[How to deal with equivocal subjects in statistical analysis of randomized controlled trials]

In randomized controlled trials, it is likely that several subjects with atypical features are registered. In this paper, it was discussed from a statistical point of view how properly the subjects with the following atypical features could be dealt with; 1) subjects who were found to be not eligible for the trial after randomization, 2) subjects who did not receive the assigned regimen by the treatment protocol, 3) subjects whose outcome was ambiguous as the designated endpoint, and 4) subjects who were ambiguous whether to be regarded as censored cases. It was also emphasized that informations on equivocal subjects should be masked in the committee which is responsible for the statistical judgement. Plausible examples were added to assist the readers to understand the rules discussed here.