Bayesian dose selection design for a binary outcome using restricted response adaptive randomization

Table 1 Operating characteristics for different treatment effect assumptions: (a) all four active doses are ineffective; (b) only the maximum dose duration is effective; (c) linear dose-response trend; and (d) all four active doses are equally effective

	Tx effect	N	P(Dose)	Uncond. power	Cond. power
a	29.0	48	24.9	41.0	41.2
b	18.4	85	78.7	76.1	81.6
c	18.8	64	50.3	84.3	87.5
d	19.1	47	24.2	93.8	94.0

The table presents the median observed effect size for the optimal dose arm (Tx effect), the average number of subjects randomized to that arm (N) and the unconditional and conditional power defined respectively as the percentage of simulations where P(θ _j−θ ₀>0)≥0.8 conditional and unconditional on whether j was the true optimal dose

ISSN: 1745-6215