Far East Journal of Theoretical Statistics
Volume 13, Issue 1, Pages 47 - 65
(May 2004)
|
|
IMPACT OF MULTIPLE ENDPOINTS ON TYPE I ERROR RATE AND POWER OF TEST STATISTIC IN NON-SUPERIORITY CLINICAL TRIALS*
Abdul J. Sankoh (U. S. A.) and Mohammad F. Huque (U. S. A.)
|
Abstract: The efficacy of a new drug may be demonstrated by showing it is clinically equivalent or non-inferior to a standard active drug. To arrive at such a non-superiority efficacy conclusion, one must (ideally) first establish that both the standard and test drugs have efficacy profiles that are superior to placebo or no treatment in the prospective clinical trial. The statistical analysis approach for the demonstration of clinical equivalence or non-inferiority involves the construction of a confidence interval that rules out any possibility of the test drug being less efficacious than the standard drug by more than a pre-defined clinically irrelevant amount called an inferiority margin. In this paper we review statistical inference and methodological concepts for clinical equivalence and non-inferiority clinical trials. Additionally, we also show, at least for two correlated endpoints, multiplicity due to multiple endpoints could negatively impact the power of the test statistics. |
Keywords and phrases: clinical equivalence, multiple endpoints, non-inferiority, power, sample size, type I error rate. |
|
Number of Downloads: 257 | Number of Views: 792 |
|