Article
Rasch analysis of the gross motor function measure: validating the assumptions of the rasch model to create an interval-level Measure1,

https://doi.org/10.1016/S0003-9993(02)04896-7Get rights and content

Abstract

Avery LM, Russell DJ, Raina PS, Walter SD, Rosenbaum PL. Rasch analysis of the Gross Motor Function Measure: validating the assumptions of the Rasch model to create an interval-level measure. Arch Phys Med Rehabil 2003;84:697-705.

Objectives:

To describe the Rasch analysis of the Gross Motor Function Measure (GMFM-88) and to demonstrate how the assumptions of unidimensionality, sample-free measurement, and test-free measurement were validated to create an interval level measure.

Design:

Cross-sectional and longitudinal (12-mo) data from a prospective study of motor development in children with cerebral palsy (CP) were used for the analysis.

Setting:

Motor assessments were completed at 18 children’s ambulatory rehabilitation centers in Ontario, Canada, by pediatric physical therapists trained in the use of the GMFM-88.

Participants:

The first 537 of 682 children enrolled into a longitudinal study of motor development in children with CP. Children had a mean age of 6.43±2.75 years (range, 11mo–12y) with varying types and severity of CP.

Interventions:

Not applicable.

Main Outcome Measure:

The GMFM-88.

Results:

The Rasch analysis, in conjunction with clinical decisions, identified 66 items from the GMFM-88 that formed a unidimensional measure (GMFM-66). Assumptions of sample-free and test-free measurement were confirmed, and a user-friendly scoring program was developed.

Conclusions:

The GMFM-66 is an interval-level measure of gross motor function for children with CP; it should improve the scoring, interpretation, and overall clinical and research utility over the original GMFM.

Section snippets

Subject selection

Data from a large prospective study of motor development in children with CP using the GMFM were used. A detailed description of the sampling procedure is presented elsewhere.12 In brief, 18 of the 19 children’s treatment centers in Ontario, Canada, and 1 additional pediatric treatment facility participated in this study. The vast majority of children with CP in Ontario attend 1 of these 19 treatment centers. Sampling was random within centers, and the overall sample is thought to be

Defining a unidimensional set of items

The choice between critical values for the infit statistic was made based on the properties of the items selected with each threshold. A critical value of 2.0 resulted in a group of 52 items, and a critical value of 3.0 resulted in a group of 68 items. The properties of these 2 groups, plus those of the original 88 items, are summarized in table 2. The result of the repeated-measures ANOVA was not statistically significant for the effect of item group, but the mean change score over 12 months

Discussion

The GMFM-88 is a widely used clinical and research instrument developed with principles of classical test theory. The decision to apply Rasch analysis to the GMFM-88 was made to determine whether we could improve the scoring and interpretability of this already established measure without sacrificing its reliability, validity, and responsiveness to change over time. Unlike new measures created with this method, working with an existing health outcome measure posed some additional challenges. We

Conclusion

The application of Rasch analysis is by no means a pure science. Because decisions must be made by using a balance between statistical and clinical reasoning, there is no single correct approach. It is therefore important to make decisions with care, considering a number of possible options before deciding on an appropriate course of action. A review of the procedures after their completion can further ensure the suitability of the decisions. Validation of the assumptions of the Rasch model,

Acknowledgements

We thank the project coordinator, Barbara Galuppi, the assessing therapists, and all the children and families who participated in the Ontario Motor Growth Study.

References (24)

  • D Russell et al.

    Improved scaling of the Gross Motor Function Measure for children with cerebral palsyevidence of reliability and validity

    Phys Ther

    (2000)
  • R.J Palisano et al.

    Validation of a model of gross motor function for children with cerebral palsy

    Phys Ther

    (2000)
  • Cited by (150)

    • Development of the Computerized Adaptive Test of Motor Development (MD-CAT) Adopting Multidimensional Rasch Analysis

      2021, Archives of Physical Medicine and Rehabilitation
      Citation Excerpt :

      The development of the MD-CAT makes an additional contribution to clinical and research purposes. That is, the MD-CAT also inherits the advantages of Rasch analysis, such as the interval properties of rating scales and the children's motor skills and item difficulties being comparable.24,25 The estimated abilities are thus useful for monitoring children's progress and comparing the results with others.

    View all citing articles on Scopus

    Supported in part by the National Institutes of Health (grant no. R01-HD 34947).

    1

    A commercial party with a direct financial interest in the results of the research supporting this article has conferred or will confer a financial benefit upon the authors or one or more of the authors. The authors will receive royalties from the sale of a book published by Mac Keith Press that describes the development of the GMFM-66.

    View full text