Premium
A framework for two‐stage adaptive procedures to simultaneously test non‐inferiority and superiority
Author(s) -
Koyama Tatsuki,
Sampson Allan R.,
Gleser Leon J.
Publication year - 2005
Publication title -
statistics in medicine
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.996
H-Index - 183
eISSN - 1097-0258
pISSN - 0277-6715
DOI - 10.1002/sim.2118
Subject(s) - computer science , test (biology) , statistics , stage (stratigraphy) , econometrics , mathematics , biology , paleontology
Abstract In clinical trials it is often desirable to test for non‐inferiority and for superiority simultaneously. For such a situation a two‐stage adaptive procedure may be advantageous to a conventional single‐stage procedure because a two‐stage adaptive procedure allows the design of stage II, including the main study objective and sample size, to depend on the outcome of stage I. We propose a framework for designing two‐stage adaptive procedures with a possible switch of the primary study objectives at the end of stage I between non‐inferiority and superiority. The framework permits control of the type I error rate and specification of the unconditional powers and maximum sample size for each of non‐inferiority and superiority objectives. The actions at the end of stage I are predetermined as functions of the stage I observations, thus making specification of the unconditional powers possible. Based on the results at the end of stage I, the primary objective for stage II is chosen, and sample sizes and critical values for stage II are determined. Copyright © 2005 John Wiley & Sons, Ltd.