Simulating a Two-Arm Random Allocation Model Using a Bayesian Dynamic Linear Model Approach

Random allocation models used in clinical trials can help reduce the between group bias which may occur when comparing multiple treatment groups to determine the preferred treatment method. Often however, this determination leaves researchers battling ethical issues of providing patients with unfavorable treatments. Many methods such as Play the Winner and Randomized Play the Winner Rule have historically been utilized to determine patient allocation, yet, these methods are prone to increased unfavorable assignments. Recently a new Bayesian Method using Decreasingly Informative Priors has been proposed but, this method can be time consuming when multiple individual Markov Chain Monte Carlo (MCMC) updating methods are required. We present an experimental design study via simulation using Dynamic Linear Models as an alternative method to increase allocation speed while also decreasing patient allocation samples necessary to identify the more favorable outcome and we illustrate the effectiveness of this new method. Furthermore, a sensitivity analysis is conducted on multiple parameters to demonstrate the method is robust. Finally, a Bayes Factor criterion is used to determine termination through decisive evidence in favor of the better treatment assignment.