| Abstract | A web-based implementation that generates Simon's admissible two-stage designs (Simon, 1989; Jung et al., 2004) for Phase II single arm clinical trials with relaxed probability of stopping for futility (Ivanova and Deal, 2013).
The primary outcome is the probability of treatment response (TR), where treatment response is defined as complete response (CR) or partial response (PR), TR = CR + PR. The null hypothesis is that the TR rate is p0T and the alternative is that the TR rate is pAT. However, the trial is stopped for futility only if the number of patients with disease control (DC) is less than or equal to stage 1 futility boundary, where disease control is defined as DC = TR + SD, with SD denoting stable disease. The trial is also stopped for futility if the number of TRs in stage 1 is such that it is impossible to reject H0 after stage 2, that is, if the number of TRs is less than r2 - (n - n1).
The user needs to input lower and upper bounds for the probability of stable disease, pSL and pSU , 0 ≤ pSL ≤ pSU ≤ 1 - pAT. We assume that pS ~ Uniform(pSL, pSU).
The optimal design minimizes the expected sample size under H0, that is, given p0T and over the distribution of p0S.
The design guarantees the type I error rate not exceeding α for given p0T and pSU, and required power for given pAT and pSL.
When pSL=pSU=0 the design is equivalent to the Simon's two-stage design.
|
