ABSTRACT
The frequent use of the standard two-arm randomized clinical trial is due in part to its relative simplicity of design and interpretation. Conclusions are straightforward: either the two arms are shown to be different or they are not. Complexities arise with more than two arms; with four arms there are 6 possible pairwise comparisons, 19 ways of pooling and comparing two groups, 24 ways of ordering the arms, plus the global test of equality of all four arms. Some subset of these comparisons must be identified as of interest; each comparison has power, level, and magnitude considerations; the problems of multiple testing must be addressed; and conclusions can be difficult, particularly if the comparisons specified to be of interest turn out to be the wrong ones.
