It is interesting that baseline SF 36 physical function may help predict response. This doesn't seem to be the case for the Chalder Fatigue Scale.
I didn't notice that before you pointed it out. It does seem to suggest that those with more severe illness are unlikely to improve - contrary to the commonly held myth (among doctors) of "most patients improve over time".
Of course, because the Chalder scale was never designed to measure absolute fatigue, but rather, short term perceptions of changes in fatigue. (and it has never been tested for patient acceptability, understandability etc - they never bothered to find out whether patients believe that it accurately reflects their symptoms or not)