Testing Superiority after the Confirmation of Noninferiority for Three-arm Clinical Trials with Multiple Experimental Treatments

  • 鐘 俊江

Student thesis: Doctoral Thesis

Abstract

The objective of a non-inferiority (NI) trial is to assert the efficacy of an experimental (new) treatment compared with a reference (standard) treatment by showing that the experimental treatment retains a substantial proportion of the efficacy of the reference treatment The marginal loss of efficacy of adopting the experimental treatment has to be justified by its other benefits such as the alleviation of side effects reduction of costs and introduction of less complicated regimens In NI trials it is common to have multiple experimental treatments when the efficacy of different doses of a new drug or different combinations of several new drugs are being investigated Statistical methods have been developed to test NI of multiple experimental treatments in three-arm NI trials consisting of experimental treatments a reference treatment and a placebo However there is a shortage of procedures that examine both NI and superiority of the new treatments as compared with the reference treatment The objective of this thesis is to develop testing procedures which examine the superiority of an experimental treatment after it has been shown to be NI to the reference treatment Hence at the end of the entire test procedure we will be able to classify all the experimental treatments into three groups; namely inferior non-inferior and superior to the standard treatment As compared to previous methods the advantage of the proposed test procedures is the additional ability to identify superior treatments while retaining an NI testing power comparable to that of existing testing procedures Both single-step and stepwise procedures are derived and the computational details regarding the required critical values are provided Further a comparative study is conducted to investigate the merits of each procedure To illustrate the proposed procedures three clinical trial examples are discussed Finally the computation of power and the determination of sample size before the onset of an NI study is given
Date of Award2017 Jan 9
Original languageEnglish
SupervisorMiin-Jye Wen (Supervisor)

Cite this

'