549
The Motor Control Assessment: An Instrument to Measure
Motor Control in Physically Disabled Children
Kathleen 0. Steel, MSc, Joan E. Glover, BSc(PT), Robert A. Spasoff, MD
ABSTRACT. Steel KO, Glover JE, Spasoff RA. The motor control assessment: an instrument to measure motor
control in physically disabled children. Arch Phys Med Rehabil 1991;72:549-53.
l The Motor Control Assessment has been developed as a research tool for the assessment of motor skills in physically
disabled children. In tests of concurrent validity it was found to be highly correlated with the results of the Physical
Abilities Chart (Pearson coefficient = .96), and satisfactorily correlated with ranking of motor ability assigned using
the clinical judgment of physiotherapists (Spearman rank coefficients = .63 to -97 in diagnostic subgroups). Estimates
of both interrater and intrarater reliability were also high, with intraclass coefficients above .95.
KEY WORDS: Motor skills; Physical thempy; Research
Evaluating a treatment mode or medical facility for disabled
children is complicated by the heterogeneity of the population,
in terms of both diagnosis and level of disability. Even within
the most common diagnosis, cerebral palsy, there is enormous
variability in the distribution and extent of the expressed
disability.
Program evaluation frequently requires an assessment which
can be applied to groups of children exhibiting such diversity.
Before 1980 most instruments were inappropriate as measures
of outcome in studies of this kind, often because they consisted
of a narrow rangele3 or small sample2T4of motor skills. Al-
though multipoint rating scales were sometimes sug-
gested, 1*3.5-7instruments often used insensitive, dichotomous
ratings for items. 2*4*8*9
Few instruments were tested for validity
and reliability.3*9*‘0 Those which assessed developmental
age 2,4*9~11were unsatisfactory for children with cerebral palsy,
whose development may be disordered12 rather than simply
delayed.
For ten years an increasing number of papers has stressed
the need for carefully developed tools,13 which are beginning
to appear. Russell and colleagues14 recently presented their
Gross Motor Function Measure for children with cerebral palsy.
It is a measure of independence in completing 85 tasks, with
excellent reliability and demonstrated responsiveness to change.
THE MOTOR CONTROL ASSESSMENT
The Motor Control Assessment (MCA) was developed in
response to the need for an instrument which would measure
the outcomes of alternate treatment approaches in a population
From the Department of Epidemiology and Biostatistics, University of Western
Ontario. London, Ontario (Ms. Steel); the Arbutus Society for Children, Victoria,
British Columbia (Ms. Glhver); and’ihe Department of fipidemiology and Com-
munity Medicine, University of Ottawa, Ottawa, Ontario, Canada (Dr. Spasoff).
This project was conducted from the Department of Epidemiology and Com-
munity Medicine, University of Ottawa, and was supported by Demonstration Model
Grants 296 and 677 of the Ontario Ministry of Health.
Submitted for publication July 12, 1989. Resubmitted in revised form April 19,
1990. Accepted in revised from August 20, 1990.
No commercial party having a direct or indirect interest in the subject matter of
this article has conferred or will confer a benefit upon the authors or upon any
oreanization with which the authors are associated.
Reprints will not be available.
of physically disabled children.15 The assessment, as tested in
this study, consists of 113 test items ranging from early infant
skills to activities requiring the coordination of an able-bodied
five-year-old child. Items were drawn from Bobath’s First As-
sessment from the Western Cerebral Palsy Center (unpub-
lished) and from the instruments of Croslands and Miller and
associates.5
Wherever possible, activities were broken down into basic
contributing postures, following the concept of Bobath and
Bobathi2 that movements are effective only if they can be
supported and controlled by appropriate postures. Performance
on right and left sides was assessed separately. No attempt
was made to establish developmental levels for test items be-
cause of the difficulties in using such levels in children whose
development is disordered, and because of the problems in
validating age norms.
The grading scale for each activity is based on that of Footh
and Kogan,3 with the addition of a level for activities which
cannot be achieved passively because of muscle contractures
or deformity. The result is a six-point scale. A five-point scale
is used for automatic reactions. The average of item scores
represents the overall status of a child with respect to motor
control. The complete assessment, scoring systems, and man-
ual are available from the authors. Selected items are illus-
trated in table 1.
The MCA is designed for use by physical therapists with
experience in the normal and abnormal development of chil-
dren. It is appropriate for children with physical handicaps
ranging from mild to severe, but it is not appropriate for clumsy
children. It is not intended to be used for children less than
two years of age (for whom a developmental scale is more
appropriate).
The assessment requires a treatment room with an adjacent
staircase, a few toys, a mat, a stool, a child’s chair, an adult
chair, a table, a tricycle, and a rope. Administration time is
between 30 and 60 minutes, depending on the level and co-
operation of the child.
The MCA is designed primarily as a measure of motor con-
trol, rather than functional ability. Since the purpose of or-
thotic aids frequently is to increase level of function or adaptation
at a given level of motor control, and since abilities can often
Arch Phys Med Rehabil Vol72, July 1991
550 MOTOR CONTROLASSESSMENT,Steel

Table 1: Sample items from the Motor Control Table 3: Agreement Between Orlglnal and Revised
Assessment. Diagnoses of Partlclpatlng Children
Item score 0 1 2 3 4 5 Revised diagnosis
SUPINE (11 items) Spastic Nonspastic
2. Rolls to side cerebral Cerebral Spina
8. Can grasp and manipulate toy (R) hand Original diagnoses palsy Palsy biflda Other
PRONE (18 items) Spastic cerebral
13. Forearm support,head up palsy 70 0 0 1
26. Crawls reciprocally Nonspastic cerebral
SI’lTING (30 items) palsy 5 18 0 3
31. Gets to sitting over (R) side Spina bifida 0 0 28
33. Ring sits with arm support Other 0 0 0 3:
42. Can obtain small object by raking (R)
44. Pivots in sitting
KNEELING(7 items) pletely paralytic muscles (as in the case of some children with
62. Knee walks alone spina bifida), where physiotherapy is focused on improving
STANDING/STANDING UP (18 items) the motor performance of muscles other than the paralyzed
67. Stands from stool without using arms
76. Stands with feet parallel ones.
80. Picks up object from floor This study reports the validity, intrarater reliability, and in-
83. Stands on (R) leg 5 seconds terrater reliability of the MCA.
WALKING (10 items)
85. Cruises at furniture
86. Walks unaided on floor METHODS
92. Walks upstairs alternate feet
GENERAL COORDINATION(14 items)
97. Walks backward Subjects
105. Pedals a tricycle
107. High jumps, 6” The study was conducted simultaneously at two children’s
Scoring treatment centers (referred to as center A and center B) in
5 = Child performs in a normal way. eastern Ontario. Those eligible were children aged 24 months
4 = Child performs activity independently but slowly and awkwardly. to 17 years who experienced a permanent motor deficit; mental
3 = Child performs activity with minimal assistance.
2 = Child performs activity but requires maximal assistance. retardation did not exclude a child from the study.
1 = Child cannot perform activity even when given the advantage of posi- Lists of eligible children were provided by the treatment
tioning and assistance. centers. Subjects were classified into the following four di-
0 = Cannot be achieved passively. agnostic groups: spastic cerebral palsy; cerebral palsy other
than spastic; spina bifida; and other diagnoses, which included
change dramatically through the use of aids alone, children other neuromuscular disorders, congenital anomalies, and in-
should generally be examined without them. An exception to juries. It was intended that each group would consist of ap-
this general rule is the use of aids which compensate for com- proximately 40 children, giving a total sample size of 160.
However, most children were found to have spastic cerebral
Table 2: Diagnoses of Chlldren In Study Sample palsy, and final sample sizes reflected the distribution of di-
as Provided by the Treatlng Theraplst (n[%])
agnoses among children at the treatment centers.
Spastic cerebral palsy 75 (46.6)
Included in the analysis were 161 children-75 with spastic
Quadriplegic 31 (19.3)
Diplegic 28 (17.4) cerebral palsy, 18 with cerebral palsy of a type other than
Hemiplegic 13 (8.1) spastic, 28 with spina bifida, and 40 with other diagnoses. In
Paraplegic 2 (1.2) children with spina bifida, the affected neurosegment levels
Monoplegic 1 (0.6) ranged from T5 to Sl; 16 (57.1%) of this group had hydro-
Nonspastic cerebra1 palsy 18 (11.2) cephalus. Diagnoses of the children and age and gender dis-
Athetoid 8 (5.0) tribution are presented in tables 2, 3, and 4.
Ataxic 4 (2.5)
Mixed* 6 (3.7) The diagnoses of nine children were changed during the
Spina bifida 28 (17.4)
Table 4: Age and Gender Dlstrlbutlon of Children
Myelomeningocele
Thoracict 6 (3.7) In Four Dlagnostlc Groups @I[%])
Lumbar 17 (10.6) Spastic Nonspastic
Cerebral Cerebral Spina
Other Palsy Palsy bifida Other
Thoracic 2 (1.2)
Lumbar 3 (1.9) Age (yrs)
2-6 52(69.3) 6(33.3) 16(57.1) 25(62.5)
Other diagnoses 40 (24.8) 7-11 12(16.0) 6(33.3) 12(42.9) 12(30.0)
Other neuromuscular disorders 26 (16.1) 12-17 ll(14.7) 6(33.3) 3 (7.5)
9 (5.6) fJ(-)
Congenital and genetic anomalies Mean age 6.4 8.8 5.5 5.9
Traumatic injury 5 (3.1) Gender
Total 161 Male 33(44.0) 7(38.9) 1l(39.3) 26t65.0)
*These children show signs of more than one form of cerebra1 palsy; + children Female 42(56.0) ll(61.1) 17(60.7) 14(35.0)
were classified by their highest level of involvement. Total 75 18 28 40

Arch Phys Med Rehabil Vol72, July 1991

 MOTOR CONTROLASSESSMENT, Steel
course of the study; four children with “other diagnoses” had
originally been classified with spastic (n = 1) or nonspastic
(n = 3) cerebral palsy. Five children with spastic cerebral palsy
had originally been diagnosed with cerebral palsy other than
spastic. Table 3 shows original and revised diagnoses. The
revised diagnoses are considered more accurate, and are re-
ported in table 2 and in most results. However, the data for
the ranked analysis (see below) had been collected before re-
classification, and these results must; of necessity, be pre-
sented according to the original diagnoses.
All 161 children participated in the validation phase. Reli-
ability was tested in a random subsample of 35 of the children
as follows: nine with spastic cerebral palsy, six with cerebral
palsy other than spastic, and ten from each of the other two
diagnostic groups. This subsample consisted of 16 boys (45.7%)
and 19 girls (54.3%); age distribution was similar to that of
the total sample.
Therapists’ Training
An initial training session was held to familiarize partici-
pating physiotherapists with the MCA and the conduct of the
study. Therapists were asked to use the assessment for three
or four children not included in the formal study, and to record
any problems in definition or procedure which they experi-
enced. Items which were unclear were clarified in the assess-
ment manual.
Validity Testing
In the absence of a “gold standard” to measure motor con-
trol, we selected the Physical Abilities Cha@ and rankings
of motor ability, assigned using the clinical judgment of phy-
siotherapists, against which to test concurrent validity. Similar
in format to the MCA, the Physical Abilities Chart consists of
77 skills representing low, medium, and high levels of motor
function. A four-point rating scheme is used.
Rankings were determined by the physiotherapists working
with the children. The population was first stratified by treat-
ment center, so that children were judged only by therapists
who worked with them, and then by diagnostic group, so that
children would be compared only against others with similar
kinds of disability. The result was eight strata, each of which
was examined separately in this analysis. Within each stratum,
children were categorized according to ambulatory skills using
the following four levels: nonwalker, physiologic walker,
household walker, and community walker.” Within each level,
therapists used their clinical judgment to rank children from
lowest to highest level of overall control, working until con-
sensus was reached. Because level of ambulation was ordered,
ranking within levels automatically ranked all children within
their diagnosis-center stratum.
All rankings were done before any testing. Each child was
then assessed in a single session by his treating therapist using
the MCA, followed by the Physical Abilities Chart.
Reliability Testing
For each child participating in the reIiabi1ity phase, two
appointments were booked seven to ten days apart. The child’s
551
usual treating therapist and an observing therapist completed
both the MCA and the Physical Abihties Chart in the first
session. The two therapists graded activities simultaneously;
no communication between them was permitted during testing.
The child’s usual therapist was responsible for all positioning
and instructions to the child. In the second session the child’s
usual therapist repeated the MCA.
Calculation of Scores
A Motor Control Assessment score was the average of all
items of the MCA, with items assessing head righting and
protective reactions adjusted (X 5/4) to weight them equally
with other items. Items for which ratings were missing were
excluded from the calculations. The range of possible scores
was 0 to 5.
A Physical Abilities Chart score was the average of all items
of the Physical Abilities Chart. Missing values were excluded
from the calculations. The range of possible scores was 1 to
4.
Calculation of Validity and Reliability Coeffkients
A Pearson product-moment correlation was used to deter-
mine the correlation between MCA and Physical Abilities Chart
scores. The main analysis used results of the therapists’ as-
sessments of the children they usually treated; these were
available for all children. Additional results were provided by
the observing therapist for children included in the reliability
phase. A Spearman rank correlation was performed on MCA
scores and the rankings assigned by physiotherapists.
The intraclass correlation coefficientI was used to estimate
the reliability of assessment scores. By using the intraclass
correlation, consistency between observers (which may inflate
an interclass correlation coefficient such as the Pearson) is not
mistaken for reliability. In this study, calculations were based
on a one-way ANOVA, resulting in an intraclass correlation
which regards the study therapists as a random sample of ther-
apists working with these children. The intraclass correlation
was also used to examine the reliability of individual items.
Intrarater reliability was estimated from the scores of the
treating therapist during the first and second sessions. Inter-
rater reliability was estimated from the scores of the pairs of
therapists observing children simultaneously. A further test of
interrater reliability used the scores of the observing therapist
in the first session and the treating therapist in the second
session. This analysis reduces the influence that one therapist
may have had over another during the testing situation.
RESULTS
MCA Scores
As desired, the study included children with a wide range
of motor abilities; scores ranged from 0.16 (extremely low) to
5.00 (normal) (table 5).
Arch Phys Med Rehabll Vol72, July 1991
552 MOTOR CONTROLASSESSMENT,Steel

Table 5: Dlstrlbution of Motor Control Assessment Scores Table 7: Spearman Correlation Coefficients Between
Among 161 Study Children Motor Control Assessment Scores and Ranks
scores n (%) Center A Center B
0 - .49 4 (2.5) Spastic cerebral palsy* .94 (54) .94 (17)
.50 - .99 6 (3.7) Cerebral palsy other than spastic .93 (13) .87 (13)
1.00 - 1.49 7 (4.3) Spina bifida .97 (17) .85 (11)
1.50 - 1.99 16 (9.9) Other diagnoses .91 (24) .63 (12)
2.00 - 2.49 21 (13.1) *Original diagnostic groups are used in thisanalysis.
2.50 - 2.99 16 (9.9)
3.00 - 3.49 26 (16.2)
3.50 - 3.99 28 (17.4) correlation coefficients for the former ranged from .67 to .99;
4.00 - 4.49 21 (13.1) 77% were .90 or higher, and all but three, or 97%, were higher
4.50 - 5.00 16 (9.9) than .75. Interrater reliability coefficients ranged from .24 to
.98; 41% were .90 or higher, and 80% were higher than .75.
Validity of the MCA Score Only one item had a value below .40.
A third set of correlations was performed from the ratings
Correlation coefficients between the MCA and Physical of the observing therapist on the first occasion and the treating
Abilities Chart were more than .9 (table 6) for the entire sam- therapist on the second occasion. Intraclass correlation coef-
ple and within each of the four diagnostic groups. Results from ficients were lower again, as expected in this stringent test.
the MCA and Physical Abilities Chart completed by the ob- However, 72% were higher than .75, and, again, only one
serving therapists in the reliability study also yielded a cor- was below .40.
relation coefficient of .96. The following three items had intraclass correlation coeffi-
Comparison of MCA scores and therapists’ rankings are cients less than .50 in one of the three analyses: SUPINE: Can
presented in table 7. Diagnostic groups do not coincide with reach for and attain toy: (R) hand; SUPINE: Can reach for
those given by the treating therapist (table 2) because of the and attain toy: (L) hand; and PRONE: Forearm support: head
reclassifications previously mentioned. High Spearman coef- up. These items were felt to be important and were retained;
ficients were obtained for all diagnostic groups at center A an improved description of grades was incorporated into the
and for all groups at center B except children of “other” manual.
diagnoses.

Reliability of the MCA Score
Mean scores and standard deviations were 3.04 (1.18) for
the primary therapist at the first observation, 2.97 (1.14) for
the observing therapist, and 3.09 (1.16) for the primary ther-
apist at the second observation. Differences were not signifi-
cant by ANOVA.
Intraclass correlations yielded an intrarater reliability coef-
ficient of .99 and an interrater reliability coefficient of .97.
Correlation of results of the observing therapist at the first
examination with the results of the usual treating therapist at
the second examination yielded an intraclass correlation coef-
ficient of .96.
Reliability of MCA Items
Intrarater reliability was assessed from the item ratings of
the same therapist on two occasions, and interrater reliability
from the ratings of different therapists on the same occasion
using intraclass correlations. Predictably, intrarater reliability
was found to be higher than interrater reIiability. Intraclass
Table 6: Pearson Correlation Coefficients Between
Motor Control Assessment and Physical Abilities Chart
II
All children 160
Spastic cerebral palsy 75
Cerebral palsy other than spastic 18
Spina bifida’ 27
Other diagnoses 40
*For one child the Physical Abilities Chart could not be done.
DISCUSSION
In tests of validity, MCA scores correlated well with results
of the Physical Abilities Chart. The MCA contains a wider
range of skills and uses a six-point, rather than a four-point,
rating scale; for these reasons it is possibly a more sensitive
instrument than the Physical Abilities Chart. Ranked MCA
scores correlated well with rankings of overall motor control
judged by physiotherapists. The lowest correlation was seen
in the mixed group of children with diagnoses other than ce-
rebral palsy and in the group of children with spina bifida,
which was the most difficult group to rank because of their
heterogeneity.
The MCA appears to be a reliable measure of motor control
in physically disabled children. Because one therapist was re-
sponsible for positioning and instructing a child while a second
therapist only observed, it may be said that true interrater
reliability was not tested. This is a valid criticism. The as-
sessment was lengthy and tiring for many of the children;
because there are a multitude of potential factors which can
affect a child’s performance at different times, we felt that it
was necessary to assess interrater reliability in a single session.
It can also be said that the two therapists influenced each other
during the assessment. However, the results of different ther-
apists at different observations produced an equally high coef-
Correlation ficient, which indicates that scores are both reliable and stable
coefficient
for an individual over a short period of time.
.96
.97
Ratings of individual items were naturaIly less reliable. Fleiss19
.96 considers coefficients above .75 to represent excellent reli-
.96 ability, those between .40 and .75 to represent fair to good
.95 reliability, and those below .40 to be poor. All but three items
had excellent intrarater reliability, and 80% had excellent in-

Arch Phys Mad Rehabll Vol72, July 1991

 MOTOR CONTROLASSESSMENT, Steel
terrater reliability. The largest proportion of disagreements was
seen between grades 1 (unable to perform) and 2 (requires
maximal assistance), and between grades 2 and 3 (requires
minimal assistance). It would seem, therefore, that these are
difficult judgments for therapists to make. Better definitions
of maximal and minimal have been incorporated into the man-
ual in an attempt to improve intrarater agreement. With respect
to interrater reliability, it may be that one therapist is simply
unable to assess the amount of assistance provided by another.
The stability of scores, in spite of item discrepancies, makes
it imperative that the assessment be applied in its entirety and
that scores, rather than items, be used as outcome measures.
Missing items, through exclusion, are given the average
score of nonmissing items. If one or two items are missing,
the effect on the child’s overall score is small. However, the
possibility of bias increases as the number of missing items
increases, and investigators should make every attempt to see
that assessments are complete.
The MCA is a measure of quantity of independent activity,
rather than quality of movement. Many participating therapists
felt that this was a weakness. Given that the goals of treatment
are to make significant improvements in the physical status of
disabled children, we felt that it was first necessary to develop
a measure of the extent of motor performance. A measure of
the quality of movement is presently being developed (Boyce,
personal communication) as an extension of the Gross Motor
Performance Measure. l4
It is intended that multiple measures of the MCA will enable
documentation of the changes in a child and that change scores
can be analyzed as measures of outcome in evaluative studies.
The numeric score derived from the MCA does indicate the
status of a child-the child who is able to do very little achieves
a very low score, and the child who has nearly normal control
will be assigned a high score. However, the scale is ordinal,
rather than interval, with respect to both the overall score and
gain score. In any study of the efficacy of treatment, therefore,
it is imperative that study and comparison children be matched
as closely as possible with respect to age, diagnosis, and level
of ability if change scores are to be compared in a meaningful
way.
CONCLUSION
In designing this study it was our belief that the instrument
must first be shown to produce a meaningful and reliable mea-
sure of a significant construct, motor control. In this respect,
the results have been satisfactory for the MCA. In addition,
the MCA was able to demonstrate significant improvement in
children receiving increased time in therapy compared to a
control group. l5 Although the study was admittedly imperfect,
its results support the potential of the MCA as a measurement
of outcome. Its performance as an outcome measure and its 17. Bleck EE. Orthopedic management of cerebral palsy. Philadel-
capacity and sensitivity in detecting change, especially within
diagnostic subgroups, should be further examined.‘O
Acknowledgments: We extend our thanks to Dr. Jean Alexander, whose
original project to evaluate alternate forms of treatment for disabled children
provided the impetus for the development of this measure. We wish to thank
Sophie Levitt for her permission to use the Physical Abilities Chart, and to
express our appreciation to the physiotherapists who performed the 200 as-
sessments: Patricia Ahearn, Lin Arthur, Freda Bailey, Diane Caulfeild, Linda
553
Clark, Janine Dudding, Linda Kealey, Gail Machin, Linda Martell, Helen
McConnachie, Sandy Nicholson, Jill Nowell, and Francooise Rainville. We
are grateful to Rena McEwen, who tabulated data for us, and especially to
our research assistant, AndrCe Casabon-Hotz.
References
1. Hoberman M, Cicenia EF, Stephenson GR. A useful measure-
nent tool in the physical rehabilitation program of preschool
xthopedically handicapped children. Arch Phys Med 1951;32:456-
51.
2. lohnson MK, Zuck FN, Wingate K. The motor age test: mea-
surement of motor handicaps in children with neuromuscular
disorders such as cerebral palsy. J Bone Joint Surg [Am]
1951;33:698-707.
3. Footh WK, Kogan KL. Measuring the effectiveness of physical
therapy in the treatment of cerebral palsy. Phys Ther 1%3;43:867-
73.
4. Ingram AJ, Withers E, Speltz E. Role of intensive physical and
occupational therapy in the treatment of cerebral palsy: testing
and results. Arch Phys Med Rehabil 1959;40:429-38.
5. Miller AS, Stewart MD, Murphy M-A., Jantzen AC. An evalu-
ation method for cerebral palsy. Am J Occup Ther 1955;9:105-
11.
6. Semans S, Phillips R, Romanoli M, Miller R, Skillen M. A
cerebral palsy assessment chart: instructions for administration
of the test. Phys Ther 1965;45:463-8.
7. Zausmer E. The evaluation of motor development in children.
Phys Ther 1964;44:247-50.
8. Crosland JH. The assessment of results in the conservative treat-
ment of cerebral palsy. Arch Dis Child 1951;26:92-5.
9. Hoskins TA, Squires JE. Developmental assessment: a test for
gross motor and reflex development. Phys Ther 1973;53:117-
26.
10. Semans S. Specific tests and evaluation tools for the child with
central nervous system deficit. Phys Ther 1965;45:456-62.
11. Milani-Comparetti A, Gidoni EA. Routine developmental ex-
amination in normal and retarded children. Dev Med Child Neu-
rol 1967;9:631-8.
12. Bobath B, Bobath K. Motor development in the different types
of cerebral palsy. London: William Heineman Medical Books
Ltd, 1975.
13. Berk RA, DeGangi GA. Technical considerations in the evalu-
ation of pediatric motor scales. Am J Occup Ther 1979;33:24&
4.
14. Russell DJ, Rosebaum PL, Cadman DT, Gowland C, Hardy S,
Jarvis S. The gross motor function measure: a means to evaluate
the effects of physical therapy. Dev Med Child Neurol
1989;31:341-52.
15. Steel KO, Spasoff RA, Alexander JO. The satellite clinic: a
model for the treatment of handicapped children in towns and
rural areas. Dev Med Child Neural 1985;27:355-63.
16. Levitt S. Treatment of cerebral palsy and motor delay. Oxford:
Blackwell Scientific Publications, 1977.
phia: Saunders, 1979.
18. Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing
rater reliability. Psycho1 Bull 1979;86:420-8.
19. Fleiss JL. The design and analysis of clinical experiments. New
York: Wiley, 1986.
20. Guyatt G, Walter S, Norman G. Measuring change over time:
assessing the usefulness of evaluative instruments. J Chronic Dis
1987;40:171-8.
Arch Phys Med Rehabll Vol72, July 1991