Reanalyzing the data.

To the Editor:

We were excited to read the September 2001 issue of Physical Therapy in which Ketelaar et al reported the outcomes of a randomized controlled clinical trial on functional therapy in children with cerebral palsy (CP). (1) To our knowledge, this is the first study ever published on functional therapy in pediatric physical therapy.

We would like to emphasize the value of evidence showing improvements in "activities" as compared with evidence showing changes in impairments. Ketelaar and colleagues reported that therapy designed to enhance function has positive effects on performance of daily functional motor skills as measured by the Pediatric Evaluation of Disability Inventory (PEDI). However, no differences between the intervention group and the reference group were found for improvement in basic gross motor abilities such as standing, walking, running, and jumping.

In the "Discussion" section, the authors addressed the differences in the types of outcome measurements provided by the PEDI and the Gross Motor Function Measure (GMFM), such as differences in the method of administration (interview of the parent [PEDI] versus observation of the child [GMFM]) and the setting (performance in the daily environment [PEDI] versus a standardized environment at the therapist's clinic [GMFM]).

The study, however, was conducted between 1995 and 1997, and the authors wrote their manuscript in 2000. Since then, 2 major developments in the field of outcome measurement in pediatric rehabilitation have occurred that may have implications for analysis and interpretation of the data from this study and for future research.

First, further development and improvement of the GMFM as an outcome measure (from an ordinal (2) scale into an interval scale) has been reported. (3) Ketelaar et al used the original 88-item measure (GMFM-88), described by Russell et al, (2) which was the only version available at the time of the study. In the past decade, some of the limitations of the measure have became become apparent, in particular, the interpretation of change in total scores. Due to a "ceiling effect," the responsiveness of the total GMFM-88 score to change in children with relatively advanced motor skills is limited. Based on the reported dimension scores in their article, this might well be the case in the trial of Ketelaar et al. We agree with the authors that such high scores could have masked development of skills. Furthermore, the interpretation of change scores between groups with the GMFM-88 is not straightforward, because it is an ordinal scale and not an interval scale. Therefore, with current knowledge, we would recommend a review of the data with the use of the now-available revised GMFM (GMFM-66), which is an interval scale and has improved properties in terms of interpretability. (3)

We have done such an analysis (Table) and have found positive change scores of 9.1% and 9.7% on the GMFM-66 in the reference and experimental groups, respectively. Although these change scores are in accordance with the changes measured by the total GMFM-88 score (7.5% and 8.2%, respectively), the interpretation of the differences between the GMFM-66 and GMFM-88 scores is different. The lower mean scores at follow-up (79.2 and 79.0) imply less of a ceiling effect with the GMFM-66 compared with the GMFM-88 (94.1 and 94.6, respectively).

Second, information on patterns of motor development in children with CP has become available. (4) Outcomes of interventions in children, we believe, should be judged with respect to the patterns of motor development in comparable children with CP. Cross-sectional data on children with CP indicate that development of motor abilities (as measured by the GMFM-88) is related to age and the severity of CP (as described by means of the Gross Motor Function Classification System). (4) Furthermore, the rate of increase in GMFM score is related to age, with the steepest slope during infancy and early childhood and flattening after about 5 years of age. (4) We think that age- and severity-specific analyses in terms of change in motor activity between the 2 groups are now possible and appropriate.

Last, we would like to emphasize that we believe there has been progress in knowledge about the development of children with CP in terms of "activities" and the effects of functional approaches in rehabilitation. We believe that the results of the trial on therapy designed to enhance function will have consequences in pediatric rehabilitation and should influence future research to improve the outcomes of children with CP and their families through further studies of "functional" approaches based on children's and families' identified goals.

Table.

Comparison of the Original Gross Motor Function Measure (GMFM-88)
and Revised Gross Motor Function Measure GMFM-66) Scores (a)

                                         Pretest        Follow-up 1

                               N      [bar]X     SD     [bar]X    SD

Total GMFM-88 Scores (%)
  Referral group              27      86.6     13.6     90.5     9.7
  Functional physical
   therapy group              28      86.4     13.3     90.3     8.7

Total GMFM-66 Scores (%)
  Referral group              27      70.1     12.2     73.2     9.7
  Functional physical
   therapy group              28      69.3     10.4     72.8    10.2

                             Follow-up 2       Follow-up 3

                            [bar]X     SD     [bar]X     SD

Total GMFM-88 Scores (%)
  Referral group             92.5      8.1     94.1      6.5
  Functional physical
   therapy group             93.2      6.9     94.6      5.9

Total GMFM-66 Scores (%)
  Referral group             76.4      9.7     79.2      9.5
  Functional physical
   therapy group             77.3     11.2     79.0     10.9

(a) The authors thank Dr Marjolijn Ketelaar for providing the data
on the GMFM for additional analysis.
Jan Willem Gorter, MD, PhD
Physician in Physical Medicine and
  Rehabilitation
Post-doctoral Fellow at CanChild Centre
  for Childhood Disability Research
Rehabilitation Centre De Hoogstraat
Rembrandtkade 10
3583 TM Utrecht, the Netherlands

Peter L Rosenbaum, MD, FRCP(C)
Professor of Paediatrics
Canada Research Chair in Childhood
  Disability
Faculty of Health Sciences
McMaster University
CanChild Centre for Childhood Disability
  Research
IAHS, Room 408
1400 Main St W
Hamilton, Ontario, Canada, L8S 1C7

References

(1) Ketelaar M, Vermeer A, 't Hart H, et al. Effects of a functional therapy program on motor abilities of children with cerebral palsy. Phys Ther. 2001;81:1534-1545.

(2) Russell DJ, Rosenbaum PL, Cadman DT, et al. The Gross Motor Function Measure: a means to evaluate the effects of physical therapy. Der Med Child Neurol. 1989;31: 341-352.

(3) Russell DJ, Avery LM, Rosenbaum PL, et al. Improved scaling of the Gross Motor Function Measure for children with cerebral palsy: evidence of reliability and validity. Phys Ther. 2000;80:873-885.

(4) Palisano RJ, Hanna SE, Rosenbaum PL, et al. Validation of a model of gross motor function for children with cerebral palsy. Phys Ther. 2000;80:974-985.

Author Response:

I am very grateful for the comments of Dr Gorter and Dr Rosenbaum on our study of effects of a functional therapy program on motor abilities of children with cerebral palsy. (1) The developments that have occurred since the study was conducted, specifically with respect to the Gross Motor Function Measure (GMFM) and classification of children with cerebral palsy (CP), are important.

First, the improved scaling of the GMFM by the 66-item version of the GMFM has many advantages. The hierarchical structure and interval scaling improve the interpretation of total scores and change scores by using the item difficulty map. The GMFM-66 allows calculation of a child's total score and the standard error around an individual's score and can estimate a child's score if some items are missing. (2)

By calculating the GMFM-66 total scores of the data in our study, as Gorter and Rosenbaum have done, it is clear that the ceiling effect is less when compared with the total scores for the original 88-item measure (GMFM-88). However, there are no differences between the 2 scoring methods with regard to the change scores. In my opinion, this might be related to one of the characteristics of the GMFM-66 scores: there are no longer separate scores for each dimension. The item maps allow looking at the dimensions, but these are not reflected in dimension scores. In our study, we explicitly decided not to use total scores, because the children who participated in the study mainly had problems in the standing and walking domain. The use of dimension scores makes the GMFM more sensitive to changes, which may be an artifact of the number of items. Maybe we should think about using goal total scores (3) to make the GMFM sensitive to changes relevant for the individual child, as was done in 2 recent studies. (4,5) Or maybe we should make use of the good psychometric properties, the hierarchical structure, and the interval scale as provided by the GMFM-66.

Currently, there are different scoring systems (GMFM-66 total scores, GMFM-88 total scores, dimension scores, and goal total scores). Depending on the purpose (eg, describing gross motor development in the population of children with CP or effect studies or use in clinical practice), a scoring system should be selected that best fits the question.

The second development Gorter and Rosenbaum mention is the information on patterns of motor development in relation to the development of the Gross Motor Function Classification System (GMFCS) for children with CP. Unfortunately, when we started our study, the GMFCS was not known to us. Although the GMFCS is an important step forward in uniform classification of children with CP, I do not think that severity-specific (using the GMFCS) analysis of the data in our study would have influenced the results of our study because most of the children in our study had only mild involvement in terms of GMFCS level I. Please note that our study was performed in primary health care physical therapy in the Netherlands, which implies that only children with mainly ambulation problems are treated. Children with more complex problems are referred to rehabilitation centers for examination and treatment.

Studies of patterns of motor development, such as that of Palisano and colleagues using cross-sectional data (5) and those now in preparation using longitudinal data, are important. We agree with Gorter and Rosenbaum that outcomes of intervention should be judged with respect to motor development in comparable children with CP. Presently, few data are available on development of gross motor function in relation to age and severity. Therefore, it is important that, in the future, more longitudinal studies be performed in which children are classified using the GMFCS and in which standardized instruments such as the GMFM and the Pediatric Evaluation of Disability Inventory (PEDI) are applied. In the Netherlands, we have just started a large research program (Pediatric Rehabilitation Research in the Netherlands [PERRIN]) on children with CP. In this program, there are 3 (of 5) projects in which children and adolescents in different age groups (0-4 years, 9-15 years, and 16-20 years) will be followed for some years.

Marjolijn Ketelaar, PhD
Research Coordinator
Rehabilitation Center De Hoogstraat
Rembrandtkade 10
3583 TM Utrecht, the Netherlands
([e-mail])

References

(1) Ketelaar M, Vermeer A, 't Hart H, et al. Effects of a functional therapy program on motor abilities of children with cerebral palsy. Phys Ther. 2001;81:1534-1545.

(2) Russell DJ, Avery LM, Rosenbaum PL, et al. Improved scaling of the Gross Motor Function Measure for children with cerebral palsy: evidence of reliability and validity. Phys Ther. 2000; 80:873-885.

(3) Russell DJ, Rosenbaum PL, Cadman DT, et al. The Gross Motor Function Measure: a means to evaluate the effects of physical therapy. Dev Med Child Neurol. 1989;31: 341-352.

(4) Bower E, Michell D, Burnett M, et al. Randomized controlled trial of physiotherapy in 56 children with cerebral palsy followed for 18 months. Dev Med Child Neurol. 2001;43:4-15.

(5) Mall V, Heinen F, Kirschner J, et al. Evaluation of botulinum toxin A therapy in children with adductor spasm by Gross Motor Function Measure. J Child Neurol. 2000;15:214-217.

(6) Palisano RJ, Hanna SE, Rosenbaum PL, et al. Validation of a model of gross motor function for children with cerebral palsy. Phys Ther. 2000;80:974-985.