Table 2.
Consider non-inferiority trials designed with primary analysis to rule out the non-inferiority margin, δo, assuming that the experimental vaccine (EXP) would have 60% vaccine efficacy versus placebo.
| From placebo-controlled RCT |
For the non-inferiority trial |
||||||||
|---|---|---|---|---|---|---|---|---|---|
| Number events | Vaccine efficacy | Hazard ratio | HR upper limit 95% CI | NI margin δ | NI margin δo | HA: EXP/PLAhazard ratio | NI trial# events* | Max EXP/AC hazard ratio ruling out | |
| δ ** | δ o ** | ||||||||
| 150 | 70% | 0.30 | 0.4272 | 1.530 | 1.958 | 0.40 | 304 | 1.217 | 1.552 |
| 350 | 70% | 0.30 | 0.3781 | 1.626 | 2.213 | 0.40 | 180 | 1.207 | 1.630 |
| 150 | 60% | 0.40 | 0.5620 | 1.334 | 1.489 | 0.40 | 271 | 1.050 | 1.170 |
| 350 | 60% | 0.40 | 0.4997 | 1.415 | 1.674 | 0.40 | 164 | 1.039 | 1.226 |
| 150 | 50% | 0.50 | 0.6972 | 1.198 | 1.200 | 0.40 | 259 | 0.938 | 0.939 |
| 350 | 50% | 0.50 | 0.6216 | 1.268 | 1.346 | 0.40 | 158 | 0.926 | 0.983 |
RCT: randomized controlled trial; CI: confidence interval; HR: hazard ratio; NI: non-inferiority; PLA: placebo.
*
Calculated assuming 90% power when the experimental vaccine (EXP) has 60% vaccine efficacy, using a statistic having 2.5% false positive error when δo is the true EXP/AC hazard ratio.
**
This represents the highest estimated experimental (EXP) to active control (AC) estimated hazard ratio that yields a positive result in the non-inferiority trial.