Choice of Best Design

The pharmaceutical literature is full of theories and methods for arriving at a “best�? design (1-17), but ultimately the choice of which method to use depends on the goal of the clinical trial. Sheiner (18) provides a useful framework for establishing the question an experiment should address. Experiments can be performed to quantitate some feature of the system being investigated (LEARN) or con�?rm an observation about the system (CONFIRM). “What is the bias in an estimated parameter?�? is an example of a learning question. “Does the drug have a signi�?cant effect compared to placebo?�? is an example of a con�?rming question. With this framework in mind, design properties can be identi�?ed for optimization using analytical or simulation techniques