Evidence-based sizing of non-inferiority trials using decision models

Area of Science:

• Clinical Trials
• Biostatistics
• Health Economics

Background:

• Non-inferiority trials require large sample sizes to demonstrate minimal inferiority of new interventions compared to standard treatments.
• Determining an appropriate non-inferiority margin (δ) is crucial but complex, involving trade-offs between benefits and burdens.

Purpose of the Study:

• To present a novel strategy for designing non-inferiority trials by developing a method to determine the appropriate non-inferiority margin (δ).
• To explicitly balance the benefits (e.g., lower recurrence, better survival) against the burdens (e.g., toxicity, morbidity) of interventions in trial arms.

Main Methods:

• A decision analytic approach was used to simulate trial outcomes.
• The non-inferiority margin (δ) was defined as the tipping point where quality-adjusted life-years (QALYs) are equal between arms, balancing event rates and morbidity.
• Simulations systematically varied outcome incidence under alternative interventions.

Main Results:

• Applied to the European Polyp Surveillance trial, the method suggested a 0.42% non-inferiority margin for 10-year colorectal cancer risk, differing from the original 0.50% design.
• The calculated non-inferiority margin decreased with an increased assumed burden of colonoscopies.

Conclusions:

• The proposed decision analytic method for determining non-inferiority margins is feasible in real-world clinical trials.
• This approach offers a more explicit and rigorous quantification of factors influencing non-inferiority margin and sample size determination.