SIGNIFICANCE: Research in stroke rehabilitation supports a focus of providing task-based hand training as early after stroke as possible. However, we have little information about the relative difficulty of different hand movements and types of functional grasp. In addition, successful motor rehabilitation cannot focus on arm and hand function in isolation of other abilities. Hence, we need practical tools to assess multiple factors that have an impact on motor performance after stroke. The Motor Assessment Scale (MAS) is, overall, a well-respected tool; however, previous studies have found weaknesses in its items related to hand function. In this research, we provide a new way to assess hand function within the context of the MAS.

INNOVATION: Our research builds on the task-oriented paradigm for helping stroke survivors develop motor function that they can use for independent activity performance and participation. We have identified and provided a solution for one area of weakness in an assessment that otherwise provides therapists with an efficient, holistic, and function-based evaluation of motor function after stroke.

APPROACH: Our research objectives were as follows: (1) to develop and test a hierarchical scoring system for assessing hand movements and hand activities in stroke survivors; and (2) to determine a hierarchy of difficulty levels, in stroke survivors, for hand movements and common hand activities.

Two previous studies found serious limitations in the original behavioral hierarchies on the MAS for assessing hand function in stroke survivors. Using the paradigm that hand use requires both an ability to produce specific movements and skill in integrating these movements for task performance, we developed two behavioral hierarchies: for hand movements and for hand activities.

METHOD: In this methodological study, we tested 332 stroke survivors at eight occupational therapy departments for their ability to perform a series of behavioral criteria associated with hand function. Data collectors were occupational therapists who completed a training program and passed a rater qualification examination, which required them to achieve consistency with gold standard scores, determined by the researchers, when assigning pass or fail scores to videotaped stroke survivors who performed the tested items. Data were analyzed with the Dichotomous Rasch Model to assess dimensionality, test reliability, and test hierarchy of difficulty for the behavioral criteria.

RESULTS: We determined a set of 10 behavioral criteria for hand movements, which exhibited adequate measurement properties: person-separation of 2.11, strata of 3.15, acceptable infit values, and acceptable principal components analysis (PCA) results. The first dimension explained 72% of the variance. Scores ranged from 0.0 to 10.0 (M = 7.0, SD = 2.6). We determined a set of eight behavioral criteria for hand activities with the following measurement properties: person-separation of 2.31, strata of 3.41, acceptable infit values, and acceptable PCA results. The first dimension explained 82% of the variance. Scores ranged from 0.0 to 8.0 (M = 4.7, SD = 2.4).

CONCLUSION: Following a multistep process of scale development, refinement, and Rasch analysis, we have redesigned the hand items on the MAS. Our proposed hand movements and hand activities items each measure a unidimensional construct and can reliably measure individuals with different levels of ability. The hierarchical ordering of behaviors proposed to score each item is supported by Rasch analysis.