Table 3.
Consider non-inferiority trials designed with primary analysis to rule out the non-inferiority margin, δ, under the assumption that efficacy of the experimental (EXP) and active control (AC) vaccines is equal. While such trials properly are powered to rule out δ only when vaccine efficacy of the EXP vaccine truly is greater than or equal to that of the active control (AC), the trial is powered to rule out the non-inferiority margin, δo, when the vaccine efficacy of the experimental (EXP) is only 10% less than that on the active control (AC). Results are presented corresponding to 175 or 350 events in the placebo-controlled trial of the AC.
| Events in placebo-controlledtrial of AC | Vaccineefficacy of AC | NI trial #events | H0: true HR = margin |
HA: |
Power under HA |
|||
|---|---|---|---|---|---|---|---|---|
| δ | δo | EXP vaccineefficacy | HR | To ruleout δ | To ruleout δo | |||
| 175 | 95% | 34 | 3.421 | 9.790 | 95% | 1 | 90% | >99% |
| 90% | 2 | 28% | 90% | |||||
| 80% | 4 | <1% | 40% | |||||
| 90% | 54 | 2.561 | 5.486 | 90% | 1 | 90% | >99% | |
| 80% | 2 | 13% | 85% | |||||
| 70% | 3 | <1% | 41% | |||||
| 80% | 112 | 1.875 | 2.940 | 90% | 0.5 | >99% | >99% | |
| 80% | 1 | 90% | >99% | |||||
| 70% | 1.5 | 20% | 90% | |||||
| 60% | 2 | <1% | 45% | |||||
| 350 | 95% | 31 | 3.700 | 11.454 | 95% | 1 | 90% | >99% |
| 90% | 2 | 22% | 90% | |||||
| 80% | 4 | <1% | 46% | |||||
| 90% | 48 | 2.724 | 6.207 | 90% | 1 | 90% | >99% | |
| 80% | 2 | 16% | 87% | |||||
| 70% | 3 | <1% | 48% | |||||
| 80% | 97 | 1.974 | 3.260 | 90% | 0.5 | >99% | >99% | |
| 80% | 1 | 90% | >99% | |||||
| 70% | 1.5 | 25% | 93% | |||||
| 60% | 2 | <1% | 57% | |||||
HR: hazard ratio; NI: non-inferiority.