Dose Finding by the Continual Reassessment Method

This book presents the continual reassessment method (CRM) as a tool for dose-finding studies. With a focus on the implementation and practice of the CRM and its variations, it explains how the CRM may be calibrated and extended to suit common clinical settings. The book includes examples of real clinical trials data to illustrate the calibration techniques and shows how R can be used to carry out the techniques. It reviews the literature, related methodology, and theoretical properties of the CRM. It also explores alternatives for situations where the CRM fails.

As clinicians begin to realize the important role of dose-finding in the drug development process, there is an increasing openness to "novel" methods proposed in the past two decades. In particular, the Continual Reassessment Method (CRM) and its variations have drawn much attention in the medical community, though it has yet to become a commonplace tool. To overcome the status quo in phase I clinical trials, statisticians must be able to design trials using the CRM in a timely and reproducible manner.

A self-contained theoretical framework of the CRM for researchers and graduate students who set out to learn and do research in the CRM and dose-finding methods in general, Dose Finding by the Continual Reassessment Method features:

• Real clinical trial examples that illustrate the methods and techniques throughout the book
• Detailed calibration techniques that enable biostatisticians to design a CRM in timely manner
• Limitations of the CRM are outlined to aid in correct use of method

This book supplies practical, efficient dose-finding methods based on cutting edge statistical research. More than just a cookbook, it provides full, unified coverage of the CRM in addition to step-by-step guidelines to automation and parameterization of the methods used on a regular basis. A detailed exposition of the calibration of the CRM for applied statisticians working with dose-finding in phase I trials, the book focuses on the R package ‘dfcrm’ for the CRM and its major variants.

The author recognizes clinicians’ skepticism of model-based designs, and addresses their concerns that the time, professional, and computational resources necessary for accurate model-based designs can be major bottlenecks to the widespread use of appropriate dose-finding methods in phase I practice. The theoretically- and empirically-based methods in Dose Finding by the Continual Reassessment Method will lessen the statistician’s burden and encourage the continuing development and implementation of model-based dose-finding methods.