Methods to improve efficiency and robustness of clinical trials using information from real-world data with hidden bias

项目摘要/摘要 随机对照试验(RCT)是评价fi的安全性和有效性的金标准方法。 癌症和神经疾病等疾病的治疗。由于疾病的异质性- 遗传、疾病罕见或登记差异，可用于临床试验的患者有限， 使药物开发既昂贵又耗时，试验失败率很高。21世纪 Cres Act呼吁监管机构和药物开发商考虑创新的临床试验 设计将传统临床试验与真实世界数据(RWD)联系起来，以克服其中的一些问题 限制。RWD的外部控制(EC)已用于构建CON中的比较器臂。 fi的试验性试验最终获得了监管决策者的批准。然而，人们担心的是 关于RWD与RCT的有效性和可比性，限制了它们在更广泛的背景下的使用 到目前为止。选择偏差、变量Defi的差异和未测量的混杂可导致 如果RWD与RCT合并，则治疗效果估计有偏差，推断不正确。目标 在此项目中，重点解决集成RWD时隐藏的偏见，以提高fi的效率 临床试验。我们将开发一种新的敏感性分析框架，用于外部现实世界的使用 用于评估结果对隐藏偏差的稳健性的控制，以及稳健和有效的fi分析 有选择地借用和调整数据差异的方法，以减轻隐藏的双 ASES.我们将积极从事这些新方法的传播和翻译给研究人员 从工业、学术和监管机构到模范应用程序、最先进的软件、 统计分析计划模板、足智多谋的网站、研讨会和辅导课程。我们的终极- Mate的目标是促进在监管决策中更有力地使用真实世界的证据。

PROJECT SUMMARY/ABSTRACT Randomized controlled trials (RCTs) are the gold-standard method of evaluating the safety and effi- cacy of treatments for diseases, such as cancer and neurological disorders. Due to disease hetero- geneity, disease rarity, or enrollment disparities, there are limited patients available for clinical trials, making drug development costly and time-consuming with trials failing at a high rate. The 21st Century Cures Act has called for regulatory agencies and drug developers to consider innovative clinical trial designs that bridge conventional clinical trials with real-world data (RWD) to overcome some of these limitations. External controls (ECs) from RWD have been used to construct the comparator arm in con- firmatory trials that eventually received approval from regulatory decision-makers. However, concerns regarding the validity and comparability of RWD with RCTs have limited their use in a broader context thus far. Selection bias, differences in variable definitions, and unmeasured confounding can lead to biased treatment effect estimates and incorrect inference if RWD are integrated with RCTs. The aims in this project focus on addressing hidden biases when integrating RWD to improve the efficiency of clinical trials. We will develop a novel sensitivity analysis framework for the use of external real-world controls to assess the robustness of results to hidden biases, as well as robust and efficient analysis methods that selectively borrow and adjust for data discrepancies to mitigate the impact of hidden bi- ases. We will actively engage in dissemination and translation of these new methods to researchers from industry, academic and regulatory agencies through exemplary applications, state-of-art software, statistical analysis plan template, resourceful website, and workshops and tutorial sessions. Our ulti- mate goal is to facilitate more robust use of real-world evidence in regulatory decision-making.