Perceptual & Motor Skills: Physical Development & Measurement

2014, 119, 1, 305- 319. © Perceptual & Motor Skills 2014

SENSITIVITY TO FUNCTIONAL IMPROVEMENTS OF GMFM-88,

GMFM-66, AND PEDI MOBILITY SCORES IN YOUNG CHILDREN
WITH CEREBRAL PALSY1

JOOYEON KO

Department of Rehabilitation Medicine

CHA Bundang Medical Center

Summary.—This study assessed the sensitivity to functional change of the total

score on age- and severity-relevant dimensions (Goal Total score) of the Gross Motor
Function Measure (GMFM)-88 compared with GMFM-88 Total, GMFM-66, and Pedi-
atric Evaluation of Disability Inventory (PEDI) Mobility scores in children with cere-
bral palsy (CP). Correlations among the four parameters were calculated to assess
how sensitivity may differ according to the severity of CP. 64 children with CP (M
age = 43.8 mo., SD = 16.5, range = 21 to 84 mo.; 36 boys, 28 girls) were recruited. The
GMFM and PEDI assessments were performed over an interval of 6 mo. The effect
sizes for changes over time were large (0.88 to 1.26) for the selected GMFM-88 Goal
Total scores. The minimally important differences of the GMFM-88 Goal Total scores
were within the mean range of change, with CP severity categorized as GMFCS Lev-
els I/II, Level III, and Levels IV/V. The selected GMFM-88 Goal Total scores showed
from poor to good correlations with GMFM-88 Total, GMFM-66, and PEDI Mobil-
ity scores. The results indicated that age- and severity-relevant GMFM-88 Goal Total
scores were the optimal parameter to detect meaningful change in children with CP
for clinical and research use.

Cerebral palsy (CP) is a leading cause of disability in early childhood

(Reddihough & Collins, 2003). In children, CP is a complex condition re-
sulting from damage to the immature brain with primary features of move-
ment limitations and impairment in postural control (Rosenbaum, Paneth,
Leviton, Goldstein, Bax, Damiano, et al., 2007). Various types of physical
therapy remain the most important intervention (Ko & Kim, 2013) used to
improve the motor impairment. To monitor change in motor function af-
ter an intervention, an evaluative measure is required (Rosenbaum, Rus-
sell, Cadman, Gowland, Jarvis, & Hardy, 1990). An instrument's ability to
detect meaningful change over time, i.e., “sensitivity,” is vital to its use-
fulness in this context (Mokkink, Terwee, Gibbons, Stratford, Alonso, Pat-
rick, et al., 2010).
Only two measures, the Gross Motor Function Measure (GMFM) and
the Pediatric Evaluation of Disability Inventory (PEDI) have been vali-
dated as responsive tools specific to CP (Ketelaar, Vermeer, & Helders,

Address correspondence to Jooyeon Ko, Department of Rehabilitation Medicine, CHA

1

Bundang Medical Center, CHA University, 351 Yatap-dong, Bundang-gu, Seongnam-si,

Gyeonggi-do, 463-712 South Korea or e-mail (7806218@hanmail.net).

DOI 10.2466/03.25.PMS.119c14z1 ISSN 0031-5125

27-PMS_Ko_140022.indd 305 02/08/14 9:17 AM

 306 J. KO

1998). These two measures can be used in treatment goal-setting in daily

physical therapy and are useful to guide conversation with a child's fam-
ily about prognosis (Novak, Hines, Goldsmith, & Barclay, 2012). The
GMFM has two versions: the original version (GMFM-88) and the more
recent version called the GMFM-66, which comprises 66 items chosen us-
ing a Rasch analysis. Overall, the GMFM-66 is considered a better mea-
sure than the GMFM-88 because its scoring is based on an interval scale
and accounts for differences in the difficulties of the skills. However, the
GMFM-66 provides only a total score and not scores for each dimension.
In most CP studies, the GMFM score has been used in several ways in re-
search and clinical settings: the GMFM-88 Total score, one or more rele-
vant GMFM-88 Goal Total score, or the GMFM-66 score). The PEDI mea-
sures both capability and performance of self-care, mobility, and social
ability in disabled children, including those with CP, in daily situations
(Haley, Coster, Ludlow, Haltiwanger, & Andrellos, 1992). In this study, the
Mobility content domain was of interest.
Many studies, including psychometric and intervention research in
heterogeneous CP populations, have tried to assess which GMFM and/
or PEDI scores best detect clinically meaningful changes in function. For
example, Russell, Avery, Rosenbaum, Raina, Walter, and Palisano (2000)
found that in 228 children, those with mild CP had greater changes in
GMFM-66 scores than those more than 5 years old and with more severe
CP. Nordmark, Jarnlo, and Hägglund (2000) examined the sensitivity of
the GMFM and the PEDI in 18 spastic diplegic CP patients who under-
went selective dorsal rhizotomies; the PEDI Mobility score was more sen-
sitive to changes than the GMFM Total and Goal Total scores in the CP
children with Gross Motor Function Classification System (GMFCS) lev-
els II and III. Vos-Vromans, Ketelaar, and Gorter (2005), studying 33 chil-
dren with CP, found that both the GMFM-88 Total score and the PEDI
Mobility score were responsive to changes in motor ability over time,
especially in children younger than 4 years old. In 65 children with CP,
the GMFM-66 reportedly was more responsive than the GMFM-88 To-
tal score, as assessed by consistency with a therapist's judgment, partic-
ularly in children younger than 5 years (Wang & Yang, 2006). Lundkvist,
Jarnlo, Gummesson, and Nordmark (2009) investigated the longitudinal
construct validity of the GMFM-88 Total, GMFM-88 Goal Total, and GMFM-
66 scores and found that the GMFM-88 Total and Goal Total scores showed
larger sensitivities in gross motor function than did the GMFM-66 scores.
Boyd, Dobson, Parrott, Love, Oates, Larson, et al. (2001) reported that the
GMFM-88 Goal Total and GMFM-66 scores showed similar improvements
in 39 children with CP who were administered botulinum toxin type A in-
jections. Knox and Evans (2002) reported that functional improvements
were detected only with the GMFM-88 Total and GMFM-88 Goal Total

27-PMS_Ko_140022.indd 306 02/08/14 9:17 AM

 GROSS MOTOR FUNCTION, CEREBRAL PALSY 307

scores, not the GMFM-66 or PEDI Mobility scores. Understanding the in-
struments' differences in sensitivity to change and complimentary interac-
tion between them would be useful in choosing evaluative parameters for
various clinical cases.
Based on the findings of several studies of the responsiveness of
GMFM and PEDI, it was expected that sensitivity of the scores to function-
al change would be affected by several variables (Beurskens, de Vet, Köke,
van der Heijden, & Knipschild, 1995). The amount of expected change
varies depending on the age of the child, the length of follow-up period,
the sex of the child, and the assessment instrument (Greenfield & Nelson,
1992). In this study, because CP is a heterogeneous group of motor disor-
ders, first the GMFCS was used to categorize the severity of CP in each
child. Because of the small sample, GMFCS Levels were combined into
Levels I/II, III, and IV/V. To detect a treatment effect, the post-evaluation
was conducted after 6 mo. (Lundkvist, et al., 2009). Effect size, minimally
important difference, and the amount of change were calculated to exam-
ine the clinical importance of the changes (Pardasaney, Latham, Jette, Wa-
genaar, Ni, Slavin, et al., 2012). The purposes of the study were to compare
the responsiveness of the age- and severity-relevant GMFM-88 Goal Total
scores with the GMFM-88 Total score, GMFM-66 score, and PEDI Mobility
score. In addition, the correlations among the measures were calculated
for children with CP according to GMFCS Levels.
Hypothesis 1. The GMFM-88 Goal Total score will be the most sen-
sitive score among the four parameters regardless of GMFCS
Level.
Hypothesis 2. Younger children with CP will have higher respon-
siveness than older children.
Hypothesis 3. Differences in responsiveness will be found between
boys and girls.
Hypothesis 4. The GMFM and PEDI will show a complementary
relationship according to GMFCS Level.

METHOD
Participants
In total, 64 children with CP (M age = 43.8 mo., SD = 16.5, range = 21 to
84 mo.; 36 boys, 28 girls) diagnosed by a medical doctor at a clinic in South
Korea participated. Hopkins (2000) suggested that a sample size of 30–50
should be sufficient to conduct a psychometric study. Informed consent
was obtained from the mothers of the children after a full explanation of
the study and its procedures was provided. The inclusion criteria were as

27-PMS_Ko_140022.indd 307 02/08/14 9:17 AM

 308 J. KO

follows: ability to follow verbal commands, absence of nerve block (e.g.

Botox) injections within the previous 6 mo., and the absence of orthope-
dic surgery (e.g. muscle-lengthening surgery) within the previous 6 mo.
Measures
Gross Motor Function Measure (GMFM)-88.—The GMFM-88 has a well-
known scoring system that assesses changes in gross motor function. The
GMFM does not have age limits, and consists of 88 items that are cate-
gorized into five gross motor function dimensions: (A) lying and rolling,
(B) sitting, (C) crawling and kneeling, (D) standing, and (E) walking, run-
ning, and jumping (Russell, Rosenbaum, Avery, & Lane, 2002). Each item
is scored on a 4-point scale (0-1-2-3). The raw score for each dimension
(“Goal”) is converted into a percentage score. Age- and severity-relevant
Goals were chosen per age group and CP severity class (based on GMFCS
Levels as shown in Table 1). The GMFM-88 Total score is the unweighted
average of all five dimension scores. Decimals are rounded off to the near-
est whole number (Russell, et al., 2002). After choosing as Goals the dimen-
sions relevant to each age group and CP severity class, each dimension
score, the Goal Total score (sum of Goal scores), and Total score (sum of all
dimension scores) were calculated. All scores range from 0% to 100%.
Goal-selection procedures for calculating the GMFM-88 Goal Total
score were as follows: the treating pediatric physical therapist and child/
family selected age-and severity-relevant dimensions as Goals according
to GMFCS Level considering the child's age and in terms of activities and
participation according to the International Classification of Functioning,
Disability and Health for Children and Youth (ICF-CY). Ages were divid-
ed into four subgroups (< 2 years, 2–4 years, 4–6 years, and 6–12 years),
based on the GMFCS age classifications (Table 1). Then, individualized
treatment programs were planned based on the Goal(s) for the purpose
of the movement quality control and task-oriented approaches for each
child. Each child's Goal(s) remained unchanged over the 6-mo. period. In
the present study, the Korean version of the GMFM-88 was used (GMFM-
88–K). Regarding inter-rater reliability of the GMFM-88–K, ICC was .975–
.997 and test-retest reliability was .998–1.000 (Ko & Kim, 2013; Table 1).
Gross Motor Function Measure (GMFM)-66.—GMFM-66 is the updated
version of GMFM-88, in which the original 88 items were reduced to 66
items. The total score was calculated using a computer program called the
Gross Motor Ability Estimator. A higher score represents better gross mo-
tor function.
Pediatric Evaluation of Disability Inventory.—For the objectives of the
present study, the PEDI Mobility content domain was used. PEDI Mobility
was administered through an interview with the mothers of the children.
The measure was designed for use with children under the age of 8 years;

27-PMS_Ko_140022.indd 308 02/08/14 9:17 AM

 GROSS MOTOR FUNCTION, CEREBRAL PALSY 309

TABLE 1
SELECTION OF GOAL DIMENSIONS (NUMBER OF CHILDREN) FROM THE
GROSS MOTOR FUNCTION MEASURE-88 BY AGE GROUP AND SEVERITY
(GMFCS LEVELS)

GMFCS Age Group

n
Level < 2 years 2–4 years 4–6 years > 6 years
I/II 24 B, C (2) B, C (1) C + D (2) D + E (1)
C + D (2) C, D (3) D (1)
D (1) D (6) D + E (1)
D + E (4)
III 17 A + B (1) A + B (6) B + C (2)
B + C (1) B + C (5)
C + D (2)
IV/V 23 A (3) A (1) A (2)
A + B (2) A + B (5) A + B (2)
B (4) B + C (3)
C (1)
Note.—GMFCS = Gross Motor Function Classification System; GMFM
= Gross Motor Function Measure, dimensions: A = lying and rolling,
B = sitting, C = crawling & kneeling, D = standing, E = walking, run-
ning & jumping. In the analyses, GMFCS Levels were grouped due to
small numbers of participants: Levels I/II, Level III, and Levels IV/V.

however, it can be also used in older children with functional status below
that of normally developed 8-year-olds. The score ranges from 0 to 100, with
higher scores indicating better function (Haley, et al., 1992). PEDI inter-rater
and intra-rater reliabilities were ICC = 0.95–0.99 (Berg, Jahnsen, Frøslie, &
Hussain, 2004). The assessments before and after the intervention were ad-
ministered by a senior pediatric physical therapist who did not take part in
goal-setting or treating the participants during the 6-mo. research period.
Procedure
Motor development of children with CP is affected by the functional
severity of CP. In this study, the Korean version of the Gross Motor Func-
tion Classification System (K–GMFCS) was used to classify CP severity
as GMFCS Levels I/II, Level III, and Levels IV/V. Two pediatric physical
therapists assisting this study rated the GMFCS Level of each child. Other
information such as age, gestational age at birth, and birth weight were
obtained through interviews with the mothers of the children.
All children were scheduled to participate in two separate test sessions
using the K–GMFM and the PEDI Mobility over a period of 6 mo. to in-
vestigate responsiveness. During this time, the children were given physi-
cal therapy three times per week, 30 min. per session. Two assistants with
more than five years of experience in treating and evaluating children with
CP measured gross motor function at baseline and upon completion of the
study at 6 mo., according to the standardized manual.

27-PMS_Ko_140022.indd 309 02/08/14 9:17 AM

 310 J. KO

Analysis
Analyses were performed using the SPSS (Version 12.0.1; SPSS Inc.,
Chicago, IL, USA). Scores obtained on GMFM-88, GMFM-66, and PEDI
Mobility were described using means and standard deviations. To deter-
mine test sensitivity, the paired t test, the effect size, and the minimally im-
portant difference were calculated. Effect size is a measure of functional
change from pre- to post-intervention; the minimally important difference
is the lower boundary of change that has been defined as clinically impor-
tant (Berg & Norman, 1996). Any amount of change greater than the mini-
mally important difference threshold is considered meaningful.
The paired t test was used to measure the statistical significance of
changes during the 6 mo. period. The effect size was calculated by divid-
ing the mean change by the SD at baseline. According to Cohen's criteria,
an effect size of 0.2 reflects a small change, 0.5 a moderate change, and 0.8
a large change (Cohen, 1988). The minimally important difference was cal-
culated using three commonly used effect size estimates: 0.3 × Sb, 0.5 × Sb,
and 0.8 × Sb, where Sb is the standard deviation of the baseline scores
(Oeffinger, Bagley, Rogers, Gorton, Kryscio, Abel, et al., 2008; Adair, Said,
Rodda, & Morris, 2012; Pardasaney, et al., 2012). According to Haley and
Fragala-Pinkham (2006), the minimally important difference is considered
more informative when expressed as a range of values rather than a single
value. Pearson correlation coefficients were calculated among the four test
scores. A correlation coefficient of r ≥ .8 indicates a high correlation, r = .6–
.8 indicates a strong correlation, r = .4–.6 indicates a moderate correlation,
and r < 0.4 indicates a weak correlation (Meyer, 1979).
RESULTS
Characteristics of the participants and descriptive statistics for the
GMFM-88 Total, GMFM-88 Goal Total, GMFM-66, and PEDI Mobility
scores are presented in Tables 2 and 3. The results of the paired t test for
the data gathered at baseline and 6 mo. after baseline indicated significant
change (Table 3), indicating functional changes were detected using all
test scores at each severity level.
The sensitivity of the change scores at 6 mo. after baseline assessment
is shown in Table 4. In GMFCS Levels I/II, the effect sizes of the function-
al change scores on the GMFM-88 Goal Total and GMFM-66 were 1.26
and 0.89, respectively, both large effects. All minimally important differ-
ences exceeded the corresponding 0.3 SD of baseline, 0.5 SD of baseline,
and 0.8 SD of baseline values. In GMFCS Level III, the effect size exceed-
ed 1.0 for all four change scores, and all minimally important differences
remained in the mean change range. For GMFCS Levels IV/V, GMFM-88
Goal Total and PEDI Mobility effect sizes were 0.88 and 0.73, respectively,

27-PMS_Ko_140022.indd 310 02/08/14 9:17 AM

 GROSS MOTOR FUNCTION, CEREBRAL PALSY 311

TABLE 2
SUMMARY OF SAMPLE CHARACTERISTICS
GMFCS I/II GMFCS III GMFCS IV/V Total (N = 64)
Variable (n = 24) (n = 17) (n = 23)
M SD M SD M SD M SD
Sex
Boys 14 58.3 9 52.9 13 56.5 36 56.3
Girls 10 41.7 8 47.1 10 43.5 28 43.7
Age, mo. 40.6 16.9 36.7 10.5 52.4 16.7 43.8 16.5
Gestational age, wk. 36.5 4.4 33.0 4.9 32.5 4.5 34.1 4.9
Birth weight, kg 2.6 0.9 1.9 0.8 1.9 0.7 2.2 0.9
Note.—Values are n (%) or M (SD). Gestational age is weeks of gestation at birth. Classes of
CP severity are derived from the five Levels of the GMFCS: I&II, III, and IV&V. Goal Total
scores for each age group and severity class are the average of the total scores for the rele-
vant Goal dimensions.

but only the minimally important difference of GMFM-88 Goal Total re-
mained in the mean change range (Table 4).
The difference between younger and older children is shown in Ta-
bles 4 and 5. When considering mean age of the participant (Table 2), the
effect size was smaller in GMFCS Levels IV/V than in the two other sub-
groups. That is, scores of younger children with CP indicated larger func-
tional change with therapy than scores of older children.
In comparing differences between boys and girls, boys in the GMFCS
Levels I/II and III showed larger effect sizes in the GMFM-88 Goal Total
score and GMFM-66 score than girls at these same levels (Table 5). Also, the
minimally important difference remained in the mean range in the GMFM-
88 Goal Total score. In GMFCS Levels IV/V, the effect size of the GMFM-88
Goal Total score was 0.78 for boys and 1.05 for girls; the minimally impor-
tant difference for girls remained in the mean change range (Table 5).
Complementary correlations among the four parameters are shown
in Table 6. In the three subgroups of GMFCS, the correlations of GMFM-
88 Total and GMFM-66 were r = .98, .97, and .96, respectively, indicating in
psychometric terms that the scores are equivalent. Correlations in the dif-
ferent severity levels for children with CP varied widely: for GMFM-88
Goal Total vs PEDI Mobility, they ranged from zero to .58; for GMFM-88
Goal Total vs GMFM-88 Total, they ranged from .39 to .64; and for GMFM-
88 Goal Total vs GMFM-66, they ranged from .46 to .70 (Table 6).
DISCUSSION
This study examined the responsiveness of and correlation among
the GMFM-88 Total, GMFM-88 Goal Total, GMFM-66, and PEDI Mobility
scores in children with CP. The results indicated that the GMFM-88 Goal

27-PMS_Ko_140022.indd 311 02/08/14 9:17 AM

27-PMS_Ko_140022.indd 312

312
TABLE 3
MEANS AND STANDARD DEVIATIONS FOR GMFM-88, GMFM-66, AND PEDI MOBILITY BY GMFCS LEVELS I/II, III, AND IV/V
GMFCS I/II (n = 24) GMFCS III (n = 17) GMFCS IV/V (n = 23)
Score Baseline After 6 mo. Baseline After 6 mo. Baseline After 6 mo.
M SD M SD M SD M SD M SD M SD
(A) Lying & rolling a
97.4 3.3 99.2* 1.7 89.7 12.0 93.7‡ 9.9 71.9 26.1 80.2‡ 23.1
(B) Sittinga 92.9 8.2 97.7* 4.5 71.7 19.9 86.2† 13.9 36.6 22.6 50.2‡ 25.2
(C) Crawling & kneelinga 87.7 13.2 96.5* 4.7 60.4 17.0 74.8† 18.9 16.3 24.0 25.2‡ 27.8

J. KO
(D) Standinga 69.7 18.3 81.3‡ 15.6 19.8 14.3 36.8‡ 18.3 3.6 5.3 8.6* 12.0
(E) Walking, running, &
46.5 23.9 59.7‡ 28.4 8.7 6.9 14.1† 9.5 1.3 3.8 1.9* 4.5
jumpinga
GMFM-88 Goal Total 72.3 9.1 83.8‡ 8.3 67.1 14.5 83.1‡ 12.3 49.4 16.1 63.6‡ 16.5
GMFM-88 Total 78.8 11.9 87.0‡ 9.8 50.1 10.1 61.1‡ 9.0 25.9 14.2 33.2‡ 16.3
GMFM-66 61.7 7.9 69.5‡ 10.8 45.8 4.1 50.3‡ 3.9 31.9 9.6 36.1‡ 9.6
PEDI Mobility 56.9 16.3 70.6‡ 12.9 34.6 8.7 44.5‡ 10.5 18.5 10.3 26.1‡ 11.2
Note.—GMFCS = Gross Motor Function Classification System; GMFM = Gross Motor Function Measure; PEDI = Pediatric Evaluation of Disabil-
ity Inventory; Sb = baseline standard deviation. a GMFM Dimension. *p < .05. †p < .01. ‡p < .001.
02/08/14 9:17 AM
 GROSS MOTOR FUNCTION, CEREBRAL PALSY 313

TABLE 4
EFFECT SIZES (ES) AND MINIMALLY IMPORTANT DIFFERENCES FOR GMFM-88, GMFM-66, AND
PEDI MOBILITY BY GMFCS LEVELS I/II, III, AND IV/V
Change Score Minimally Important Difference
Score ES
M SD 0.3 × Sb 0.5 × Sb 0.8 × Sb
GMFCS I/II (n = 24)
GMFM-88 Goal Total 1.26 11.5 5.7 2.73 4.55 7.28
GMFM-88 Total 0.60 7.2 4.2 3.57 5.95 9.52
GMFM-66 0.89 7.1 3.9 2.37 3.95 6.32
PEDI Mobility 0.63 10.4 8.7 4.89 8.15 13.04
GMFCS III (n = 17)
GMFM-88 Goal Total 1.10 16.0 8.9 4.35 7.25 11.6
GMFM-88 Total 1.11 11.1 5.5 3.03 5.05 8.08
GMFM-66 1.09 4.5 2.3 1.23 2.05 3.28
PEDI Mobility 1.13 9.9 4.9 2.61 4.35 6.96
GMFCS IV/V (n = 23)
GMFM-88 Goal Total 0.88 14.2 8.9 4.83 8.05 12.88
GMFM-88 Total 0.51 7.3 3.8 4.26 7.1 11.36
GMFM-66 0.43 4.2 2.9 2.88 4.8 7.68
PEDI Mobility 0.73 7.6 6.4 3.09 5.15 8.24
Note.—GMFCS = Gross Motor Function Classification System; GMFM: Gross Motor Func-
tion Measure; PEDI = Pediatric Evaluation of Disability Inventory; Sb = baseline standard de-
viation.

Total score was the most responsive parameter, regardless of functional

severity. Younger children and boys were more responsive than older chil-
dren and girls. The four scores all showed moderate-to-strong relation-
ships with functional motor changes in children with CP over 6 mo.
Test sensitivity to functional change over time is particularly impor-
tant in detecting the progress of the children with CP after interventions
for both clinical and research purposes. The results indicated that only the
GMFM-88 Goal Total score showed a large effect size; i.e., more than 0.88
for all three GMFCS levels (I/II, III, and IV/V). In the severely impaired
CP groups (GMFCS Levels IV/V), the GMFM-88 Goal Total score alone
showed large changes (effect size = 0.88), while the GMFM-88 and PEDI
Mobility showed medium effect sizes (0.51 and 0.73), and the GMFM-66
had a small effect size of 0.43.
The GMFM-66 has been used in many studies of children with CP
whose functional CP severities were classified as GMFCS Levels I to III
(e.g., Yi, Hwang, Kim, & Kwon, 2012). In the study by Oeffinger, et al.
(2008) on 381 ambulatory children with CP to examine responsiveness
and clinically important differences in two assessments with an interval

27-PMS_Ko_140022.indd 313 02/08/14 9:17 AM

27-PMS_Ko_140022.indd 314

314
TABLE 5
EFFECT SIZE (ES) AND MINIMALLY IMPORTANT DIFFERENCES OF THE GMFM-88, GMFM-66, AND PEDI MOBILITY CHANGE SCORES BY SEX
Boys Girls
Minimally Important Minimally Important
Score Change Score Change Score
ES Difference ES Difference
M SD 0.3×Sb 0.5×Sb 0.8×Sb M SD 0.3×Sb 0.5×Sb 0.8×Sb
GMFCS I/II (n = 24)
GMFM-88 Goal Total 1.23 11.7 6.5 2.88 4.80 7.68 1.23 11.2 3.9 2.71 4.52 7.23
GMFM-88 Total 0.78 6.8 3.9 2.61 4.35 6.96 0.55 7.9 5.0 4.32 7.19 11.51
GMFM-66 1.09 7.1 3.8 1.95 3.25 5.20 0.8 7.2 4.4 2.69 4.48 7.17
PEDI Mobility 0.93 10.3 9.0 3.33 5.55 8.88 0.67 10.5 8.5 4.68 7.80 12.48
GMFCS III (n = 17)

J. KO
GMFM-88 Goal Total 1.05 16.57 7.62 4.74 7.90 12.65 1.10 15.33 10.70 4.18 6.97 11.15
GMFM-88 Total 1.27 13.0 6.74 3.07 5.12 8.20 0.85 8.94 2.82 3.16 5.26 8.42
GMFM-66 1.15 5.34 2.75 1.4 2.33 3.73 0.99 3.59 1.17 1.08 1.81 2.89
PEDI Mobility 1.21 10.3 6.34 2.55 4.25 6.8 1.02 9.52 3.03 2.80 4.67 7.47
GMFCS IV/V (n = 23)
GMFM-88 Goal Total 0.78 14.45 9.46 5.53 9.22 14.75 1.05 13.78 8.70 3.95 6.58 10.53
GMFM-88 Total 0.52 7.10 3.58 4.08 6.81 10.89 0.48 7.55 4.25 4.69 7.82 12.51
GMFM-66 0.44 4.0 2.93 2.75 4.59 7.34 0.42 4.44 3.03 3.17 5.28 8.45
PEDI Mobility 0.79 8.10 7.32 3.07 5.12 8.19 0.63 6.91 5.18 3.29 5.48 8.77
Note.—GMFCS = Gross Motor Function Classification Measure; GMFM: Gross Motor Function Measure; PEDI = Pediatric Evaluation of Disability
Inventory; Sb = baseline standard deviation.
02/08/14 9:17 AM
 GROSS MOTOR FUNCTION, CEREBRAL PALSY 315

TABLE 6
CORRELATION MATRIX FOR GMFM-8, GMFM-6, AND PEDI
MOBILITY ACCORDING TO GMFCS LEVELS I/II, III, AND IV/V
Pearson Correlation
GMFCS Level and Score PEDI GMFM-88
Mobility Goal Total Total
GMFCS Levels I/II
GMFM-88 Goal Total .30
GMFM-88 Total .76† .60†
GMFM-66 .73† .58† .98†
GMFCS Level III
GMFM-88 Goal Total −.04
GMFM-88 Total .56* .39
GMFM-66 .54* .46 .97*
GMFCS Levels IV/V
GMFM-88 Goal Total .58†
GMFM-88 Total .68† .64†
GMFM-66 .70† .70† .96†
Note.—GMFCS = Gross Motor Function Classification System;
GMFM = Gross Motor Function Measure; PEDI = Pediatric Eval-
uation of Disability Inventory. *p < .05, †p < .01.

of 1 year using GMFM-66 and dimensions D and E of the GMFM-88, me-

dium or large effect sizes were seen for GMFCS Level III. The study of
Wang and Yang (2006) on 65 CP children, whose CP severity ranged from
GMFCS Levels I to V, evaluated responsiveness of the GMFM-88 and the
GMFM-66 with a mean interval of 3.5 mo. with no intervention, and re-
ported that the GMFM-66 had superior sensitivity. It is difficult to com-
pare their results with the present study, because the functional CP sever-
ity of the participants, dependent variables, interval of assessment, and
statistical methods all differed.
More recently, the use of the GMFCS was emphasized in an effort
to minimize the effects of the heterogeneous nature of the CP severity in
most CP studies. Also, selection of goal area(s) and item(s) of the GMFM-
88 for individual children according to the GMFCS Level is necessary to
give insight in setting realistic short-term goals for physical therapists, chil-
dren, and families and to encourage goal-directed approaches, regarded as
a “green light” intervention in the literature (Novak, Mcintyre, Morgan,
Campbell, Dark, Morton, et al., 2013). In a study by Lundkvist, et al. (2009),
effect sizes generally were lower for the GMFM-66 scores than for the
GMFM-88 Total and Goal Total scores in both mild CP (GMFCS Levels I to
III) and severe cases (GMFCS Levels IV and V). In a study comparing sen-

27-PMS_Ko_140022.indd 315 02/08/14 9:17 AM

 316 J. KO

sitivity between the GMFM-88 Total score and Goal Total score by Ko and
Kim (2013) in 84 children with CP, the effect size was large for GMFM-88
Goal Total scores and medium for the GMFM-88 Total score for subgroups
with GMFCS Levels I and II and GMFCS Levels III to V. These results are
consistent with the present study and may suggest that the GMFM-88 Goal
Total score is more sensitive to functional change than the three other vari-
ables regardless of the functional severity level.
In the current study, the effect sizes in the three GMFCS levels for
the GMFM-88 Goal Total score were large, and the range of minimally
important difference (0.3 SD, 0.5 SD, and 0.8 SD at baseline) was within
the mean range of change for that score. Several terms have been used
to describe the clinically meaningful change, such as ‘minimal detectable
change,’ ‘minimally clinically important difference,’ and ‘minimal impor-
tant difference,’ or estimated using standard error of measurement (SEM)
and intraclass correlation coefficient (ICC) in some cases (Baker, McGinley,
Schwartz, Thomason, Rodda, & Graham, 2012). According to Haley and
Fragala-Pinkham (2006), the minimally important difference is better pre-
sented as a range of values, using the ranges of effect size defined by Co-
hen (1988), than as a single value, because any single value has some un-
certainty. For example, to be considered clinically significant, the change
score in GMFM-88 Goal Total had to exceed 2.73 for GMFCS Levels I and
II in this study (Table 4). The actual change score for GMFM-88 Goal Total
was 11.5 after the intervention; this can be interpreted as 6 mo. of physi-
cal therapy resulting in a clinically meaningful change. Using the above
standard values for interpretation, in GMFCS Levels IV/V, the GMFM-88
Goal Total score ranged from 4.83 to 12.88 and the actual average amount
of change after 6 mo. (14.2) exceeded the minimally important difference,
so it too can be interpreted as a clinically important change (Table 4).
In this study, older children with a mean age of more than 52 mo. had
more severe CP than the younger ones, and the effect sizes of the four
tests' change scores were lower than for those for children with milder CP.
In Vos-Vromans, et al.'s (2005) study of 55 children with CP to compare
sensitivity of the GMFM and the PEDI, the two instruments were more re-
sponsive to changes in motor ability over time in children younger than 4
years old than children older than 4 (Tables 2, 4, and 5).
In comparing the responsiveness of the four instruments' change
scores between boys and girls, those with more severe CP (GMFCS Lev-
els IV/V) showed a moderate effect size (< 0.5), while those with milder
involvement (GMFCS Levels I/II) showed a large effect size (> 0.8). These
results are consistent with a number of other studies and suggest that goal
area(s) and item(s) are more responsive in the case of children with func-
tionally severe CP, because it is unlikely that a summary score is sufficient-
ly sensitive to detect subtle changes as CP severity increases.

27-PMS_Ko_140022.indd 316 02/08/14 9:17 AM

 GROSS MOTOR FUNCTION, CEREBRAL PALSY 317

Regarding the correlations among different instruments' change scores

by GMFCS Levels, the GMFM-88 Goal Total score showed correlations with
GMFM-88 Total, GMFM-66, and PEDI Mobility scores ranging from zero to
.70. This suggests that the GMFM-88 Goal Total score and the three other
scores should be used in a complementary way in assessing motor function
in children with CP. For example, the correlation coefficient between the
GMFM-88 Goal Total score and the GMFM-66 score was .70 in this study,
indicating that the two parameters showed a good correlation in evaluating
motor ability changes in children with CP. However, they do not assess ex-
actly the same aspects of motor changes (they share about 50% of their vari-
ance), so it is better to use the two parameters together to obtain a complete
picture of motor ability in CP (Ketelaar, et al., 1998).
Researchers have often reported only the summary effect size and min-
imally important difference statistic for the GMFM-88. However, it is im-
portant to provide a more detailed description of the results, not just sum-
mary scores, for monitoring gross motor changes. The distribution of effect
sizes and minimally important differences for functional change scores,
and graphs indicating patterns of the changes, can demonstrate meaning-
ful changes for an individual child or group (Beaton, 2000). Recently, the
Goal Attainment Scale (GAS) has often been used. Steenbeek, Gorter, Kete-
laar, Galama, & Lindeman (2011) investigated responsiveness of the GAS
in comparison with the GMFM and the PEDI and stated that the GAS, the
PEDI, and the GMFM-66 were complementary in their ability to assess indi-
vidual changes in children with CP over time. An advantage of the GAS is
its availability; in fact, it can be used even by assessors who are unfamiliar
with the participant. Further study is necessary to examine the sensitivity of
the GMFM, the PEDI, and the GAS to interventions with more detailed con-
trol of the heterogeneous nature of CP in the sample.
The present study had some limitations. First, the study sample in the
lower age range would be expected to show larger changes in measures
at 6 mo. after baseline evaluation vs an older CP population, so results are
not generalizable to that group. Second, a convenience sampling method
at one CP clinic was used to select the participants for this study.
In conclusion, it is important to use an assessment instrument with
sensitivity to outcome measures in children with CP, taking into consid-
eration the CP severity. The GMFM-88 Goal Total score was sensitive to
change in the current study, regardless of the children's GMFCS Levels.
It is possible to set realistic short-term and long-term goals based on goal
area(s) and item(s) considering both movement quality and goal-directed
approaches for each child. It also provides a useful way to guide a conver-
sation with the family of the child about the effectiveness of an interven-
tion using objective numerical data.

27-PMS_Ko_140022.indd 317 02/08/14 9:17 AM

 318 J. KO

REFERENCES
ADAIR, B., SAID, C. M., RODDA, J., & MORRIS, M. E. (2012) Psychometric properties of func-
tional mobility tools in hereditary spastic paraplegia and other childhood neuro-
logical conditions. Developmental Medicine & Child Neurology, 54, 596-605.
BAKER, R., MCGINLEY, J. L., SCHWARTZ, M., THOMASON, P., RODDA, J., & GRAHAM, H. K. (2012)
The minimal clinically important difference for the Gait Profile Score. Gait & Pos-
ture, 35, 612-615.
BEATON, D. E. (2000) Understanding the relevance of measured change through studies
of responsiveness. Spine, 25, 3192-3199.
BERG, K. O., & NORMAN, K. E. (1996) Functional assessment of balance and gait. Clinics
in Geriatric Medicine, 12, 705-723.
BERG, M., JAHNSEN, R., FRØSLIE, K. F., & HUSSAIN, A. (2004) Reliability of the Pediatric
Evaluation of Disability Inventory (PEDI). Physical & Occupational Therapy in Pedi-
atrics, 24, 61-77.
BEURSKENS, A. J., DE VET, H. C., KÖKE, A. J., VAN DER HEIJDEN, G. J., & KNIPSCHILD, P. G. (1995)
Measuring the functional status of patients with low back pain: assessment of the
quality of four disease-specific questionnaires. Spine, 20, 1017-1028.
BOYD, R. N., DOBSON, F., PARROTT, J., LOVE, S., OATES, J., LARSON, A., BURCHALL, G., CHON-
DROS, P., CARLIN, J., NATTRASS, G., & GRAHAM, H. K. (2001) The effect of botulinum
toxin type A and a variable hip abduction orthosis on gross motor function: a
randomized controlled trial. European Journal of Neurology, 8(Suppl. 5), 109-119.
COHEN, J. (1988) Statistical power analysis for the behavioral sciences. (2nd ed.) Hillsdale,
NJ: Erlbaum.
GREENFIELD, S., & NELSON, E. C. (1992) Recent developments and future issues in the use
of health status assessment measures in clinical settings. Medical Care, 30(Suppl.),
MS23-41.
HALEY, S. M., COSTER, W. J., LUDLOW, L. H., HALTIWANGER, J. T., & ANDRELLOS, P. J. (1992)
Pediatric Evaluation of Disability Inventory (PEDI). Boston, MA: New England Med-
ical Center Hospitals.
HALEY, S. M., & FRAGALA-PINKHAM, M. A. (2006) Interpreting change scores of tests and
measures used in physical therapy. Physical Therapy, 86, 735-743.
HOPKINS, W. G. (2000) Measures of reliability in sports medicine and science. Sports
Medicine, 30, 1-15.
KETELAAR, M., VERMEER, A., & HELDERS, P. J. (1998) Functional motor abilities of children
with cerebral palsy: a systematic literature review of assessment measures. Clini-
cal Rehabilitation, 12, 369-380.
KNOX, V., & EVANS, A. L. (2002) Evaluation of the functional effects of a course of bobath
therapy in children with cerebral palsy: a preliminary study. Developmental Medi-
cine & Child Neurology, 44, 447-460.
KO, J., & KIM, M. (2013) Reliability and responsiveness of the Gross Motor Function
Measure-88 in children with cerebral palsy. Physical Therapy, 93, 393-400.
LUNDKVIST, J. A., JARNLO, G. B., GUMMESSON, C., & NORDMARK, E. (2009) Longitudinal con-
struct validity of the GMFM-88 total score and goal total score and the GMFM-66
score in a 5-year follow-up study. Physical Therapy, 89, 342-350.
MEYER, C. R. (1979) Measurement in physical education. New York: Ronald Press.
MOKKINK, L. B., TERWEE, C. B., GIBBONS, E., STRATFORD, P. W., ALONSO, J., PATRICK, D. L.,
KNOL, D. L., BOUTER, L. M., & HENRICA, C. W. DE VET. (2010) Inter-rater agreement
and reliability of the COSMIN (COnsensus-based Standards for the selection of

27-PMS_Ko_140022.indd 318 02/08/14 9:17 AM

 GROSS MOTOR FUNCTION, CEREBRAL PALSY 319

health status Measurement INstruments) checklist. BMC Medical Research Meth-

odology, 10, 82.
NORDMARK, E., JARNLO, G. B., & HÄGGLUND, G. (2000) Comparison of the Gross Motor
Function Measure and Pediatric Evaluation of Disability Inventory in assessing
motor function in children undergoing selective dorsal rhizotomy. Developmental
Medicine & Child Neurology, 42, 245-252.
NOVAK, I., HINES, M., GOLDSMITH, S., & BARCLAY, R. (2012) Clinical prognosis messages
from a systematic review on cerebral palsy. Pediatrics, 130, e1285-1312.
NOVAK, I., MCINTYRE, S., MORGAN, C., CAMPBELL, L., DARK, L., MORTON, N., STUMBLES, E., &
GOLDSMITH, S. (2013) A systematic review of interventions for children with cere-
bral palsy: state of the evidence. Developmental Medicine & Child Neurology, 55,
885-910.
OEFFINGER, D., BAGLEY, A., ROGERS, S., GORTON, G., KRYSCIO, R., ABEL, M., DAMIANO, D.,
BARNES, D., & TYLKOWSKI, C. (2008) Outcome tools used for ambulatory children
with cerebral palsy: responsiveness and minimum clinically important differ-
ences. Developmental Medicine & Child Neurology, 50, 918-925.
PARDASANEY, P. K., LATHAM, N. K., JETTE, A. M., WAGENAAR, R. C., NI, P., SLAVIN, M. D., &
BEAN, J. F. (2012) Sensitivity to change and responsiveness of four balance mea-
sures for community-dwelling older adults. Physical Therapy, 92, 388-397.
REDDIHOUGH, D. S., & COLLINS, K. J. (2003) The epidemiology and causes of cerebral
palsy. Australian Journal of Physiotherapy, 49(1), 7-12.
ROSENBAUM, P., PANETH, N., LEVITON, A., GOLDSTEIN, M., BAX, M., DAMIANO, D., DAN, B.,
& JACOBSSON, B. (2007) A report: the definition and classification of cerebral palsy.
Developmental Medicine & Child Neurology, 49(Suppl. 109), 8-14.
ROSENBAUM, P. L., RUSSELL, D. J., CADMAN, D. T., GOWLAND, C., JARVIS, S., & HARDY, S. (1990)
Issues in measuring change in motor function in children with cerebral palsy: a
special communication. Physical Therapy, 70, 125-131.
RUSSELL, D. J., AVERY, L. M., ROSENBAUM, P. L., RAINA, P. S., WALTER, S. D., & PALISANO, R. J.
(2000) Improved scaling of the Gross Motor Function Measure for children with
cerebral palsy: evidence of reliability and validity. Physical Therapy, 80, 873-885.
RUSSELL, D. J., ROSENBAUM, P. L., AVERY, L. M., & LANE, M. (2002) Gross Motor Function Measure
(GMFM-66 & GMFM-88) user's manual. (2nd ed.) London, UK: Mac Keith Press.
STEENBEEK, D., GORTER, J. W., KETELAAR, M., GALAMA, K., & LINDEMAN, E. (2011) Respon-
siveness of goal attainment scaling in comparison to two standardized measures
in outcome evaluation of children with cerebral palsy. Clinical Rehabilitation, 25,
1128-1139.
VOS-VROMANS, D. C., KETELAAR, M., & GORTER, J. W. (2005) Responsiveness of evalua-
tive measures for children with cerebral palsy: the Gross Motor Function Measure
and the Pediatric Evaluation of Disability Inventory. Disability & Rehabilitation, 27,
1245-1252.
WANG, H. Y., & YANG, Y. H. (2006) Evaluating the responsiveness of two versions of the
Gross Motor Function Measure for children with cerebral palsy. Archives of Physi-
cal Medicine and Rehabilitation, 87, 51-56.
YI, S. H., HWANG, J. H., KIM, S. J., & KWON, J. Y. (2012) Validity of pediatric balance scales
in children with spastic cerebral palsy. Neuropediatrics, 43, 307-313.

Accepted June 3, 2014.

27-PMS_Ko_140022.indd 319 02/08/14 9:17 AM