This RT topic will focus on issues related to the specification of intent-to-treat versus per-protocol analysis populations, with particular emphasis on non-inferiority studies. Additional design considerations such as blinding and adjustments for multiplicity in this context will also be discussed, as well as variations in approaches favored by FDA vs other Health Authorities.

1. What is the preferred approach to the specification of the primary analysis population(s) for non-inferiority studies? 2. Discuss the practical (eg, definitions for population exclusions) and statistical (eg, multiplicity adjustments) considerations for the chosen analysis population(s)?