|
Abstract:
|
In small to moderate sized multi-center clinical trials, as the number of variables to be balanced increases, simple or stratified randomization could result in serious treatment imbalances. Adaptive randomization (AR) may be used. When both baseline variables and study centers need to be considered, the question arises as how should centers be handled in AR. Two AR methods are investigated: Pocock-Simon (PS) and Pocock-Simon-Zelen (PSZ). In PS, study center is treated similarly as baseline variables. In PSZ, baseline variables are initially used to make provisional treatment assignment using PS method. Treatment balances are then checked against study centers using Zelen's algorithm. Computer simulated data and chi-square goodness-of-fit test are used to compare the methods. Results show that PS method achieved better treatment balances with respect to baseline variables and centers.
|
