Choosing Appropriate Estimands in Clinical Trials
Choosing Appropriate Estimands in Clinical Trials
复制标题
DOI：
10.1177/2168479014567317
复制
发表时间：
2015-07-01
影响因子：
1.5
通讯作者：
Benda, Norbert
中科院分区：
文献类型：
--
作者：
Leuchs, Ann-Kristin;Zinserling, Joerg;Benda, Norbert
缺乏对研究方案的依从性和缺失数据在临床试验中通常是不可避免的，并且两者都增加了区分理想治疗效果（如果按照指示服用药物）和存在实际依从性模式的治疗效果的需要。在这方面，被估量已成为人们关注的焦点。被估量就是被估计的对象。在治疗获益的背景下，被估量可以涉及疗效或有效性方面。在决定试验设计和主要分析之前，定义感兴趣的被估量是必不可少的一步。被估量的选择对任何临床试验计划阶段要考虑的各种其他因素都有影响。这项研究提出了一个流程图，包括规划过程中要考虑的所有方面。决定主要被估量后，应指定试验设计，然后进行主要分析。两者都应该适当地解决所选择的被估量。最后，应考虑敏感性分析。提供了所有涉及的规划步骤的建议，特别是在疗效和有效性之间的选择，并从临床实践中的相关例子来说明他们。建议在任何临床试验的计划过程中牢记流程图，并为研究方案中的每个决定提供合理的理由。
Lack of adherence to study protocol and missing data are often unavoidable in clinical trials, and both increase the need to differentiate between the ideal treatment effect if the medication is taken as directed and the treatment effect in presence of the actual adherence pattern. In this regard, estimands have become the focus of attention. An estimand is simply that which is being estimated. In the context of treatment benefit, an estimand may address either efficacy or effectiveness aspects. Defining the estimand of interest is an essential step to take before deciding on trial design and primary analysis. The choice of estimand has consequences for various other factors to be considered during any clinical trial's planning phase. This study presents a process chart including all aspects to consider during planning. After deciding on the primary estimand, the trial design should be specified, followed by the primary analysis. Both should appropriately address the chosen estimand. Finally, sensitivity analyses should be taken into account. Provided are suggestions for all the planning steps involved, especially on choosing between efficacy and effectiveness, and relevant examples from clinical practice to illustrate them. It is recommended that one bear in mind the process chart during planning of any clinical trial and give reasonable justification for each decision in the study protocol.