Interpreting Results of a Non-Inferiority Clinical Trial

 

A non-inferiority clinical trial is done when:

(1) there is an effective and accepted therapy for a disorder

(2) a placebo control is considered unethical or otherwise unacceptable

(3) the new drug is compared against a control group that is being treated with the accepted therapy (active control)

 

In a non-inferiority clinical trial a new drug is compared against an "active control" (group being treated with an existing, accepted therapy for the condition). The study is intended to only show that the new drug is not inferior to the existing therapy. These trials are not designed to show that it is better.

 

Parameters:

(1) mean outcome measure with the new drug being evaluated

(2) mean outcome measure with the existing therapy

(3) outcome measure expected if the existing therapy had no effect (M1)

(4) outcome measure that is considered the level that is clinically acceptable (M2); above this the outcome is not acceptable. M2 is less than M1.

 

difference between the new drug and the active control =

= (outcome for the active control) - (outcome measure for the new drug)

 

The 95% confidence interval (CI) is calculated for the difference between the control and the new drug.

Difference Between the Control and the Drug	Upper Limit of the 95% Confidence Interval	Interpretation
<= 0	<= M2	new drug effective (non-inferiority demonstrated)
<= 0	> M2 and <= M1	uncertain (difference suggests non-inferiority but high 95% CI indicates an unacceptable loss in clinical effect)
<= 0	> M1	unlikely scenario with difference and CI discrepant
> 0 and <= M2	<= M2	existing drug favored with new drug providing acceptable clinical effect
> 0 and <= M2	> M2 and <= M1	existing drug favored with new drug showing an unacceptable loss in the desired clinical effect
> 0 and <= M2	> M1	existing drug favored with no evidence that new drug effective
> M2	NA	performance of new drug unacceptable