Enrichment Bayesian design for randomized clinical trials using categorical biomarkers and a binary outcome

Table 4 Comparisons of decisions at the end of the trial according to the rules when K=2

			Subset A		Subset B
Scenarios	Interaction method	Go with EP ^∗	go with SP (efficacy/interaction) ^∗	n_A	go with SP (efficacy/interaction) ^∗	n_B	False positive
Scenario 1	Millen	0.9622	0.0177	399.1692	0.0201	400.8308	0.0378
	Gail and Simon	0.9454	0.0282	400.2944	0.0264	399.7056	0.0546
Scenario 2	Millen	0.7259	0.2707	447.1501	0.0034	352.8499	0.0034
	Gail and Simon	0.7928	0.1934	446.8688	0.0138	353.1312	0.0138
Scenario 3	Millen	0.2025	0.7970	558.2047	0.0000	241.7953	0.0000
	Gail and Simon	0.5834	0.4080	507.3405	0.0086	292.6595	0.0086
Scenario 4	Millen	0.0067	0.9933	655.0378	0.0000	144.9622	0.0000
	Gail and Simon	0.2894	0.7106	601.0114	0.0000	198.9886	0.0000

The total sample size is set at n= 800, with π=0.5 and q_B=q_A=0.5.
^*EP: entire population;
^*SP: subpopulation (efficacy/interaction) due to the detection of interaction with efficacy in subset k;
n_Aand n_Bare the mean sample size in each subset at the end of the study

ISSN: 1471-2288