violated. Therefore, a Greenhouse-Geisser correction
factor was applied to all P values. The interaction effect
(group time) and main effect for group were not
significant. The power of this study to detect a clinically
meaningful effect size of 8 degrees for the interaction
effect was .81. There was a main effect for time
(P.0001). The results of the mixed-model ANOVA
using an intention-to-treat analysis did not differ from
the results of the on-protocol analysis.
Post hoc testing for the main effect of time was