样本量计算之如何确定非劣效性界值

In clinical research, the determination of a clinically meaningful difference, denoted by δ, is critical in clinical trials such as equivalence/noninferiority trials. In therapeutic equivalence trials, δ is known as the equivalence limit, while δ is referred to as the noninferiority margin in noninferiority trials. The noninferiority margin reflects the degree of inferiority of the test drug under investigation as compared to the standard therapy that the trials attempt to exclude.

A different choice of δ may affect the sample size calculation and may alter the conclusion of clinical results. Thus, the choice of δ is critical at the planning stage of a clinical study. In practice, there is no golden rule for the determination of δ in clinical trials. As indicated in the ICH E10 Draft Guideline, the noninferiority margin cannot be chosen to be greater than the smallest effect size that the active drug would be reliably expected to have compared with placebo in the setting of the planned trial, but may be smaller based on clinical judgment (ICH, 1999). The ICH E10 Guideline suggests that the noninferiority margin be identified based on past experience in placebo-controlled trials of adequate design under conditions similar to those planned for the new trial. In addition, the ICH E10 Guideline emphasizes that the determination of δ should be based on both statistical reasoning and clinical judgment, which should not only reflect uncertainties in the evidence on which the choice is based, but also be suitably conservative.

Most recently, FDA published a draft guidance on Non-Inferiority Clinical Trials, which recommends that two noninferiority margins, namely, M1 and M2, should be considered (FDA, 2010b). The 2010 FDA draft guidance indicated that M1 is based on (i) the treatment effect estimated from the historical experience with the active control drug, (ii) assessment of the likelihood that the current effect of the active control is similar to the past effect (the constancy assumption), and (iii) assessment of the quality of the noninferiority trial, particularly looking for defects that could reduce a difference between the active control and the new drug

In clinical trials, the choice of δ may depend upon absolute change, percent change, or effect size of the primary study endpoint. In practice, a standard effect size (i.e., effect size adjusted for standard deviation) between 0.25 and 0.5 is usually chosen as δ, if no prior knowledge regarding clinical performance of the test drug is available. This recommendation is made based on the fact that the standard effect size of clinical importance observed from most clinical trials is within the range of 0.25 and 0.5.