|
Abstract:
|
In the setting of binomial rates that differ under H0 and are equal under HA, we show that the marginal response rate and the noninferiority margin are a convenient "design pair" for noninferiority trials. Further, the minimum overall sample size, N, and optimal allocation ratio associated with fixed error rates depend on how the margin is parameterized. Since investigators commonly use the difference between test and control response rates (Delta) for design and the log-odds of response (logPsi) for analysis, we quantify effects on power of switching parameterizations of the margin. We also model the ratio N_Delta/N_Psi as a function of a wide range of design pairs; the regression estimates can be used at the design stage to identify pairs for which the margin's parameterization should not be interchanged between design and analysis. Last, we propose ways to quantify the marginal rate.
|
