Statistical Methods and Adaptive Trial Designs for Cardiovascular Outcomes with Information Sharing

项目概要/摘要。K 01奖项的申请描述了研究和指导计划， 亚历山大凯泽博士的课程，他是该大学生物统计学和信息学的助理教授 科罗拉多州安舒茨医学院（CU-AMC），以实现对独立研究的进步， 使用适应性设计进行心血管结局临床试验，促进信息共享 不同的数据来源，以提高评估新疗法的统计效率。开发过程 传统上，有效的新疗法通过一系列研究和阶段进行。传统上，每个 阶段与以前的阶段独立处理，以前的阶段传统上只用于设计阶段， 重新审判这意味着对所有可用数据的使用可能效率低下，而这些数据可以在 设计阶段，代表了可能包括多种治疗的新试验设计的重要限制 在单个协议的上下文中，但是其中比较仅使用同时收集的数据。统计 本申请中提出的方法和试验设计通过开发新的方法来解决这一限制 以促进信息共享沿着平台试验设计的应用。在这个奖项中， 统计方法将与心血管系统生物学的正式培训相结合，以确保 这些新方法在心血管结局研究的设计中具有无缝应用。实现 根据本K 01中列出的培训目标和研究目标，组建了一个由三名导师组成的研究团队。 博士CU-AMC的生物统计学和信息学教授John Kittelson是临床试验设计专家 并在心血管试验实施和分析方面拥有丰富的经验。格雷戈里·施瓦茨医生 CU-AMC医学教授和VA医疗中心心脏科主任，是 提出、实施和传播心血管结局临床试验。Robert Eckel博士，教授 他是CU-AMC的医学博士，美国心脏协会的前任主席，美国心脏协会的当选主席。 糖尿病协会，是一个专家在脂质和脂蛋白代谢和糖尿病。 在目标1中，我们将开发统计方法，以纳入来自补充来源的数据，例如过去的数据。 试验，进入基于它们的交换率的当前研究的分析（即，调整后， 协变量（也称为信息共享）。在目标2中，我们将开发自适应平台试验设计， 与共用对照组相比，考虑新的治疗组。为了提高新方法的可及性， 将开发方便用户的软件（目标3）。从这些目标的方法和设计将进行评估 通过严格的模拟研究，了解它们在各种场景下的小样本特性。方法 将通过应用于NHLBI提供的先前进行的心血管试验数据进行说明 BioLINCC.总之，这些用于信息共享和适应性试验设计的方法将提高效率 研究过程中，充分利用现有的信息进行统计推断。

Project Summary/Abstract. This application for a K01 award describes the research & mentoring plans and coursework for Dr. Alexander Kaizer, an Assistant Professor of Biostatistics and Informatics at the University of Colorado-Anschutz Medical Campus (CU-AMC), to achieve advancement towards independent research in the use of adaptive designs for cardiovascular outcome clinical trials that facilitate information sharing across different sources of data to improve the statistical efficiency of evaluating new therapies. The process to develop effective novel treatments traditionally proceeds through a series of studies and phases. Conventionally, each phase is treated independently from previous phases, which are traditionally only used in the design stage of a new trial. This represents a potentially inefficient use of all available data that could be incorporated beyond the design stage and represents an important limitation for newer trial designs that may include multiple treatments within the context of a single protocol, but where comparisons only use concurrently collected data. The statistical methodologies and trial designs proposed in this application address this limitation by developing new methods to facilitate information sharing along with applications to platform trial designs. In this award, the development of statistical methods will be coupled with formal training in the biology of the cardiovascular system to assure that these new methods have seamless application in the design of cardiovascular outcomes research. To achieve the training goals and research aims laid out in this K01, a research team of three mentors has been assembled. Dr. John Kittelson, Professor of Biostatistics and Informatics at CU-AMC, is an expert in clinical trial design and has extensive experience with cardiovascular trial implementation and analysis. Dr. Gregory Schwartz, Professor of Medicine at CU-AMC and Chief of the Cardiology Section at the VA Medical Center, is a leader in proposing, implementing, and disseminating cardiovascular outcome clinical trials. Dr. Robert Eckel, Professor of Medicine at CU-AMC, past President of the American Heart Association, and President-elect of the American Diabetes Association, is an expert in lipid and lipoprotein metabolism and diabetes. In Aim 1, we will develop statistical methods for incorporating data from supplemental sources, such as past trials, into the analysis of a current study based on their exchangeability (i.e., equivalence) after adjusting for covariates (also known as information sharing). In Aim 2, we will develop adaptive platform trial designs that consider new treatment arms compared to a shared control arm. To improve the accessibility of the new methods, user-friendly software will be developed (Aim 3). The methods and designs from these aims will be evaluated via rigorous simulation study to understand their small sample properties under various scenarios. Methods will be illustrated through application to previously conducted cardiovascular trial data available from the NHLBI BioLINCC. Together, these methods for information sharing and adaptive trial designs will improve the efficiency of the research process and take fuller advantage of available information for statistical inference.