|
Abstract:
|
Clinical trials operations require randomization of treatments to subjects to be efficient and appropriate for any study's particular needs. Commonly used randomization methods include stratified permuted block, complete randomization, and covariate adjusted methods that assign treatments to reduce treatment imbalances within randomization factors. The covariate adjusted approach extends the minimization method of Taves (1974) and sequential assignment with balancing of Pocock & Simon (1975). It requires the user to select relative weights for imbalances within each randomization factor, sites, and strata. We compare the performance of a range of randomization plans, as the number of factors and the relative weights for imbalances are varied. Criteria include minimizing selection bias, enabling the blinding of treatments to observers, and minimizing overall imbalance.
|
