EcoSta 2017: Start Registration
View Submission - EcoSta2017
A0741
Title: Non-inferiority and equivalence tests in a sequential multiple-assignment randomized trial (SMART) design Authors:  Bibhas Chakraborty - Duke-NUS Medical School, National University of Singapore (Singapore) [presenting]
Palash Ghosh - Duke-NUS Medical School (Singapore)
Abstract: In a sequential multiple-assignment randomized trial (SMART), a patient is given a sequence of treatments over the trial period. SMART designs are applicable in various branches of medical science like depression, behavioral science, cancer etc. The treatments/drugs used in a SMART could be from multiple pharmaceutical companies. Generally, pharmaceutical companies are reluctant to do a comparison of different marketed drugs from different companies by a traditional one/two-sided test, because a negative result may damage their business severely. In this context, a general perception is that a SMART design may not be a suitable choice for pharmaceutical companies. Contrary to that general perception, we argue that a SMART design could be profitable for pharmaceutical companies. It may be a win-win situation for all the companies involved in a SMART. We develop the non-inferiority and the equivalence test procedures to compare treatment-sequences in a SMART, to facilitate the uptake of SMART designs in the pharma-sponsored trials. We show that the trade-off between the efficacy and the quality of life (QOL) from a treatment-sequence can be addressed by a non-inferiority/equivalence test. Sample size and power issues have been discussed with supporting simulations. We show the applicability of the developed test procedures in the SMART context using the STAR*D data.