Assessing the responsiveness of functional scales to clinical change: An analogy to diagnostic test performance