Abstract:
|
In clinical trial, a test treatment is often compared with a reference treatment for non-inferiority or equivalence. The comparison is often carried out by comparing the difference or ratio of treatment means against a constant non-inferiority margin. For continuous endpoints, the data is often assumed to be normally distributed. In a superiority test setting, when the normality assumption is violated, we may use a non-parametric method or data transformation strategy to carry out some proper tests. However, with the non-inferiority test setting, we may encounter difficulties when carrying out the proper test. We will illustrate the problems with some examples.
|