Simulation Practices for Adaptive Trial Designs in Drug and Device Development
Simulation Practices for Adaptive Trial Designs in Drug and Device Development
复制标题
DOI：
10.1080/19466315.2018.1560359
复制
发表时间：
2019-04-22
影响因子：
1.8
通讯作者：
Cicconetti, Greg
中科院分区：
文献类型：
--
作者：
Mayer, Cristiana;Perevozskaya, Inna;Cicconetti, Greg
适应性临床试验是现代药物和设备开发的基石。美国最新立法呼吁提高临床试验设计效率，重点是扩大复杂创新设计的利用，而这些设计必须通过模拟才能开发。众所周知，临床试验模拟是探索、比较和理解不同设计中嵌入的操作特征、统计特性和自适应决策以回答给定研究问题的基本工具。本文提供了业界对由一组统计学家在药物信息协会自适应设计科学工作组的赞助下编写的模拟报告的见解。这项工作旨在说明确保在进行和报告适应性临床试验模拟时具有更高的一致性和清晰度所需的关键共同要素，消除与不同受众沟通技术设计方面的不必要障碍，并促进候选设计的利弊评估。适用于特定类型的适应性试验的模拟报告的设计相关元素通过剂量递增设计、剂量范围研究、样本量重新估计和早期停止规则的试验以及验证性多阶段设计的示例来呈现。
Adaptive clinical trials are the cornerstone of modern drug and device development. The most recent US legislation calls for higher efficiency in designing clinical trials with emphasis on expanding the utilization of complex innovative designs that cannot be developed without simulations. It is well recognized that clinical trial simulation is a fundamental tool to explore, compare, and understand the operating characteristics, statistical properties, and adaptive decisions embedded in different designs to answer the given research questions. This article provides insights from industry on the development of a simulation report from a group of statisticians brought together under the sponsorship of the Drug Information Association Adaptive Design Scientific Working Group. This effort intends to illustrate the key common elements required to ensure higher consistency and clarity in conducting and reporting simulations of adaptive clinical trials, eliminate unnecessary barriers in communicating technical design aspects to different audiences, and facilitate the assessment of pros and cons of candidate designs. The design-dependent elements of a simulation report applicable to specific types of adaptive trials are presented with the examples of dose-escalation designs, dose-ranging studies, trials with sample size re-estimation and early stopping rules, and confirmatory multistage designs.