Abstract
Crossover is increasingly favored in trials of cancer therapies; even those that seek to establish the basic efficacy of novel drugs. Crossover is done in part for trial recruitment, but also out of a sense of doing the right thing-offering the investigational agent to more patients. In this paper, we argue that this ethical inclination-that crossover is a preferred trial choice-is misguided. In seeking to sate the desires of participants, we might undermine a trial's ability to answer a meaningful clinical question. When a trial is incapable of answering a question, it becomes unethical. Using a crossover strategy in oncology clinical trials can make trials less ethical, not more.