A plea for improved use of the tools available for the evaluation of anticancer drugs

Abstract

Superb methods of anticancer drug evaluation exist, but they are still being used inadequately. In comparative trials, it is essential that the statistical design and analysis eliminate the possibility of allocation (group membership) bias, assessment (measurement) bias and chance before it is possible to conclude that a real difference exists between the two treatments under test. Oncologists should examine these aspects critically before accepting that a trial shows a genuine treatment effect.