Psychometric properties of the Dependence Scale in large randomized clinical trials of patients with mild and moderate Alzheimer's disease.

Abstract

The need for assistance from others is a hallmark concern in Alzheimer's disease (AD). The psychometric properties of the Dependence Scale (DS) for measuring treatment benefit were investigated in large randomized clinical trials of patients with mild to moderate AD. Reliability, validity, and responsiveness of the DS were examined. Path models appraised relationships and distinctiveness of key AD measures. The responder definition was empirically derived. Generally acceptable reliability (α ≥ .65), significant (P < .001) known-groups tests, and moderate to strong correlations (r ≥ .31) confirmed the DS psychometric properties. Path models supported relationships and distinctiveness of key AD measures. A DS change of ≤ 1 point for patients with limited home care and ≤ 2 points for patients with assisted living care best described stability of the level of dependence on caregivers. The DS is a psychometrically robust measure in mild to moderate AD. The empirically derived responder definition aids in the interpretation of DS change.