ABSTRACT
In this chapter we discuss the continual reassessment method (CRM) for conducting phase I trials.1 In particular, we emphasize the use of graphical presentations for the application of a Bayesian approach.2 For many years, phase I trials of anticancer drugs have not involved statisticians because the conventional design with threepatient cohorts is still widely used.3,4 A first step is for statisticians to explain the differences between conventional and CRM designs to clinicians, emphasizing the potential benefits of the CRM. This first challenge is difficult because the conventional design is very simple to understand, whereas the CRM is more complicated. Clinicians may fear that what they now do routinely under the three-patient cohort method will no longer be allowed by the CRM. Many clinicians either do not understand the definition of maximum tolerated dose (MTD) or believe that it automatically corresponds to 33% of the dose limiting toxicity (DLT) occurrence rate. Clinicians may be unfamiliar with ongoing collaboration with a statistician throughout the trial and with the information a clinician should provide to aid in decisions to escalate or deescalate dose levels. In order to overcome these issues, a statistician
needs tools to communicate with nonstatisticians. Graphical presentations are one such powerful tool that can be used to elicit prior assumptions and to convey the posterior results from the CRM.
