Test-Retest Reliability of the Gross Motor Function ...

Phys Occup Ther Pediatr Downloaded from informahealthcare.com by University of Washington on 08/18/14 For personal use only.

CLINICAL CONCERNS

Test-Retest Reliability of the Gross Motor Function Measure in Children with Cerebral Palsy Kristie F. Bjornson Catherine S. Graubert John F. McLaughlin Cheryl I. Kerfeld Elaine M. Clark

Kristie F. Bjornson, MS, PT, PCS, is Research Coordinator, Spasticity Management Clinic, Children’s Hospital & Regional Medical Center. Catherine s. Graubert, PT, Senior Research Physical Therapist, Spasticity Management Clinic, Children’s Hospital & Regional Medical Center. John E McLaughlin, MD, Professor of Pediatrics, University of Washington; Director ot the Neurodevelopmental Program, Children’s Hospital & Regional Medical Center; and Director o f the Clinical Training Unit, Center on Human Development arid Disability, UW. Cheryl I. Kerfeld, MS, PT, is Research Physical Therapist, Spasticity Management Clinic, Children’s Hospital & Regional Medical Center. Elaine M. Clark, MPT, is Research Physical Therapist, Spasticity Management Clinic, Children’s Hospital & Regional Medical Center. Address Correspondence to: Kristie F. Bjornson, MS, PT, PCS, Children’s Hospital & Regional Medical Center, 4800 Sand Point Way NE, Seattle, WA 98105. This research was supported by: National Institute of Neurological Disorders and Stroke, NIH NINDS R01-27867. Physical & Occupational Therapy in Pediatrics, Vol. 18(2) 1998 0 1998 by The Haworth Press, Inc. All rights reserved.

51


52

PWSICAL c ! OCCUPATIONAL THERAPY IN PEDIATRICS

ABSTRACT. The purpose of this study was to describe the test-retest reliability of the Gross Motor Function Measure (GMFM) in children with cerebral palsy. The GMFM was administered twice, within a oneweek time period, to twenty-one children with cerebral palsy. The time of day, therapists, and evaluation setting were held constant. Intra-class correlations ranged from .76 to 1.00. The data suggest that the GMFM is consistent in the measurement of gross motor skills and that children with cerebral palsy exhibit stable gross motor skills during repeat measurement. This information will be helpful in future analysis of treatment efficacy using the GMFM as an outcome measure. (Article copies available for a fee Jkoin The Haworth Docwnent Delivery Service: 1-800-342-9678.E-mail address: getiff?@kawort~ipressitic.coti~]

INTRODUCTION Pediatric therapists use a variety of outcome measurement tools to document treatment efficacy in clinical and research settings. These outcome measures need to be consistent with their intended purpose. Kirshner and Guyattl suggest a framework based on the following applications of outcome measures. A discriminative measure, such as the Bayley Developmental Scales,2 is indicated to discriminate between subjects on a particular characteristic. A predictive measure is designed to estimate future outcome or prognosis. For example, the Bleck3 scale predicts ambulation skills at age seven based on the presence of postural and tonic reflexes in a child. Evaluative measures are used to measure the amount of change in a characteristic over time. Wright and Nicholson4 developed and applied an evaluative measure of motor skills to study the effect of physical therapy on children with spastic cerebral palsy (CP). The criterion-referenced measure designed by Wright and Nicholson would not be appropriate to ascertain whether or not a child’s motor skills were within the normal range. Likewise it does not have the ability to predict future level of motor skill. The Gross Motor Function Measure (GMFM)5 is an example of an evaluative tool designed to document changes in motor function over time in children with CP. The development of an evaluative instrument requires an examination of the test-retest reliability or, in other words, the within-patient differences unrelated to true within-patient change.6 This is necessary in order to differentiate between true clinical changes and the variability in performance of the patient or the measurement tool. From a measurement perspective, the reproducibility or


Clitzicul Cotzcerris

53

consistency of scores is a reflection of the measurement tool’s reliability. The numerous sources of variation, which may affect the consistency or reproducibility of scores, collectively known as the total variance. The sources of variation in scores can be due to subject performance, observers, and/or random test error.7 This paper will focus on the variability due to subject performance, also known as the ‘test-retest’ reliability of the measurement tool. The GMFM is a standardized observational tool designed to measure change in motor skills over time in children with cerebral p a l ~ y . ~ . ~ Criterion-referenced, the GMFM was developed to assess how much of a motor task a child can complete (quantity) rather than how well it is performed (quality). The GMFM consists of 88 items, which cover five dimensions of motor function. These five dimensions are lying/ rolling, sitting, crawling/kneeling, standing, and walk/run/jump. Each of the 88 items is scored on a 4 point Likert scale of zero to 3. The categories assigned these values range from: 0 = does not initiate, 1 = initiates (< 10% task), 2 = partially completes (10 to < 100% task), and 3 = completes an activity. An example of an item from the lying/rolling dimension is noted in Figure 1. Item ## 15 has a starting position of prone with arms and legs comfortably positioned and preferably with the head down. The child is encouraged to roll to supine over the left side either by request or by following demon~tration.~ The test materials needed to administer the GMFM are commonly available in a pediatric therapy department and the authors designed the measure to be administered by experienced pediatric physical therapists. The specific criteria for scoring each item are given in the manual, and the authors state that the manual is essential for scoring. Each item of the GMFM is administered up to a maximum of three trials with the best performance scored in order to elicit the child’s optimal perforFIGURE 1. GMFM Item #15. Prone: Rolls to Supine Over Left Side 0. does not initiate rolling 1. initiates rolling 2. rolls part way to supine

3. rolls to supine over left side


54

PHYSICAL & OCCUPATIONAL THERAPY IN PEDIATRICS

mance. Scoring the GMFM requires a tally of each item’s scores. A percent score for the dimension is derived by dividing the obtained raw score by the possible raw score. For example, if a child received a raw score of 33 in the lyingholling dimension out of a possible raw score of 51, the percent score for that dimension would be 65%. A total percent score is calculated by adding the percent scores of the five dimensions together and dividing by five.5 A five year old child with normal gross motor ability is expected to score a 3 on all 88 items. Since there are no ‘gold standard’ evaluative measures for children with CP, the responsiveness of the GMFM was addressed by correlating the tool’s measurement of change over several months’ time to the independent judgements of change in motor function by parents, treating therapists, and therapists unfamiliar with the children via videotape e ~ a l u a t i o n .The ~ study sample consisted of 111 children with cerebral palsy, 25 with head injury and 34 who were non-disabled. A priori hypotheses of correlation were that the correlation between change on the GMFM and non-treating therapist’s judgements would be the greatest of the three comparisons. The correlation between the GMFM and videotape judgements of change of the treating therapists would be intermediate, with the parents’ judgement of change having the lowest correlation with the GMFM. Total GMFM scores correlated with non-treating therapist video evaluation at r = 0.82, the treating therapists’ judgement at r = 0.65 and the parents’ judgement at r = 0.54. All were significantly different at p < .05. The authors report inter-rater reliability coefficients ranging from 0.87 to 0.99 for the five dimensions and total scores. On test-retest administration to ten children with cerebral palsy, intra-rater coefficients of .99 were found for all dimensions and total percent scores, except for the standing dimension which was .92.s The GMFM has been used as an outcome measure to address efficacy of treatments for children with cerebral The baseline variability of motor skills in children with CP must also be accounted for when attempting to document treatment efficacy. The purpose of this study was to describe the test-retest reliability of the GMFM in children with cerebral palsy within a one-week time period. This protocol uses procedures suggested by Tuckerman13 to demonstrate that a measurement tool is reliable or consistent. The design addresses

55

Clinical Coricrriis

test-retest reliability as well as the variability of the child’s motor skills over a one week time period.


METHODS The GMFM was prospectively administered to twenty-one children with spastic cerebral palsy twice, one week apart, by the same therapist. Therapist, time of day, and evaluation setting were held constant for each child. Practice effect and familiarity with test items was judged to be minimal since all motor activities tested could be part of the subjects’ daily activities. The two participating therapists achieved and maintained a > 90% point by point agreement of inter-rater reliability for the study period by viewing a minimum of six GMFM administration videos prior to data collection. The therapists were trained by and established inter-rater reliability with a ‘gold standard’ senior physical therapist as part of a randomized clinical trial of the efficacy of selective dorsal rhizotorny.l4 The two participating physical therapists have four and ten years of pediatric experience, respectively. The children were recruited from an outpatient clinic that focused on the management of spasticity. Thirteen children had spastic diplegia and eight had spastic quadriplegia (see Table 1). The average age of the children with diplegia and quadriplegia was 7.5 years (4.4-15.4 yr.) and 9.3 years (4.5-17.7 yr.) respectively. The average age of the whole sample of children was 8.3 years (4.4-17.7 yr.). Nine of the TABLE 1. Subject Characteristics

N

Mean Age Independent Gender Baseline (Range) Arnbulator (male) GMFM Total % (range)

Spastic Diplegia

13

7.4 (4.4-15.4)

9

5

88.5 (48-97)

Spastic Quadriplegia

8

9.3 (4.5-17.7)

0

3

48.1 (22-67)

Total

21

8.3 (4.4-17.7)

9

8

68.3 (22-97)


56


children with spastic diplegia were independent ambulators while none of the children with quadriplegia were independent ambulators. Independent ambulators were defined as children able to ambulate 50 feet without orthoses or assistive devices. Five children with diplegia were male while three with quadriplegia were male. The children with diplegia and quadriplegia had mean baseline GMFM percent total scores (range) of 88.5 (48-97) and 48.1 (22-67) respectively. The mean baseline GMFM percent total score (range) for the total sample was 68.3 (22-97).

ANALYSIS To analyze test-retest reliability of the scores, an intraclass correlation coefficient (ICC model 1,1) was calculated from a one-way analysis of variance A coefficient of 0 indicates no reliability/ consistency of motor performance and a coefficient of 1 indicates perfect reliability/con~istency.~

RESULTS The intraclass correlation coefficients for the total sample of 21 children are shown in Table 2. The correlations of the percent scores for the five dimensions and total percent scores ranged from .80 to 1.00 (p < .001). The thirteen children with spastic diplegia are presented in Table 3. The correlations ranged from .80 for the sitting TABLE 2. Correlations for the Total Sample (n = 21)

GMFM Dimension

I

cc

P

Lyinglrolling

0.80

0.000

Sitting

0.99

0.000

Crawling/kneeling

0.99

0.000

Standing

0.99

0.000

Wal klrunljump

0.99

0.000

Total

1.oo

0.000

57

Clitiical Coricertis


TABLE 3. Correlations for Children with Spastic Diplegia (n = 13) GMFM Dimension

ICC

P

Lying/rolling

1.oo

0.000

Sitting

0.80

0.001

Crawling/kneeling

0.97

0.000

Standing

0.98

0.000

Wal k/run/jurnp

0.98

0.000

Total

0.99

0.000

dimension to 1.00 for the lying/rolling dimension (p c .OOl). The correlation data for the eight children with quadriplegia are shown in Table 4. The correlation ranged from .76 for the lyingjrolling dimension to .98 for sitting, crawlingkneeling, and total percent scores Op < .02).

DISCUSSION The GMFM exhibits test-retest reliability for this sample of children with CP that is sufficiently robust to suggest that the generalization to other samples is appropriate. Also, these results support the premise that children with spastic diplegia and spastic quadriplegia exhibit consistent performance of gross motor skills as measured by the GMFM over a one week time period. The children with spastic quadriplegia exhibited somewhat less consistency in lyingholling skills. The older children with quadriplegia may not have been as cooperative for motor skills that did not match their chronological ages (e.g., pivot prone for a 17 year old). The children with spastic diplegia were most variable in sitting skills. Test reliability has been documented for children with typical motor development. Test-retest reliability of the gross motor items of the Peabody Developmental Motor Scales (PDMS)17 on 38 typically developing children between one and 83 months of age were reported at correlations o f .94 for items administered. The Bruinincks-Oseretsky Test of Motor Proficiency (BOTMP)18 documented test-retest correlations of .64 to .80 for the four subtests of the gross motor section with

58

PHYSICAL & OCCUPATIONAL T H E M PY IN PEDIATRICS


TABLE 4. Correlations for Children with Quadriplegia (n = 8)

GMFM Dimension

ICC

P

Lying/rolling

0.76

0.021

Sitting

0.98

0.000

Crawling/kneeling

0.98

0.001

Standing

0.88

0.002

Walk/run/jump

0.81

0.014

Total

0.98

0.000

a composite correlation of .77 in 63 typical second graders. Sixtythree typical sixth graders were found to have correlations ranging from .49 to .89 with a gross motor composite of 3 5 . The correlations for the 126 typically developing children were .81 for gross motor skills as measured by the BOTMP. The data presented suggest that the consistency of motor performance in children with cerebral palsy as measured by the GMFM is similar to the normative populations’ consistency as measured by the PDMS and the BOTMP. Therapists and caregivers often describe considerable day to day variability in the motor skills of children with cerebral palsy. At times a child may walk securely while at other times the same child falls frequently or may be unable to get up from the floor without assistance. This variability in performance can result from many factors, such as stress, illness, anxiety. It can also be ascribed to fatigue during activities such as long distance ambulation or prolonged static standing. The early work of Campbell and Bell19 supported the influence of fatigue, when they documented greater energy consumption during ambulation for children with cerebral palsy as compared to typically developing children. In an assessment of the effect of orthoses on energy expenditure, Mossberg and colleaguesz0 noted three to four times greater than normal physiological cost index values for children with cerebral palsy. Using a measure of heart rate, Bleck and Alder21 reported a 2.5 times greater energy cost expenditure in children with spastic diplegia than their healthy counterparts. Children with spastic diplegia cerebral palsy have also been documented to walk at half the speed of healthy


Clinical Concerns

59

~ h i l d r e n .The ~ ~ consistency >~~ of motor performance of children with cerebral palsy may be influenced by increased energy expenditure and/or fatigue over time. This variability in motor performance could influence the ability of evaluative measures to accurately discern true treatment effects. The variability in motor performance noted by clinicians and caretakers in children with cerebral palsy may be a function of endurance over smaller time periods (e.g., at the end of a school day compared to early in the morning). It may be appropriate to look at motor skills over a shorter time frame and in the environmental context in which the child must function. This “variability” in motor performance may also be a function of quality of movement over time rather than quantity of motor skills as measured by the GMFM. A tool that measures “quality” of gross motor skill such as the Gross Motor Performance Measure (GMPM)24 may be warranted. There are limitations to the generalizability of our data. This was a convenience sample and is not a representative sample of children with cerebral palsy with regard to age and severity of motor impairment. The sample was homogeneous for its level of motor skills and the children were primarily spastic. Children with significant ataxia, athetosis, or dystonia were not sampled. Future research should address a larger sample of children that is balanced for motor skills, motor involvement (e.g., hemiplegia, diplegia), movement disorder (e.g., athetosis, dystonia, hypotonia), cognitive/perceptual abilities, as well as age. Our results provide independent confirmation of the short-term testretest reliability of the GMFM in a small sample of children with CP. It also offers preliminary support for the lack of extreme variability in the performance of gross motor skills in children with spastic cerebral palsy as measured by the GMFM. The GMFM has been documented to be responsive to changes in motor skills over time, is standardized for administration to children with cerebral palsy, and has been shown to have concurrent and construct ~ a l i d i t y . ~We > I ~conclude that the GMFM can be confidently used as a clinical outcome measure for children with spastic cerebral palsy. The GMFM appears to be a valid and reliable evaluative tool for clinical efficacy research in this diagnostic group.

60



REFERENCES 1. Kirschner B, Guyatt G. A methodological framework for assessing health indices. Jour Chronic Dis. 1985; 38: 27-36. 2. Bayley N. Bayley Scales of Itzfant Developmerit Manual. Second Edition. San Antonio, Tex: The Psychological Corporation, Harcourt Brace & Company; 1993. 3. Bleck EE. Locomotor prognosis in cerebral palsy. Dev Med Child Neurol. 1975; 17: 18-25. 4. Wright BD, Nicholson J. Physiotherapy for the spastic child: An evaluation. Dev Med Child Neurol. 1973; 15: 146-163. 5. Russell D, Rosenbaum P, Gowland C, Hardy S, Lane M, Plews N, McGavin H, Cadman D, Jarvis S. G r o ~ Motor s Function Measure Manual, 2nd Edition. Children’s Development Rehabilitation Programme, Hugh MacMillan Rehabilitation Centre, McMaster University, Hamilton, Ontario; 1993. 6. Juniper EF, Guyatt GH & Jaeschke R. How to Develop and Validate a New Health-Related Quality of Life Instrument. In B Spilker, Quality ofLife arid Pilzarnzacoeconornics in Clinical Trials, 2nd Ed, (Ch 6). Lippencott-ramen, Philadelphia; 1996. 7. Streiner DL, Norman GR: Health Mensurernerit Scales: A Practical Guide to their. Development arid Use. New York, NY, Oxford University Press Inc.; 1989. 8. Russell D, Rosenbaum P, Cadman D, Gowland C, Hardy S, Jarvis S: The Gross Motor Function Measure: A means to evaluate the effects of physical therapy. Dev Med ChildNeurol. 1989; 31: 341-352. 9. MacPhail HE, Kramer JF. Effect of isokinetic strength-training on functional ability and walking efficiency in adolescents with cerebral palsy. Dev Med Child Neurol. 1985; 37: 763-775. 10. Bower E, McLellan DL. Effect of increased physiotherapy on skill acquisition of children with cerebral palsy. Dev Med Child Neiirol. 1992; 34: 25-39. 11. Steinbok P, Reiner AM, Beauchamp R, Armstrong RW, Cochrane DD. A randomized clinical trial to compare selective posterior rhizotomy plus physiotherapy with physiotherapy alone in children with spastic diplegia cerebral palsy. Dev Med Child Neurol. 3997; 39: 178-184. 12. MacPhail HEA, Kramer JF. Effect of isokinetic strength training on functional ability and walking efficiency in adolescents with cerebral palsy. Dev Med Child Neiirol. 1995; 37: 763-775. 13. Tuckerman BW: Cotiductitig Educational Research. 2nd Ed. New York, NY, Harcourt Brace Jovanovich; 1978: 160-195. 14. McLaughlin JF, Bjornson KF, Astley SK, Graubert C, Hays R, Robert TS, Price R, Temkin N. Selective dorsal rhizotomy: Efficacy and safety in an investigator masked randomized clinical trial. Dev Med Child Neurol. (in press). 15. Bartko JJ, Carpenter W T On the methods and theory of reliability. J Nerv Meti Dis.1976; 163: 307-317. 16. Shrout PE, Fleiss JL. Intraclass correlation: Uses in assessing rater reliability. Psych Bulletin. 1979; 86: 420. 17. Folio MR, Fewell RR. Peabody Developrnental Motor Scales and Activity C a d 3 Marzual. Allen, TX: DLM Teaching Resources; 1983.


Cliiiical Cotzcerrzs

61

18. Bruinincks RH. Briritiitzcks-(Pseretsky TeJt of Motor Proficiency. Circle Pines, MN: American Guidance Service; 1978. 19. Campbell J. Ball J. Energetics of walking in cerebral palsy. Orthop Clitz North Atn. 1978; 9: 374-377. 20. Mossberg KA, Linton KA, Friske K. Ankle-foot orthoses: Effect on energy expenditure of gait in spastic diplegia children. Arch Plzys Med Rehabil. 1990; 71: 490-494. 21. Adler NS, Bleck EE. Correlations of balance, motion and efficiency of gait in cerebral palsy. Dev Med Child Neurol. 1986; 28(53): 3. 22. Butler P, Englebrecht M, Major RE, Tait JH, Stallard J, Patrick JH. Physiological cost index of walking for normal children and its use as an indicator of physical handicap. Dev Med Child Neurol. 1984; 26: 607-612. 23. Dahlberg GO, Norlin R. The effect of corrective surgery energy expenditure during ambulation in children with cerebral palsy. Eur J Appl Physiol. 3985; 25: 617-24. 24. Boyce WF, Gowland C, Rosenbaum PL, Lane M, Plew N, Goldsmith CH, Russell DJ, Wright V, Potter S, Harding D. The Gross Motor Performance Measure: Validity and responsiveness of a measure of quality of movement. Phys Ther. 1995; 75: 603-613.