Abstract:
|
When defining a scoring system for a continuous outcome during study design, a question of interest is, does the direction of scores (the larger or the smaller value, the better effect) matter in testing for superiority, non-inferiority(NI), or equivalence? In other words, will this change the power of the test? For continuous data, ratio of means and difference of means are often used as a measure to evaluate whether the test mean is superior to, NI, or equivalent to the reference mean by zero (for superiority) or by a certain margin (for NI or equivalence). Therefore, does the measure (difference or ratio) play a role in addressing this question? Our conclusions: 1) Direction of scores does not change the power in testing for superiority, NI or bioequivalence when difference of means is used as the measure; 2) However, direction of scores does change the power when ratio of means is used, especially for a NI and equivalence test. Caution should be used.
|