Abstract:
|
The situation in which the response (outcome) of interest is in an ordinal scale, having more than two categories, appears often when comparing two drug formulations. One such example is testing for bioequivalence of two formulations, using the pharmacokinetic parameter time of maximum concentration (Tmax), when this is measured in discrete units. Another example is non-inferiority testing of the degree of adhesion, in the case of transdermal products. In this case, lack of normality does not always allow for the use of the standard mixed-effects parametric approach. In both situations, the Generalized Odds Ratio (GOR) is a reasonable choice as it describes the difference between two ordered distributions in a simple manner and without the need of any parametric assumptions. Its potential for testing bioequivalence along with alternative methods will be explored. Examples of real data sets demonstrate the use of such methods in the cases of the crossover and paired-sample design cases. Advantages and potential pitfalls will be discussed.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.