Prior to randomization,
there are often some baseline prognostic factors (covariates) collected,
such as age, gender, and disease severity, which may have an important
impact on the treatment outcomes. It is thus desirable to balance
treatment allocation over these covariates so that valid and efficient
treatment comparisons can be conducted.
The simplest way to
balance the treatment assignment over important covariates is through
stratification. As shown in
Figure 9.1 Illustration of Stratified Randomization, trial participants are first grouped into mutually exclusive strata based on the values of their prognostic variables, and then they are randomly allocated to each treatment ...