Table 3.

Operating characteristics for planned adaptive trial design variants

	Mean subjects	Power	Early success	Early futility	% Randomized to best arm	Mean trial duration (months)
Design 1: early stopping only
Null	39487	0.03	0.023	0.875	1.00	60
Alternative	39513	0.93	0.895	0.005	0.63	56
One Works	39637	0.70	0.657	0.112	0.21	66
Better & Best	39649	0.73	0.669	0.082	0.21	66
Worse	35286	0.00	0.000	1.000	0.37	29
Design 2: probability-weighted adaptive randomization; no arm dropping
Null	39504	0.02	0.020	0.905	1.00	59
Alternative	39400	0.93	0.900	0.016	0.66	56
One Works	39537	0.75	0.708	0.091	0.32	63
Better & Best	39634	0.75	0.707	0.091	0.30	64
Worse	34755.3	0.00	0.000	1.000	0.34	28
Design 3: information-weighted adaptive randomization; no arm dropping
Null	39624	0.02	0.018	0.896	1.00	60
Alternative	39443	0.93	0.901	0.012	0.66	56
One Works	39595	0.75	0.708	0.093	0.28	66
Better & Best	39711	0.78	0.742	0.075	0.27	65
Worse	35118	0.00	0.000	1.000	0.34	29
Design 4: probability-weighted adaptive randomization; 5% arm dropping
Null	39557	0.02	0.018	0.897	1.00	60
Alternative	39395	0.91	0.889	0.01	0.66	57
One Works	39658	0.79	0.728	0.094	0.34	64
Better & Best	39641	0.76	0.707	0.09	0.30	65
Worse	34683	0.00	0	1	0.34	28
Design 5: information-weighted adaptive randomization; 10% arm dropping
Null	39377	0.03	0.022	0.905	1.00	59
Alternative	39503	0.92	0.885	0.013	0.66	56
One Works	39622	0.74	0.704	0.088	0.29	66
Better & Best	39726	0.74	0.7	0.07	0.27	66
Worse	35142	0.00	0	1	0.34	29
Design 6: no adaptive randomization; arm dropping 20% vs control
Null	38568	0.02	0.022	0.736	1.00	56
Alternative	39369	0.92	0.901	0.012	0.79	55
One Works	39256	0.69	0.645	0.104	0.20	63
Better & Best	39404	0.71	0.656	0.103	0.22	64
Worse	31700	0.00	0.000	0.190	0.00	26
Design 7: no adaptive randomization; predictive probability-based arm dropping
Null	38630	0.022	0.020	0.88	1.00	57
Alternative	34337	0.91	0.87	0.012	0.67	57
One Works	37303	0.69	0.64	0.11	0.25	67
Better & Best	37032	0.70	0.65	0.08	0.24	68
Worse	34323	0.00	0.00	>0.99	1.00	27