Michael W. O’Dell, Grace Kim , Lisa Rivera , Robert Fieo, Paul Christos, Caitlin Polistena , Kerri Fitzgerald, Delia Gorga
Department of Rehabilitation Medicine, Weill Cornell Medical Center, 10065 New York, USA. E-mail: mio2005@med.cornell.edu
DOI: 10.2340/16501977-1138

Objective: To further examine the psychometric properties of a 9-item version of the Arm Motor Ability Test (AMAT-9) in persons with stroke.
Subjects: Thirty-two community-dwelling persons > 6 months post-stroke undergoing robotics treatment (mean age = 56. 0 years, time post-stroke = 4. 1 years, National Institutes of Health Stroke Scale score = 4. 1, and AMAT-9 score = 1. 22).
Methods: Construct validity (including Rasch analyses) used baseline data prior to treatment (n = 32). Standardized response mean was calculated for subjects completing the protocol (n = 29). The Wolf Motor Function Test (WMFT), Fugl-Meyer Assessment (FMA), Action Research Arm Test (ARAT), and Stroke Impact Scale (SIS) were also administered.
Results: Spearman-rank correlation coefficients between AMAT-9 and the WMFT, FMA, and ARAT were strong (0. 78–0. 79, all p < 0. 001). The correlation between the AMAT-9 and SIS Hand Function sub-score was stronger than that between the AMAT-9 and the Communication sub-score (0. 40, p = 0. 025 and –0. 16, p = 0. 39, respectively). Rasch analyses provided evidence for an appropriate hierarchical structure of item difficulties, unidimensionality, and good reliability. The AMAT demonstrated a comparable standardized response mean of 0. 98.
Conclusions: The AMAT-9 is valid and responsive among subjects scoring in the lower range of the scale. It has the advantage of assessing function and by eliminating the standing item from the previous iteration, it may be more easily used with severely impaired patients.