ABSTRACT
Probability of Success. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207 8.8 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
This chapter demonstrates the gsDesign R package using two examples of trials planned with a single futility analysis. If you wish to replicate the calculations, you will need to install gsDesign 2.8-3 or later, which is available from the CRAN website https://www.cran.r-project.org. While considering
a single interim analysis may seem overly restrictive, this can be the most important analysis in a group sequential design in terms of limiting the patient volunteers required as well as research expenses for trials which have little chance of helping patients. In addition, designs with more analyses are simple variations of those shown here. The chapter is intended to both demonstrate the essential R coding and how to evaluate design bounds. The chapter was written using a combination of LATEX and the knitr R package (Xie 2014); this allowed combining R commands and text markup in the source document for the chapter. By applying the knitr package to the source, R commands were run and a pure LATEX file was produced for typesetting. Note that files with the extension .rnw in the package distribution are source files that can be used with LATEX and knitr to produce a typeset document with results from embedded R code.
