On Selection of Margin in Non-Inferiority Trials

In clinical trials, it is unethical to use a placebo control in treating patients with severe or life-threatening diseases such as cancer when approved and effective therapies (e.g., standard of care or active control agents) are available. Alternatively, an active control or non-inferiority trial is often considered. In practice, one of the key issues for a noninferiority trial is the determination of non-inferiority margin which has an impact on the power analysis for sample size calculation. In its 2010 draft guidance, the United States Food and Drug Administration (FDA) recommend a couple of margins for testing non-inferiority of a test treatment as compared to an active control agent or a standard of care treatment. In this article, several margins, which not only adjust for variability associated with the observed data but also take into consideration of the retention rate of the treatment effect, are proposed.