ABSTRACT
As we mentioned in Chapter 1, most fundamental design issues in clinical trials arise from the need to strike a balance between scientific experiment and good clinical practice in a trial. Whereas laws and regulations guard the ethical aspects of trial design, statistics addresses scientific validity and efficiency, which in turn enhance the medical ethics and practice. Although many design issues surface in the development of a clinical trial, in this chapter we focus on the major concerns for trial validity, and we discuss the concerns for trial efficiency in Chapter 3. First, we consider the various types of validity: external validity, internal validity, and repeatability. We then explore the phenomenon of regression toward the mean to illustrate a kind of bias, which can be accounted for when there is a concurrent control group in the design. Logically, we need to consider at least two observations per subject in a clinical trial, one as the baseline and the other as a followup measurement. In this two-dimensional setting, we review the notion of bivariate normal distribution and the expression of simple linear regression as we discuss the topic of regression toward the mean. Finally, we conclude this chapter by discussing several methods of randomization and blinding that are commonly used in clinical trials to make the control group fully functional for achieving internal validity.
