Power function arguments in support of an alternative approach for analyzing management trials.

Power function arguments are used in support of an alternative approach for analyzing management trials. Central to the arguments is the recognition that, based on interpreting the smallest clinically important difference as a point of indifference, management trials require different power functions than explanatory trials. This interpretation of the smallest clinically important difference implies a specific ideal power function, and the alternative approach, by testing a nonzero null hypothesis at the appropriate level, is designed to achieve a power function with intuitive appeal and some optimal properties with respect to this ideal function. The alternative approach, by moving away from the traditional null hypothesis, better reflects clinical considerations, and eliminates the conflict between statistical and clinical significance. In addition, with adequate sample size, a decision in favor of the superior treatment will be made with a predetermined high level of confidence, if the therapies differ significantly.