When inferiority meets non-inferiority: implications for interim analyses.

Area of Science:

• Clinical Trials Methodology
• Statistical Inference in Medicine
• Biostatistics

Background:

• Non-inferiority trials assess if a new treatment is not unacceptably less effective than a control.
• Unusual trial outcomes can occur where a treatment is simultaneously inferior and non-inferior.
• This scenario has implications for interim analysis and trial stopping rules.

Purpose of the Study:

• To explore issues in non-inferiority trials with potential simultaneous inferior/non-inferior results.
• To describe methods for deciding trial cessation at interim analyses due to futility.

Main Methods:

• Conditional power is used to assess the probability of demonstrating non-inferiority at trial's end.
• Simulations are employed to calculate conditional power for non-inferiority trials.
• The Bullous Pemphigoid Steroids and Tetracyclines Study (BLISTER) trial serves as a relevant case example.

Main Results:

• Stopping guidelines based on conditional power are more appropriate for non-inferiority trials than traditional superiority trial methods.
• Simulations provide a flexible and straightforward approach for calculating conditional power.

Conclusions:

• Conditional power is a suitable tool for establishing futility stopping guidelines in non-inferiority trials.
• Careful consideration of assumptions regarding future treatment efficacy is crucial.
• Stopping guidelines must be chosen to prevent inflation of the Type II error rate.