The book offers a detailed overview of hypothesis-based adaptive design for clinical trials. It includes superiority, noninferiority and equivalence trials testing which may be implemented in multiple stages. Unlike the classical design, which parameters are assumed to be fixed, adaptive design allows to change parameters dynamically depending on currently available information. This leads to considerable cost/time savings in drug development without undermining the validity and integrity of the trial. The book presents the second edition of the monograph and contains 29 chapters (12 chapters of those are new). Each chapter includes theoretical background, examples, SAS and/or R codes and problems.
In Chapter 2, the classical statistical design methods are overviewed. In Chapter 3, unified theory for adaptive designs, which covers four key statistical elements in adaptive designs: stopping boundary, adjusted \(p\)-value, point estimation, and confidence interval, is introduced. In Chapters 4–5, methods with direct combination of \(p\)-values and method with inverse-normal \(p\)-values are considered. In Chapters 6–7, classical and adaptive noninferiority designs with paired binary data are discussed, including the case of incomplete paired data.
In Chapter 8, two-stage adaptive designs are generalized to \(K\)-stage designs using analytical methods, SAS-macros, and R-functions. In Chapter 9, in addition to classical methods, a conditional error function method and conditional power are introduced. In particular, the Proschan-Hunsberger method and the Muller-Schafer method are discussed. In Chapter 10, the recursive two-stage adaptive design, which offers a closed-form solution for \(K\)-stage designs, is considered.
In Chapters 11–12, unblinded, blinded and semi-blinded sample-size reestimation designs are studied. Adaptive design with coprimary endpoint and multiple-endpoint adaptive design are presented in Chapters 13–14. In Chapters 15–16, pick-the-winners design and the add-arm design for unimodal response are presented. In Chapters 17–18, adaptive biomarker-enrichment design methods are developed for classifier, diagnosis, and predictive markers. In Chapter 19, survival modeling and adaptive treatment switching, which is motivated by an ethical consideration, are studied. In Chapter 20, response-adaptive allocation design, e.g. randomized play-the-winner, is considered.
Chapter 21 presents an introduction to the Bayesian approach in clinical trial which is used together with the frequentist approach in the rest of the monograph. It shows how to use Bayesian decision theory to further improve the efficiency of adaptive designs. In Chapters 22–23, the adaptive dose-finding designs, or dose-escalation designs, are discussed. In Chapter 24, biosimilarity trial is considered. In Chapter 25, regulatory requirements and optimal adaptive design for global multiregional clinical trial are discussed. In Chapter 26, SAS and R modules for group sequential design are presented. In Chapter 27, techniques for data analysis of adaptive trial, such as point and confidence parameter estimates, and adjusted \(p\)-values, are discussed. In Chapter 28, the logistic issues with adaptive designs are discussed. It also includes the most concurrent regulatory views and recommendations. Chapter 29 includes debates in adaptive designs.