to demonstrate re-
tor measures in studies with DCD/MI.30,36,41 Additionally, sponsiveness to change. The M-ABC has also had its psy-
the PDMS-2 has moderate to good concurrent validity cor- chometric qualities tested the most of all the measures in
relations coefficients with gross motor subtests (r ⫽ 0.63– the DCD population. Additionally, the M-ABC is quick to
0.75) in typical populations23,26 and adequate results on set up, administer, and score and requires minimal train-
factor analysis and goodness of fit tests to support its con- ing, making it a viable tool clinically. The suitability of the
struct validity.26 However, for gross motor components, M-ABC’s US normative data has also been investigated for
the MAND only correlated to a fair degree (r ⬍ 0.50) with other cultural groups in a number of studies.39,49,55–59 The
specific motor tests.25,42 In a single study, the TGMD-2 M-ABC, however, requires further support for its validity,
showed support for its concurrent/predictive validity with which should be considered when choosing tests. The M-
fair to good correlation coefficients on its subtests and total ABC also seems to have less stable concurrent validity
scores with a similar motor test27 (locomotion r ⫽ 0.63 and when testing children with DCD and learning disabilities.
object control r ⫽ 0.41, total scores r ⫽ 0.63). The The TGMD-2 has demonstrated good psychometric
TGMD-2 also has acceptable content validity (through ra- qualities27; however, it is important to note that the inves-
tionale, item analyses and discriminative tests) and con- tigations were only from the manual and were conducted
struct validity (through factor analysis).27,43,44 in children who are typically developing; thus, results have
BGMA Content (partially: expert opinion Internal consistency (all ⻫ 22 ⻫ 22 ⻫ 22 Partial: time admin 22 Partial: scoring 9
only)22 norms)22 difficulties22
Construct (all norms 7–12 y)22 Test-retest (special ed and
Abbreviations: BGMA, Basic Gross Motor Assessment; BOT-2, Bruininks-Oseretsky Test, 2nd ed.; DCD, developmental coordination disorder; MABC-2, Movement Assessment Battery for Children; MAND,
McCarron Assessment of Neuromuscular Development; MI, motor impaired; PDMS-2, Peabody Developmental Motor Scale, 2nd ed.; TGMD-2, Test of Gross Motor Development, 2nd ed.; ZNA, Zurich neuromotor
assessment.
Gen pop indicates the general population (representing typical children or unspecified group of children); phys disable, physically disabled; all norms, the entire sample used to establish normative data; part
norms, part of sample used to establish normative data; special ed and phys ed, special education and/or special physical education classes; interpret probs, interpretation problems from test scores for patients;
long admin, long administration time of the testing procedures; task probs, some tasks do not seem to be important for children; and training⫹, large amount of extra training required to use test efficiently. The
symbol ⻫ represents adequately achieved in the area; , did not adequately achieve in the area; and ?, unknown. Note: All the measures had normative data for their entire age groups and used ordinal
measurements.
Pediatric Physical Therapy Measures of Gross Motor Skills for Children With DCD 179