Comparison of two outcome measures for infants with cerebral palsy and infants with motor delays.

Key Words: Cerebral palsy, Developmental tests, Gross motor development, Measurement, Responsiveness.

Children with cerebral palsy (CP) and infants who are at risk for gross motor developmental disabilities are among the children frequently treated by physical therapists. Efforts to document the effectiveness of treatments have yielded inconclusive findings. The magnitude of change reported for infants and young children who have received physical therapy has been either mild to moderate or insignificant.[1-10] One factor that confounds the interpretation of existing research is the lack of evidence of the validity of the outcome measures.[11-13] Norm-referenced assessments have frequently been used to measure the effectiveness of interventions for children with motor disabilities.[3,5,6-10] Norm-referenced assessments are one type of discriminative measure used to distinguish between children with or without delays in development. Although the results of norm-referenced assessments can be compared among groups of children, standardized scores such as developmental quotients are based on the average performance of children without delays in development. The validity of using norm-referenced assessments to measure change over time in children with motor disabilities has been questioned.[14-17]

An evaluative measure is necessary when the purpose of testing is to measure change over time.[18] Responsiveness, a type of validity that deals with the ability to measure clinically important changes over time, becomes important when an evaluative measure is used.[19] The responsiveness of an evaluative measure should be determined specific to the clinical population of interest, such as children with CP.[13,20] The items and rating scales of an evaluative measure, therefore, must be sensitive to changes that the population of interest is capable of making within a specified period of time.

The Peabody Developmental Gross Motor Scale (PDMSGM)[21] is one of the most frequently used tests of motor development. The PDMS-GM is standardized and normed for children aged from birth through 83 months, and it has been validated for use as a discriminative measure.[21-23] Although Folio and Fewell[21] stated that a purpose of the PDMS-GM is to measure change across time or after intervention for children with motor impairments or delays, the responsiveness of the PDMS-GM had not been investigated.

We previously examined the validity of the PDMS-GM as an evaluative measure of infants receiving physical therapy.[20] The PDMS-GM was administered to 124 infants 3 times over a 6-month period. The infants were between 2 and 33 months of age at the start of the study, and they were grouped by diagnosis (ie, CP, Down syndrome, hydrocephalus, preterm with developmental delay, full term with developmental delay, and "other"). Mean scaled scores and age-equivalent scores increased for all groups over the 6-month period, supporting the use of the PDMS-GM as an evaluative measure. The results for the infants with CP, however, suggested that the PDMS-GM has limitations when used as an evaluative measure for this group. The 36 infants with CP demonstrated the smallest mean change in scaled scores (15.0) and age-equivalent scores (2.2 months) for the 6-month period among the 6 groups. Only 50% of the infants with CP had changes in raw scores that exceeded what could potentially be attributed to random variation or measurement error associated with the PDMS-GM.

Our findings on the responsiveness of the PDMS-GM raise the question of whether items are sensitive to changes among children with CP or whether the results reflect the potential for change in children with CP over a 6-month period. A method that has been proposed to address this question involves comparison of the instrument being evaluated with another measure of known responsiveness.[18,19,24-26] An advantage of this approach is that, not only can the responsiveness of the instrument be examined, but 2 measures can also be compared. Stratford et al[26] have proposed using a 3-factor repeated-measures analysis-of-variance (ANOVA) approach that directly compares the ability of 2 measures to assess change in a patient sample grouped by potential for important change.

The Gross Motor Function Measure (GMFM) is the first standardized instrument that was constructed and validated to measure change in gross motor function in children with CP.[24] The GMFM, therefore, can serve as a criterion measure to examine the validity of the PDMS-GM for measuring change over time. The GMFM consists of 88 items grouped into 5 dimensions: (1) lying and rolling, (2) sitting, (3) crawling and kneeling, (4) standing, and (5) walking, running, and jumping. Items were selected to represent motor functions typically performed by children without motor impairments by age 5 years and are scored on a 4-point Likert scale. A score of 0 indicates that the child does not initiate the movement, a score of 1 indicates that the child initiates but completes less than 10% of the movement, a score of 2 indicates that the child partially completes the movement, and a score of 3 indicates that the child successfully completes the movement. Criteria are provided for scoring each item in the manual.[24] The score for each dimension is expressed as a percentage of the maximum score for that dimension. A total score is obtained by adding the percentage scores for each dimension and dividing the sum by the total number of dimensions. Each dimension, therefore, contributes equally to the total score.

The responsiveness of the GMFM has been demonstrated by several methods of analysis.[24] The GMFM was validated on 136 children with CP, 25 children with acute head injury, and 34 children without motor delays who ranged in age from I month to 4.3 years. The measure was administered to the subjects twice over a period of 6 months. In the absence of an accepted external standard to measure motor change for children with motor dysfunction, validity was examined in several ways. Change scores on the GMFM were correlated with the ratings of change by independent physical therapists from videotapes of the initial and 6-month tests (test order was randomized) for 28 children. Each child's parent and physical therapist also rated change on a 15-point scale. The correlation between the GMFM change scores and the ratings made from the videotapes was .82. Correlations between the GMFM change scores and the parent and therapist ratings were .54 and .65, respectively. Children rated by both the parent and therapist as having made no change did not have changes on the GMFM.

Among the children without motor delays, those who were younger than 3 years of age made greater gains than did those who were older, confirming that the GMFM is weighted toward motor functions that develop during the first 3 years. Additional analyses supported the following hypotheses: (1) younger children with CP will have higher GMFM change scores than older children with CP will have, (2) children with mild CP will. have higher GMFM change scores than children with severe CP will have, and (3) children recovering from acute head injury will have higher GMFM change scores than children with CP will have.

The purpose of our study was to compare the GMFM with the PDMS-GM when used to measure change in infants with CP and infants with motor delays but no CP. Based on previous research,[20,24] we believe that the GMFM is the most appropriate evaluative measure for infants with CP. The 2 questions addressed in the study were: (1) Is there a difference between mean change scores on the PDMS-GM and GMFM for infants with CP and infants with motor delays over a period of 6 months? and (2) Which of the 2 measures is more responsive to change in infants receiving physical therapy? Our validity construct was that infants with motor delays would demonstrate more change in motor development compared with infants with CP. We hypothesized that mean change scores on the GMFM would be higher than mean change scores on the PDMS-GM.

Method

Subjects

A sample-of-convenience method was used to select 50 infants with either a diagnosis of CP or a motor delay who were receiving physical therapy. To be eligible to participate in the study, infants had to meet the following criteria:

1. Be between 6 and 24 months of age at initial testing.

2. Have a medically documented diagnosis of CP or a motor delay. Motor delay was defined as having a z score of -1.5 or lower on the PDMS-GM.

3. Be receiving outpatient physical therapy, or be enrolled in an early intervention program that included direct physical therapy, at least once every 2 weeks.

4. Demonstrate ability to imitate actions or to interact with toys and caregiver.

5. Be medically stable for therapy (as determined by the child's pediatrician).

6. Have informed consent of a parent or guardian.

Infants with uncontrollable seizures, progressive neurological diseases, fetal alcohol syndrome, acquired immunodeficiency syndrome, blindness, or deafness were excluded. The GMFM and the PDMS-GM were administered to the infants 3 times over a 6-month period (ie, initially and at 3 and 6 months after the initial test session).

Forty-two of the 50 infants completed all 3 test sessions. Data for 8 subjects who did not complete either the 3-month test session or the 6-month test session were not included in the data analysis. Of the 8 children who were dropped from the study, 3 children could not be tested within 3 weeks of their 3-month test date, and 5 children did not complete the 6-month test session (1 child relocated, 1 child was hospitalized, 2 children missed their appointments, and 1 child dropped out of the study).

The sample consisted of 24 infants with CP and 18 infants with motor delays. The characteristics of the 42 infants are presented in Table 1. The infants' mean chronological age at the onset of the study was 15.2 months (SD=5.9, range=5.2-25.0). The mean corrected age was 13.9 months (SD=6.1, range=4.2-24.2). There were 23 male infants and 19 female infants. Thirty infants were Caucasian, 5 infants were Hispanic, 3 infants were African American, and 4 infants were Arabic. Infants with motor delays included 2 infants with macrocephaly, 1 infant with fragile X syndrome, and 1 infant with Klinefelter syndrome. Forty-five percent of the infants were born prematurely (mean gestational age=29 weeks, SD=3.7, range=25-37), with birth weights ranging from 500 to 2,300 g (X=907, SD=449). Seven infants were from multiple births. During the course of the study, 4 infants in the motor delay group were diagnosed as having CP. They were therefore assigned to the group with CP.

Table 1.
Sample Characteristics

                                   Total Sample (N=42)
                                 X         SD      Range

Chronological age (mo)          15.2      5.8     5.2-25.0
Adjusted age (mo)               13.9      6.1     4.2-24.2
Sex
  Female                        19
  Male                          23
Race
  Caucasian                     30
  Hispanic                       5
  African American               3
  Arabic                         4

                                  Infants With Cerebral
                                         (n=18)
                                 X         SD      Range

Chronological age (mo)          13.7      5.7     6.1-22.1
Adjusted age (mo)               12.9      6.2     4.2-22.0
Sex
  Female                         8
  Male                          10
Race
  Caucasian                     12
  Hispanic                       2
  African American               1
  Arabic                         3

                                      Palsy (n=24)
                                 X         SD        Range

Chronological age (mo)          16.2      5.8      5.2-25.0
Adjusted age (mo)               14.6      6.2      4.2-24.2
Sex
  Female                        11
  Male                          13
Race
  Caucasian                     18
  Hispanic                       3
  African American               2
  Arabic                         1

Instrumentation

The manuals for the PDMS-GM[21] and the GMFM[24] were used to administer and score the assessments. A Panasonic VHS 180 (AF piezo) camcorder(*) was used to record and score the test sessions.

Interrater reliability and test-retest reliability were established for the GMFM and the PDMS-GM, based on data obtained for 9 children who were receiving early intervention but who were not part of the study sample. To establish interrater reliability, 6 infants were rated by the first 2 authors (THAK and RJP), who were both experienced in the administration of the tests. Reliability was calculated using intraclass correlation coefficients (ICC[2,1]) (randomly selected group of judges, with each judge rating each subject).[27] The overall ICCs were .99 tar the GMFM and .95 for the PDMS-GM. Both tests were administered to 3 additional children using the same protocol that was used in this study. The first author tested the infants. The infants' performance was scored from a videotape by the first author and one other experienced pediatric physical therapist. The ICCs were .99 for the GMFM and .98 tar the PDMS-GM.

To establish test-retest reliability, the GMFM and the PDMS-GM were again administered within 1 week to 6 of the 9 children who participated in the interrater reliability study. The first author read ministered all tests. Three children were scored from direct observation, and the other 3 children were scored from a videotape. Intraclass correlation coefficients (3,1) (one rater rates subjects) were used to determine the test-retest reliability. The overall ICCs were .97 for the GMFM and .99 tar the PDMS-GM.

Testing Procedure

We used a descriptive longitudinal design for repeated measures in our study initially and at 3 and 6 months alter initial testing. The items from the GMFM and the PDMS-GM were administered to the children during each test session. Test items were administered according to the procedure described in each of the test manuals. Standardized procedures do not require that items on the PDMS and the GMFM be administered in a specific order. Items, therefore, were administered by position (ie, prone, supine, sitting, and standing) to capitalize on natural movement sequences and to minimize position changes. Self-initiated movement transitions between positions were not restricted. To minimize fatigue, items common to both tests were administered once, but the criteria for scoring were based on each test manual. All test sessions were videotaped. All except one of the infants were tested with a parent in the room. A test session was terminated only if the infant was noncooperative. Noncooperation was operationally defined as the infant's refusal to perform more than 25% of the items administered. Most infants were tested as close to the 3-month testing date as possible, with retest days ranging from 1 day to 2.4 weeks before or after the 3and 6-month testing dates. There was no difference between the 2 groups with regard to the time of retesting. An attempt was made to test each infant in the same room and at the same time for each session. All except 7 of the infants were tested at the center at which they received therapy. Three of the 7 infants were tested in their homes only at 6 months due to scheduling problems. The other 4 infants (quadruplets) received home-based therapy, and all their testing was done at home. Each test session lasted approximately 1 to 1 1/2 hours. Overall, the infants handled the 1 to 1 1/2 hours of testing very well. Occasionally, short breaks were taken to console the infants, but fatigue was not an issue. It took less time to complete the testing with infants with severe disabilities and those below 1 year of age. The more competent the infant, the longer it took to complete testing because these infants attempted more test items than did infants who were less able.

The GMFM and PDMS-GM items were scored using the criteria described in each of the test manuals. Most of the items were scored from the videotapes. Some of the items that involved distance were scored during testing. First, the GMFM test items were scored. The videotape was then rewound, and the PDMS-GM items were scored. Although there is a considerable amount of overlap between the GMFM and the PDMS-GM items, the criteria for scoring the items are different.

Data Analysis

The repeated-measures ANOVA approach developed by Stratford and Walter[25,26] was used to compare the GMFM and the PDMS-GM. The GMFM and PDMS-GM are scored using different units of measurement. Scores, therefore, had to be converted to a common metric to permit a direct comparison of the 2 measures.[26] This procedure involved converting the raw scores for each measure to a standardized score with a mean of 0 and a pooled within-subjects standard deviation of 1. The GMFM and PDMS-GM scores were standardized using... the grand mean tar each measure (ie, the combined mean of all raw scores for the 3 test sessions) and a standard deviation based on the within-subjects error term. The within-subjects error term was obtained from a 2-factor repeated-measures ANOVA containing 1 between-group factor and 1 within-subjects factor. A between-group factor is one in which each subject is represented by only one level of the factor. In this analysis, the between-group factor was diagnosis, and the 2 levels of the factor were infants with CP and infants with motor delays. A within-subjects factor is one in which each subject is represented in all levels of the factor. In this analysis, the within-subjects factor was the repeated measurements on each infant at times 1, 2, and 3.

A 3-factor repeated-measures ANOVA (1 between-group factor and 2 within-subjects factors) was used to compare the standardized scores for the GMFM and PDMS-GM. Two analyses were performed, one to compare the GMFM percentage scores with the PDMS-GM age-equivalent scores and the other to compare the GMFM percentage scores with the PDMS-GM scaled scores. The between-group factor was diagnosis, and the 2 levels of the factor were infants with CP and infants with motor delays. The 2 within-subjects factors were measure (GMFM and PDMS-GM) and time (1, 2, and 3). The diagnosis x time interaction compared the change in motor development of the infants with CP and the infants with motor delays. The measure X time x diagnosis interaction compared the ability of the GMFM and PDMS-GM to assess differential amounts of change in the 2 diagnostic groups. The .05 probability level was used to test for statistical significance.

Results

Means and standard deviations for GMFM percentage scores and PDMS-GM age-equivalent and scaled scores are reported in Table 2. For the 6-month period, the increase in mean PDMS-GM age-equivalent scores was 3.8 months for infants with motor delays and 1.8 months for infants with CP. The increase in mean PDMS-GM scaled scores was 35 points for infants with motor delays and 13 points for the infants with CP. The increase in mean GMFM percentage scores was 12.2% for infants with motor delays and 4.2% for infants with CP.

Table 2.
Mean Test Scores and Standard Deviations by Group

                                           Infants With Motor
                                              Delay (n= 18)
Adjusted Age (mo)                       X         SD      Range
PDMS-GM(a) age-equivalent scores
  Initial                              6.6       4.8     1.0-15.0
  3 mo                                 8.5       4.4     1.0-16.0
  6 mo                                 10.4      4.5     1.0-19.0

GMFM(b) standard percentage scores
  Initial                              30.2      19.7    4.4-65.4
  3 mo                                 37.3      18.8    7.4-67.6
  6 mo                                 42.4      17.0    9.3-75.8

PDMS-GM scaled scores
  Initial                              385        63     205-453
  3 mo                                 408        44     293-458
  6 mo                                 420        42     300-470

                                         Infants With Cerebral
                                             Palsy (n=24)
Adjusted Age (mo)                       X         SD     Range
PDMS-GM(a) age-equivalent scores
  Initial                              5.4       4.1    1.0-14.0
  3 mo                                 6.3       4.2    1.0-16.0
  6 mo                                 7.2       4.4    1.0-16.0

GMFM(b) standard percentage scores
  Initial                              24.6      17.1   2.3-54.9
  3 mo                                 27.7      17.8   1.8-57.2
  6 mo                                 28.8      16.8   1.0-56.2

PDMS-GM scaled scores
  Initial                              384        41    292-447
  3 mo                                 392        39    293-457
  6 mo                                 397        42    277-455

(a) PDMS-GM=Peabody Developmental Gross Motor Scale.

(b) GMFM=Gross Motor Function Measure.

The standardized GMFM and PDMS-GM scores are presented in Table 3. The magnitude of the standardized scores was higher for each successive testing period for both the GMFM and the PDMS-GM. The results of the 3-factor repeated-measures ANOVA revealed a significant time x diagnosis interaction for both the comparison of the GMFM percentage scores with the PDMS-GM age-equivalent scores (F=8.37; df=2,80; P [is less than] .001) and the comparison of the GMFM percentage scores with the PDMS-GM scaled scores (F=9.10; df=2,80; P [is less than] .001). These findings support our validity construct that infants with motor delays would demonstrate a greater increase in motor development at each 3-month interval compared with infants with CP.

Table 3.
Standardized Gross Motor Function Measure (GMFM) and Peabody
Developmental Gross Motor Scale (PDMS-GM) Scores

                                       Infants With Motor
                                       Delay (n= 18)
Test                                     X       SD

PDMS-GM age-equivalent scores
  Initial                              -0.55    4.63
  3 mo                                  1.24    4.32
  6 mo                                  3.08    4.41
PDMS-GM scaled scores
  Initial                              -0.96    5.12
  3 mo                                  0.89    3.60
  6 mo                                  1.90    3.38
GMFM standard percentage scores
  Initial                              -0.16    3.27
  3 mo                                  1.02    3.10
  6 mo                                  1.86    2.82

                                         Infants With
                                        Cerebral Palsy
                                           (n=24)
Test                                     X        SD

PDMS-GM age-equivalent scores
  Initial                              -1.80    3.96
  3 mo                                 -0.94    4.11
  6 mo                                 -0.01    4.31
PDMS-GM scaled scores
  Initial                              -1.03    3.33
  3 mo                                 -0.40    3.18
  6 mo                                  0.06    3.39
GMFM standard percentage scores
  Initial                              -1.08    2.83
  3 mo                                 -0.57    2.95
  6 mo                                 -0.38    2.78

                                          Total Sample
                                            (N=42)
Test                                      X        SD

PDMS-GM age-equivalent scores
  Initial                               1.26    4.23
  3 mo                                 -0.01    4.25
  6 mo                                  1.27    4.58
PDMS-GM scaled scores
  Initial                              -1.00    4.14
  3 mo                                  0.16    3.39
  6 mo                                  0.85    3.47
GMFM standard percentage scores
  Initial                              -0.69    3.02
  3 mo                                  0.11    3.08
  6 mo                                  0.38    2.98

The measure x time x diagnosis interaction was not statistically significant for either the comparison of the GMFM percentage scores with the PDMS-GM age-equivalent scores (F=1.23; df=2,80; P=.289) or the comparison of the GMFM percentage scores with the PDMS-GM scaled scores (F=0.71; df-2,80; P=.497) (Figure). Our hypothesis that the GMFM is more responsive to change than the PDMS-GM was not supported. Both measures performed in a similar manner for the 2 groups of infants over the 6-month period.

[Figure ILLUSTRATION OMITTED]

Discussion

The amount of change in mean GMFM and PDMS-GM scores differed based on infant diagnosis. Although mean scores increased for both groups over the 6-month period, infants with motor delays made greater gains in motor development compared with infants with CP. The results support the validity construct that infants with motor delays have greater potential for change than infants with CP. This assumption was integral to the 3-factor repeated-measures ANOVA procedure that was used to directly compare the responsiveness of the GMFM and the PDMS-GM.

Our findings for the GMFM are consistent with the results reported by Russell et al[24] for a 6-month period, although the mean change scores for infants with CP were somewhat lower than the mean change scores reported by Russell et al. In our study, infants with motor delays demonstrated an increase of 12.2% in mean GMFM scores, whereas infants with CP showed an increase of 4.2% in mean GMFM scores for the 6-month period. As part of the validity study for the GMFM, Russell et al[24] reported a mean change score of 11.3% for 28 children between 1 month and 4.3 years of age without motor delays. The mean change scores for the 32 children with CP who were less than 3 years of age varied, based on therapist judgment of severity. Children judged to have "mild" CP had an increase in mean scores of 11.5%, children classified as having "moderate" CP had an increase in mean scores of 6.4%, and children classified as having "severe" CP had an increase in mean scores of 5.0%.

The primary objective of our study was to compare the responsiveness of the GMFM and the PDMS-GM for measuring change in motor development over a 6-month period in infants receiving physical therapy. The measure x time x diagnosis interaction was not significant, indicating that the GMFM was not more responsive than the PDMS-GM to the changes recorded for the infants in this study. The statistical power of the comparison was low (about .26 for the PDMS age-equivalent comparison and about .17 for the scaled score comparison). The finding of no difference between the change scores of the 2 tests may be explained by several factors, including the overlap in GMFM and PDMS-GM items, the administration procedure for the GMFM, the low frequency in which infants with CP made partial progress on GMFM, and the inclusion of separate items for the right and left sides on the GMFM.

There is considerable overlap in the content of items of the GMFM and the PDMS-GM. At least 50% of the items of the GMFM are similar to items of the PDMS-GM. Most of the similarity is in items in which the tasks are likely to be achieved by infants (eg, prone, supine, rolling, and sitting items). At older age levels, the emphasis of the PDMS-GM is on abilities such as running, jumping, walking on a balance beam, and ball-handling skills that exceed the potential of most children with CP. Consequently, differences in the responsiveness of the 2 measures may be more pronounced for children with CP above the age of 3 years, especially children who are unable to walk without the use of walkers or canes.

The standardized procedure for administration of the GMFM may have compromised the magnitude of the GMFM change scores. All GMFM items are administered, regardless of the child's age, and only motor behaviors observed by the examiner during testing are scored. In contrast, the standardized procedure for administration of the PDMS-GM includes criteria for determining basal and ceiling age levels. As infants progress in motor development, items from higher age levels of the PDMS-GM are administered to them. Items that are below the child's basal age level are not administered, and full credit is received. In our study, infants demonstrated some movements during the initial test session that were not observed during subsequent test sessions. For the GMFM, this occurred most often for the items in lying and rolling dimension and for some of the items in the sitting dimension. This inconsistency was particularly observed in younger infants (9-20 months of age) who exhibited improved motor development.

The following example illustrates how the change in GMFM scores was minimized when an infant did not perform items in the lying and rolling dimension that were previously passed. An infant who passed most of the items in the lying and rolling dimension during the initial testing would receive a high percentage score for this dimension. If, during subsequent testing, the infant progressed to independent sitting (eg, at 3 months) or creeping (eg, at 6 months), the infant's percentage scores for the sitting dimension and the crawling and kneeling dimension would increase. If the infant, however, refused to perform items in supine and prone positions that were previously demonstrated, the scores for the lying and rolling dimension would be lower compared with the scores obtained during the initial assessment. Because all dimensions are equally weighted, the total percentage score may show little change.

The 4-point scale for scoring items of the GMFM did not appear to be more advantageous than the 3-point scale of the PDMS-GM for infants with CP. At the second and third test sessions, the frequency at which infants with CP received GMFM item scores of 2 (partially completed the action) was lower than anticipated. At each 3-month interval, the infants with CP performed the task for an item at the criterion level, received a score of 1, or were not able to perform the task. For example, for items 46 and 47, infants with CP tended to either (1) crawl up 4 steps (score of 3) or (2) not crawl up any steps (score of 0). For item 45, infants with CP tended to either (1) crawl forward less than 0.61 m (2 ft), stop (sat up), and then crawl another foot or so to get a toy (score of 1) or (2) crawl more than 1.8 m (6 ft) (score of 3) during the same session.

The GMFM contains separate items for assessing right- and left-side body movements. Scores, therefore, may be lower for the GMFM than for the PDMS-GM for infants with asymmetrical movements. In our study, infants with asymmetric movement patterns passed PDMS-GM items but failed related items of the GMFM when required to use the affected side. During subsequent testing, the infants made gains on the PDMS-GM even if improvement in the affected side was minimal. The appropriateness of the PDMS-GM for infants with hemiplegia or infants with problems associated with asymmetrical use of the extremities should be considered when selecting a tool for evaluative purposes.

Clinical Implications

Although the findings suggest that the GMFM and the PDMS-GM are comparable when used to evaluate change in infants, the purpose of testing and the infant's age and diagnosis are important considerations in test selection. The GMFM may be less appropriate for infants (particularly between 9 and 18 months of age) who do not consistently follow verbal instructions or imitate a demonstration or for infants with behavioral problems. Because an infant can receive credit only for motor behaviors that are observed during the assessment, the GMFM score is greatly influenced by the infant's cooperation with the standardized administration procedures. As previously discussed, most of the younger infants, particularly those who at the second or third test session had progressed to more advanced motor patterns such as crawling on hands and knees, resisted performing movements performed on initial testing such as creeping on the belly. When this resistance occurs, the total GMFM score may not completely reflect an infant's gross motor capabilities.

The tendency for young infants to refuse to perform previously acquired motor patterns is not unique to the infants in our study. Gesell[28] and Bayley[29] have described infants' preoccupation with newly acquired motor abilities, when more advanced motor patterns replace the immature, less-advanced motor patterns. This was usually the case when the immature patterns were considered to be prerequisites for the new patterns. To maintain consistency in scoring the infants' true level of change in development, we propose using the modified scoring criteria presented in Table 4. The initial GMFM would be administered and scored according to the procedure stated in the manual.[24] If, during subsequent assessment sessions, an infant is observed to perform and pass higher items in the sitting, crawling and kneeling, and standing dimensions but refuses to perform items passed during the initial assessment, items that are part of a motor sequence of items passed initially would be credited. For example, if the infant passes items 31 and 32 (sits on mat with feet in front, attains 4-point kneeling over right or left side) during reassessment, then the following items, if observed initially, should be credited: items 6 and 7 (supine, reaches with right or left arm across midline), items 21 and 22 (sits on mat, supported at thorax by therapist, lifts head to midline, maintains for 3 and 10 seconds, respectively), and item 23 (sits on mat with feet in front, maintains arm propping for 5 seconds).

Table 4.
Modified Scoring Criteria for the Gross Motor Function Measure

Items Passed at 3 or 6 Months       Items Credited if
                                       Observed Initially

Item 39- 4-point kneeling,          Item 10-prone, arms at sides,
  maintains weight on hand            lifts head vertical
  and knees, for 10 seconds
                                    Item 11-prone on forearms,
                                      lifts head 90 degrees
                                      with elbows extended
Items 42 and 43-4-point             Items 12 and 13-prone on
  kneeling, reaches forward           forearms, weight on the
  with right or left hand             right or left forearm,
  above shoulder level                fully extends the
                                      opposite arm forward
Item 45-4-point kneeling,           Item 38-prone, creeps
  crawls reciprocally                 forward 1.8 m
  forward 1.8 m (6 ft)

Item 40- 4-point kneeling,          Items 14 and 15-prone,
  attains sitting position,           roll to supine position
  then maintains arms free            over right or left side

Item 24 -sits on mat with           Item 21 -sits on mat supported
  feet in front, maintains            at thorax by therapist,
arms free for 3 seconds               lifts head upright,
                                      maintains for 3 seconds
                                    Item 22-sits on mat, supported
                                      at thorax by therapist, lifts
                                      head to midline, maintains for
                                      10 seconds
                                    Item 23-sits on mat with feet
                                      in front, maintains arm
                                      propping for 5 seconds

Items 31 and 32-sits on mat         Items 6 and 7-supine, reaches
  with feet in front, attains         with right or left arm
  4-point kneeling over               across midline, plus
  right or left side                  items 21, 23, and 32

Our findings demonstrate that no one test measures all aspects of change in infants with CP. Although the infants demonstrated changes on both the GMFM and PDMS-GM, the results do not provide information on the meaningfulness of the changes. The meaningfulness of an increase in age-equivalent scores of 1.8 months on the PDMS-GM or an increase of 4.2% on the GMFM needs further exploration. According to Guyatt et al,[19] the responsiveness of a test must be directly related to the magnitude and meaningfulness of the change in the scores. Levels of assistance required to perform a task can reflect important progress in infants with motor disabilities. Neither of the tests include scoring criteria for the level of assistance required by the infant to perform a task. Children with CP often must initially rely on assistance from caretakers to perform gross motor movements that infants without disabilities perform independently. Level of caregiver assistance, therefore, may be an important component of a measure constructed to evaluate change in children with CP. The Pediatric Evaluation of Disability Inventory[30] is unique for incorporating level of assistance into scoring criteria. Until information on the meaningfulness of GMFM and PDMS-GM change scores is available, the measures must be supplemented with other observational and functional measures including, parental reports.

Research Implications

Our Findings indicate that intervention studies involving children with CP need to be at least 3 months' duration and more likely of 6 months' duration (depending on sample size) if the GMFM or the PDMS-GM is used as an outcome measure. In our study, infants with CP had an increase in mean PDMS-GM age-equivalent scores of 0.9 months for the 3-month test session and a mean increase of 1.8 months for the 6-month test session. Similarly, the infants with CP had increases in mean GMFM scores of 3.1% at 3 months and 4.2% at 6 months. Our impression from viewing the videotapes is that it took longer for the infants with CP to become proficient in performing recently acquired postures or movements independently compared with the infants with motor delays. Early movements were often guarded, and infants demonstrated uncertainty in initiating movements or maintaining certain positions independently. During subsequent testing at 3 or 6 months, the infants with CP often demonstrated confidence and mastery of previous movements, which did not result in additional points based on the criteria of the GMFM.

The results support the need for careful selection of subjects when planning effectiveness studies. As demonstrated in our study, diagnosis or severity of motor delay or disability may be an important consideration when hypothesizing the amount of change expected in children receiving physical therapy. If a sample is homogeneous, selection of outcome measures can be based on the purpose and duration of the intervention and the subjects' potential for change. In the area of childhood developmental disability, however, homogeneous samples are difficult to obtain. An alternative strategy for children with CP is to classify each subject's level of gross motor function[31] and control for the effects of severity when analyzing the data.

If the results of our study are used to estimate the expected change in motor development for infants with CP (effect size) for a 3- or 6-month period, 3 points need to be considered. First, the infants in our study were receiving ongoing physical therapy, at least once every 2 weeks. This point is particularly important when planning the frequency of intervention. Second, although no formal assessment of socioeconomic status was made, the majority of the infants were from middle-class families. Third, the purpose of the study was to compare the GMFM and the PDMS-GM, not to determine the factors contributing to the changes in the infants' motor development. In addition to physical therapy, the infants' growth and maturation and other forms of play and physical activity may have contributed to the gains in motor development. The data in this study, therefore, reflect changes in the motor development of infants from middle-class families who received physical therapy, on the average, twice weekly for 6 months.

Limitations of the Study

The study addressed only one aspect of responsiveness, which was to determine whether test scores improved over time and the magnitude of the change scores. The second aspect of responsiveness concerns the meaningfulness of the changes. The question of whether the mean score changes were large enough to constitute meaningful functional change or whether the changes were meaningful to the infants' caregivers was not addressed. The findings cannot be generalized to children over the age of 30 months or to infants with diagnoses other than CP or nonspecific motor delay. Furthermore, the sample was not randomly selected, and the statistical power of the comparison was low. The numbers of infants in each group was small, limiting the ability to generalize the results and increasing the potential for a Type II error (incorrect conclusion that there is no difference in responsiveness between the GMFM and the PDMS-GM).

Conclusions

The purpose of our study was to compare the responsiveness of the norm-referenced PDMS-GM with that of the criterion-referenced GMFM, which was constructed to measure change in infants with motor disabilities. The results revealed that both tests performed similarly with the cohort of infants in the study over a 6-month period. Our hypothesis that the GMFM would be more responsive than the PDMS-GM was not supported.

Several problems associated with the standardized test administration and scoring are believed to have compromised the magnitude of percentage-of-change scores obtained for the GMFM in this study. These problems include the overlap in GMFM and PDMS-GM items, the GMFM standard administration and scoring procedure whereby an infant can receive credit only for motor behaviors that are observed during the assessment, the low frequency in which infants with CP made partial progress on GMFM, and the inclusion of separate items for the right and left sides in the GMFM. Young infants (9-20 months of age) were particularly affected by the administration and scoring criteria, which required direct observation. The findings suggest that the GMFM may be less appropriate for infants (particularly between 9 and 18 months of age) who do not consistently follow verbal instructions or imitate a demonstration or for infants with behavioral problems. To increase the responsiveness of the GMFM for this age group, a modified scoring procedure is proposed.

Our findings could be used to calculate effect size and to estimate the expected duration of intervention in efficacy studies using the GMFM and the PDMS-GM. Methods of evaluating the meaningfulness of the change were not explored in this study. Studies designed to determine the meaningfulness and magnitude of the changes in motor development of infants with motor disabilities should include ecological observations and input from parents and therapists or other professionals involved with the child.

Acknowledgments

Appreciation is expressed to Louisa Serraydarian, PhD, for her statistical advice during the development of the project; Susan Effgen, PhD, PT, Nancy Ellis, PhD, and Jean Barr, for their input during the development and completion of the project; physical therapists at DuPage Easter Seal, Gilchrist Marchman Easter Seal, Kids in Motion, Brandecker Easter Seal, and Arlington Pediatrics for their assistance in recruitment of subjects; and the parents and infants from these centers, without whom this work could not have been accomplished.

References

[1] Harris SR. Effects of neurodevelopmental therapy on motor performance of infants with Down's syndrome. Der Med Child Neurol. 1981; 23:477-483.

[2] Ottenbacher KJ Short MA, Watson PJ. The effects of a clinically applied program of vestibular stimulation on the neuromotor performance of children with severe developmental disability. Physical and Occupational Therapy in Pediatrics. 1981;1:1-11.

[3] Simeonsson RJ, Cooper DH, Scheiner AP. A review and analysis of the effectiveness of early intervention programs. Pediatrics. 1982;69: 635-641.

[4] Jenskins JR, Sells CJ, Brady D, et al. The effects of developmental therapy on motor impaired children. Physical and Occupational Therapy in Pediatrics. 1982;2:19-28.

[] Ottenbacher KJ, Biocca Z, DeCremer G, et al. Quantitative analysis of the effectiveness of pediatric therapy: emphasis on the neurodevelopmental treatment approach. Phys Ther. 1986:66:1095-1101.

[6] Palmer FB, Shapiro BK, Wachtel RC, et al. The effects of physical therapy on cerebral palsy: a controlled trial in infants with spastic diplegia. N Engl J Med. 1988;318:803-808.

[7] Piper MC, Kunos VI, Willis DM, et al. Early physical therapy effects on the high-risk infant: a randomized controlled trial. Pediatrics. 1986;78:216-224.

[8] Goodman M, Rothberg A, Houston-McMillan JE, et al. Effect of early neurodevelopmental therapy in normal and at-risk survivors of neonatal intensive care. Lancet. 1985;2 (8468):1327-1330.

[9] Harris SR. Early intervention: does developmental therapy make a difference? Topics in Early Childhood Education. 1988;7:20-32.

[10] Piper MC. Efficacy of physical therapy: rate of motor development in children with cerebral palsy. Pediatric Physical Therapy. 1990;3: 1877-1880.

[11] Harris SR, Heriza CB. Measuring infant movement: clinical and technological assessment techniques. Phys Ther. 1987;67: 1877-1880.

[12] Campbell SK. On the importance of being earnest about measurement, or, how can we be sure that what we know is true? Phys Ther. 1987;67:1831-1833.

[13] Rosenbaum PL, Russell DJ, Cadman DT, et al. Issues in measuring change in motor function in children with cerebral palsy: a special communication. Phys Ther. 1990;70:125-131.

[14] Stern FM, Gorga D. Neurodevelopmental treatment (NDT): therapeutic intervention and its efficacy. Infants and Young Children. 1988;1: 22-32.

[15] Boyce WF, Gowland C, Rosenbaum PL, et al. Measuring quality of movement in cerebral palsy: a review of instruments. Phys Ther. 1991;71:813-819.

[16] Girolami GL. Evaluating the Effectiveness of a Neurodevelopmental Treatment Program to Improve Motor Control of High-Risk Preterm Infants. Chapel Hill, NC: University of North Carolina at Chapel Hill; 1987. Master's thesis.

[17] Lawlor MC, Henderson A. A descriptive study of the clinical practice patterns of occupational therapists working with infants and young children. Am J Occup Ther. 1989;43:755-764.

[18] Kirshner B, Guyatt G. A methodological framework for assessing health indices. Journal of Chronic Diseases. 1985;38:27-36.

[19] Guyatt G, Walter S, Norman G. Measuring change over time: assessing the usefulness of evaluative instruments. Journal of Chronic Diseases. 1987;40:171-178.

[20] Palisano RJ, Kolobe TH, Haley SM, et al. Validity of the Peabody Developmental Gross Motor Scale as an evaluative measure of infants receiving physical therapy. Phys 7'her. 1995;75:939-948.

[21] Folio MR, Fewell RR. Peabody Developmental Motor Scales and Activity Cards. Allen, Tex: DLM Teaching Resources; 1983.

[22] Palisano RJ. Use of chronological and adjusted ages to compare motor development of healthy preterm and fullterm infants. Dev Med Child Neurol. 1986;28:180-187.

[23] Provost B, Harris MB, Ross K, Michnal D. A comparison of scores on two preschool assessment tools: implications for theory and practice. Physical and Occupational Therapy in Pediatrics. 1988;8:35-51.

[24] Russell DJ, Rosenbaum PL, Gowland C, et al. Gross Motor Function Measure Manual. 2nd ed. Owen Sound, Ontario, Canada: c/o Mary Lane, Pediatric Physiotherapy Services; 1993.

[25] Stratford PW, Levy D, Gauldie S, et al. Extensor carpi radialis tendinitis: a validation of selected outcomes measures. Physiotherapy Canada. 1987;39:250-255.

[26] Stratford PW, Binkley JM, Riddle DL. Health status measures: strategies and analytic methods for assessing change scores. Phys Ther. 1996;76:1109-1123.

[27] Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reliability. Psychol Bull. 1979;86:420-428.

[28] Gesell A. The ontogenesis of infant behavior. In: Carmichael L, ed. Manual of Child Psychology. 2nd ed. New York, NY: John Wiley & Sons Inc; 1954:335-373.

[29] Bayley N. The Bayley &ales of Infant Development. New York, NY: The Psychological Corporation; 1969.

[30] Haley SM, Coster WJ, Ludlow LH, et al. Pediatric Evaluation of Disability Inventory: Development, Standardization, and Administration Manual. Boston, Mass: New England Medical Hospitals; 1992.

[31] Palisano RJ, Rosenbaum PL, Walter S, et al. Development and reliability of a system to classify gross motor function in children with cerebral palsy. Dev ivied Child Neurol. 1997;39:214-223.

THA Kolobe, PhD, PT, is Assistant Professor, Department of Physical Therapy, University of Illinois at Chicago, 1919 W Taylor St, M/C 898, Chicago, IL 60612 (USA) ([e-mail]). This study was completed in partial fulfillment of Dr Kolobe's doctoral studies in pediatric physical therapy at Hahnemann University, Philadelphia, Pa. Address all correspondence to Dr Kolobe.

RJ Palisano, ScD, PT, is Professor, Department of Physical Therapy, Allegheny University of the Health Sciences, Philadelphia, Pa, and Co-Investigator, Neurodevelopmental Clinical Research Unit, McMaster University, Hamilton, Ontario, Canada.

PW Stratford, PT, is Associate Professor, School of Rehabilitation Sciences, and Associate Member, Department of Clinical Epidemiology and Biostatistics, McMaster University.

This study was approved by Hahnemann University Human Subjects Review Committee.

This work was supported in part by grants from the Section on Pediatrics of the American Physical Therapy Association and United Nations Educational Programs for Southern Africa.

This article was submitted December 19, 1997, and was accepted April 13, 1998.

(*) Matsushita Industrial Co Inc, 1360 Higashi-Hirashima, Okayama City, Okayama, Japan.