Table 4.
Re-analysis of original trial data in trials showing no evidence of an effect on the final outcome.
| ICON3 | FOCUS | |||||||||
| one-sided significance | A vs B | A vs D | A vs E | A vs B | A vs D | A vs E | ||||
| level at stage 1 | δI | HR | decision | δI | HR | decision | ||||
| 0.5 | 1 | 1.054 | ✓ | 1 | 0.876 | 0.841 | 1.008 | ✗ | ✗ | ✓ |
| 0.4 | 0.964 | 0.965 | ✗ | 0.961 | 0.918 | 0.860 | 0.988 | ✗ | ✗ | ✓ |
| 0.3 | 0.936 | 0.825 | ✗ | 0.930 | 0.892 | 0.750 | 0.902 | ✗ | ✗ | ✗ |
| 0.2 | 0.911 | 0.806 | ✗ | 0.902 | 0.796 | 0.766 | 0.946 | ✗ | ✗ | ✓ |
| 0.1 | 0.885 | 0.837 | ✗ | 0.874 | 0.887 | 0.879 | 0.937 | ✓ | ✓ | ✓ |
δI gives the cut-off hazard ratio for the intermediate outcome at a given one-sided significance level, HR gives the observed hazard ratio at that point in time, ✓ denotes that the trial would have been correctly stopped at that stage, ✗ denotes that the trial would have been incorrectly carried on to the next stage