91
Case Study:
Futility Boundary
Sponsor desired greater efficiency when treatment effect is low
•Explored asymmetric designs with a range of boundary shape functions from unified family
–P= 0.5 (Pocock), 0.8, 0.9, 1.0 (O’Brien-Fleming)
•Compare unconditional power and ASN curves
–Rationale: Are we losing power by stopping early?
•If not, then we are not making bad futility decisions on average