|
Abstract:
|
Paired and matched data have often been used in clinical trials and epidemiological safety studies. For example, crossover and William's square designs are often used in generic drug evaluation, QTc trial, pain relieve drug evaluation and in drug abuse potential trials. Two treatments may be paired on matched to study the treatment efficacy or to perform a case-control safety study. Either the response is continuous or binary variable, the correlation existed between the matched responses is not negligible. Conventionally, the sample size is determined inefficiently using the independent information of the two groups. On the other hand, when the trial is of non-inferiority or equivalence in nature, sample size is determined by a alternative hypothesis of a fixed treatment difference between test and the active control, without prior information of the true difference.
|
