Each week, we publish layman’s abstracts of new articles from our prestigious portfolio of journals in statistics. The aim is to highlight the latest research to a broader audience in an accessible format.
The article featured today is from Statistics in Medicine, with the full article now available to read here.
A practical response adaptive block randomization (RABR) design with analytic type I error protection. Statistics in Medicine. 2021; 40: 4947– 4960. https://doi.org/10.1002/sim.9104, , , , , .
As an appealing branch of adaptive clinical trials, response adaptive randomization (RAR) design can make interim modifications of the probability of a newly enrolled subject being assigned to a treatment arm based on accumulating data. Statistical, ethical, and pragmatic rationales are supporting the advantage of using RAR as more subjects are being randomized to the more promising treatment arms. However, applications of RAR in confirmatory drug clinical trials with multiple active arms are limited largely due to its complexity, and lack of control of randomization ratios to different treatment groups. As a possible solution, authors propose a Response Adaptive Block Randomization (RABR) design with flexible design parameters to meet specific study objectives of confirmatory clinical trials with multiple active treatment groups. The block feature facilitates its application in practice, in the sense that the Interactive Response Technology (IRT) schedules can be built before current trial conduct. They provide detailed descriptions of the randomization procedures utilized by RABR. Moreover, authors analytically prove that unweighted statistics in RABR have controlled type I error rates. This feature is attractive for multi-arm studies with complex multiplicity adjustment and enables RABR to adaptively randomize each future subject based on interim data. Advantages of the proposed RABR in terms of robustly reaching target final sample size to meet regulatory requirements and increasing statistical power as compared with some comparators are demonstrated by simulation studies and a practical clinical trial design. Essentially a more efficient and ethical RAR design can be implemented with fewer patients enrolled to fulfill unmet medical needs. Guidance and discussion on how to choose hyperparameters of RABR are also provided. Authors share their R code on GitHub and develop the R package “RABR” on the Comprehensive R Archive Network (CRAN) to evaluate operating characteristics of the proposed method.