Table 2.
Mean number of subjects required in SSD/Modified SSD compared to Bayesian selection strategy in the CLL trial example
|
Scenario |
True CR (pA,pB) |
Average number of subjects |
|||
|---|---|---|---|---|---|
| |
|
SSD/Modified SSD |
Bayesian selection strategy |
||
| Arm A | Arm B | Arm A | Arm B | ||
| 1 |
(0.01,0.01) |
16.0 |
16.0 |
11.1 |
11.1 |
| 2 |
(0.1,0.1) |
25.6 |
25.6 |
21.2 |
21.2 |
| 3 |
(0.2,0.2) |
28.3 |
28.3 |
26.7 |
26.7 |
| 4 |
(0.3,0.3) |
28.9 |
28.9 |
28.4 |
28.4 |
| 5 |
(0.01,0.03) |
16.0 |
19.2 |
11.1 |
13.5 |
| 6 |
(0.01,0.2) |
16.0 |
28.3 |
11.1 |
26.7 |
| 7 |
(0.20,0.35) |
28.3 |
29.0 |
26.7 |
28.8 |
| 8 | (0.2,0.4) | 28.3 | 29.0 | 26.7 | 28.9 |
The table displays the mean number of subjects required for the Screened Selection Design (SSD)/Modified Screened Selection Design (Modified SSD) and the Bayesian Selection Strategy, obtained from the same simulation study as presented in Table 1. Number of subjects required in SWE is always 29 per arm, whereas the other two approaches, allowing for early stopping due to futility, have a lower mean number of subjects when the arms are not active.