Superiority trials are a common design in clinical trials and are well accepted as proof of efficacy for investiagational drugs. But more and more, non-inferiority trials are being used. CROS NT discusses the main reasons why these trials are used and how to design and analyze these trials according to our top biostatisticians.
There are two main aspects that influence the choice of non-inferiority design when planning clinical trials:
- The first is related to ethical considerations. If there is an existing treatment which is life-saving or prevents serious harm to the patient for a given disease, it is not ethical to use placebo treatment in a comparative trial; consequently the comparison has to be done to an active control.
- The second aspect is that in many therapeutic areas the increase in efficacy of new drugs is minor but very often these drugs have improved safety profiles and cost-benefit ratios.
These situations lead to trials where the primary objective is not to show superiority of a new drug compared to existing therapies, but to show that the new drug is not “worse” than the existing therapy.
If we define “not worse” as “better than control minus a specific margin”, then we end up in a non-inferiority trial. Consequently, the alternative hypothesis of a non-inferiority trial is stated as: “H1:C-T<M, in contrast to the hypothesis of a superiority trial: H1:C-T<0, where C is the response of the control treatment, and T is the response of the test treatment and M is the non-inferiority margin.
The effect of the treatment can be established by showing that the lower bound of the 1-sided 97.5% confidence interval (equivalent to a 2-sided 95% confidence interval) of C-T is either above zero or above M. Establishing non-inferiority margin M is one of the challenges of non-inferiority trials, the other is to establish Assay Sensitivity (AS). The margin M should be chosen as the largest clinically acceptable loss in efficacy and may depend on the medical indication of the drug. In most trials this margin is chosen to be of fixed size and is determined in advance. Antibiotic treatment success rates of 80% or more are common and thus acceptable loss in efficacy is usually 10-15%.
There are a few questions left to answer: How can we determine the effect of the control treatment? What happens if the control treatment is not effective at all?
At this point, the Assay Sensitivity concept must be applied. Assay Sensitivity means that if the study included a placebo, a control drug-placebo difference of at least M would have been demonstrated. To determine whether a trial will include the Assay Sensitivity concept, we must consider three factors:
- Evidence from historical trials of sensitivity to drug effects;
- The similarity of the new non-inferiority trial to the historical trials (the constancy assumption) concerning the study population, the treatment regimen used, etc.
- The quality of the new trial (ruling out defects that would tend to minimize differences between treatments)
Given that these assumptions are not defined, a placebo treatment arm should be included in a non-inferiority trial.
Lastly, there is the issue of how to analyze this type of trial. We already mentioned that statistical approach of confidence intervals. It is widely accepted that 1-sided 97.5% confidence interval of the treatment difference is used to accept the alternative hypothesis. In order to rule out defects of a trial that would tend to minimize the differences between treatments, it is generally accepted to use the pre-protocol analysis set, instead of the intend-to-treat analysis set in a superiority trial.
In conclusion, non-inferiority trials are a very important tool, in addition to superiority trials, but have a greater degree of freedom in terms of design and analysis. Therefore, careful consideration and planning is recommended for a non-inferiority trial and reasonable arguments must be given if a trial deviates from accepted standards.