Initial psychometric evaluation of the Arm Activity Measure (ArmA): a measure of activity in the hemiparetic arm

Objective: To evaluate the psychometric properties of the Arm Activity Measure (ArmA), a patient-reported measure of active and passive function in the paretic upper limb. Design: Psychometric evaluation study. Setting: Two specialist rehabilitation and spasticity management services. Method: Patients (n = 92) with upper limb paresis were recruited from two specialist neurorehabilitation centres. Mean age 44.5 (SD 16.7). Diagnostic distribution: stroke 48 (52%); other brain injury 28 (31%); or other neurological condition 16 (17%). Evaluation of convergent and divergent validity; unidimensionality, scaling, reliability (internal consistency and test-retest); responsiveness to change and feasibility of the ArmA were undertaken. Results: Expected convergent and divergent relationships were seen with the Leeds Adult Spasticity Impact Scale and the Disabilities of Arm Shoulder and Hand (DASH) (rho 0.5-0.63). Principal components analysis confirmed that active and passive function formed two separate constructs in each sub-scale. Mokken analysis corroborated the findings of the principal components analysis and demonstrated scaling using the monotone homogeneity model (Item H>0.5 for all items). Cronbach's alpha was 0.85 and 0.96, respectively, for the passive and active function subscales. Item level test-retest agreement ranged from 92-97.5% (quadratic-weighted Kappa 0.71-0.94). In the subgroup treated for spasticity with botulinum toxin (n = 58), the ArmA passive function scale identified a significant difference between responder and non-responder groups (Mann Whitney U = 0.85, p < 0.01). Respondents reported the ArmA to be relevant (77%), easy to use (90%) and timely to complete (83% under 10 minutes). Conclusion: The ArmA is a valid and reliable tool feasible for use in the evaluation of upper limb function in the context of treatment for spasticity.