Li, Shiansong. Bayesian statistical analysis in a phase II dose-finding trial with survival endpoint in patients with B-cell chronic lymphocytic leukemia. Retrieved from https://doi.org/doi:10.7282/T3NC5ZF3
DescriptionBayesian approaches have been widely used in designing, monitoring and analyzing clinical studies in recent years. We utilize Bayesian parametric and non-parametric statistical methods in interim monitoring and decision-making for a phase II dose-finding trial with survival endpoint. The objective of the clinical trial is to find an optimal treatment schedule at the end of the study for planning future studies, using Bayesian decision rules. The primary efficacy outcome is time to progression. Binomial-Beta model and Exponential-Gamma model are included in parametric methods. Non-parametric methods include Bayesian life-table, Beta process model, Dirichlet process model and Gibbs sampling. Simulations are conducted for each of the statistical methods under 9 different scenarios including truncated exponential entry time, and the probability of a treatment-schedule being chosen is calculated based on 1,000 simulation studies. Finally, these different statistical methods are used to find optimal treatment-schedule among 3 arms in the phase II CLL clinical trial using the most recent unblinded data.