Sample sizes for active-comparator non-inferiority study designs. The hypothesis is that the new modulator is no more than a ‘small amount’ less efficacious than an active comparator, quantified by a non-inferiority (NI) margin
| Previously observed FEV1 treatment effect*: comparator–placebo, (95% CI)* | FEV1 SD of the change in FEV1 % predicted* | % of Lowest treatment effect to preserve | NI margin† | Sample size per group‡ | |
|---|---|---|---|---|---|
| Scenario A: non-inferiority study vs an active comparator which has efficacy comparable with lumacaftor–ivacaftor | |||||
| 3.0% (1.6 to 4.4) | 7.3 | 75% | 0.75×1.6=1.2% | 778 | |
| Scenario B: non-inferiority study vs an active comparator which has efficacy comparable with ivacaftor | |||||
| 10.6% (8.6 to 12.6) | 7.0 | 50% | 0.50×8.6=4.3% | 56 | |
*Treatment effects and SDs approximated from the 6-month changes in FEV1 observed in previous trials.16 ,21
†The NI margin is derived based on preserving a percentage of the lowest possible treatment effect observed in the placebo-controlled trial, as captured by the lower bound of the 95% CI. NI margins must be negotiated with regulators and ensure that a clinically meaningful effect size will be maintained.
‡Assuming there is truly no difference between the new modulator and the active comparator, sample size estimates are generated with 90% power to ensure that the lower limit of a one-sided 97.5% CI will be above the NI margin.