Office of Surveillance and Epidemiology, Center for Drug Evaluation and Research, U.S. Food and Drug Administration, Silver Spring, MD 20993, USA.
Clin Trials. 2011 Oct;8(5):559-70. doi: 10.1177/1740774511419165. Epub 2011 Aug 30.
Randomized clinical trials (RCTs) are often positioned at the top of evidence hierarchies. Meta-analyses of RCTs aim to integrate the state of knowledge on a given scientific question, particularly for rare drug-related outcomes. However, although RCTs are valuable tools in our armamentarium, they are rarely designed to evaluate drug safety and are thus susceptible to limitations that may hamper the ability of both RCTs and meta-analyses to fully characterize the safety profiles of drugs. Their potential limitations might be exacerbated in the study of rare outcomes, often encountered in drug safety assessment, when even minor deviations from the intended randomization could impact the stability of the risk estimates.
This article considers the methodological caveats of both RCTs and meta-analyses of RCTs pertinent to the study of drug-related harms. It is intended to stimulate discussion about the impact of these caveats on interpreting findings of RCTs and meta-analyses for drug safety, which would foster more robust, critical evaluations, and thus enhance clinical and regulatory decision-making.
Pertinent issues that can influence the interpretation of drug-related harms discussed in this article were based on authors' expertise and review of the literature.
Investigators and clinicians should be cognizant of the potential limitations of the secondary use of RCTs and meta-analyses in the assessment of drug-related harms and, when applicable, should consider potential remedies to overcome these limitations.
Only few practical examples are included in the article due to the fact that many of the discussed caveats are not examined and/or reported in many publications. In addition, the confidential nature of data reviewed at a regulatory agency forestalls an in depth discussion of examples pertaining to specific drugs. Furthermore, our ability to quantify the extent of encountering, or the actual impact of, the caveats addressed in this review on the RCTs findings is limited. It is worth noting that the mere encounter of a given caveat does not mean that it will obviate the utility of drug safety information from a given trial. The extent of its impact is expected to vary based on the specifics of the trial, the drugs studied, the indications, and the nature of the adverse events.
Although some of the limitations described are inherent in RCTs, some of the sources of bias highlighted in this article could be minimized by careful RCT design, planned follow-up, and improved collection of information on adverse events. As future research sheds more light on pertinent knowledge gaps and issues, the ability to maximize the use of RCTs and meta-analyses of RCTs to address drug safety questions of interest will be greatly enhanced.
随机临床试验 (RCT) 通常处于证据层级的顶端。RCT 的荟萃分析旨在整合关于给定科学问题的知识状态，特别是对于罕见的药物相关结局。然而，尽管 RCT 是我们工具包中的宝贵工具，但它们很少被设计用于评估药物安全性，因此可能存在限制，这可能会影响 RCT 和荟萃分析充分描述药物安全性概况的能力。在药物安全性评估中经常遇到的罕见结局研究中，即使是与预期随机化的微小偏差也可能影响风险估计的稳定性，这些限制可能会加剧。
本文考虑了与药物相关危害研究相关的 RCT 及其荟萃分析的方法学注意事项。旨在激发关于这些注意事项对解释 RCT 和荟萃分析药物安全性发现的影响的讨论，从而促进更稳健、批判性的评估，从而增强临床和监管决策。
本文讨论的药物相关危害相关问题是基于作者的专业知识和文献综述。
研究人员和临床医生应该意识到在评估药物相关危害时，二次使用 RCT 和荟萃分析可能存在的潜在局限性，并且在适用的情况下，应考虑潜在的补救措施来克服这些局限性。
由于许多讨论的注意事项在许多出版物中没有被检查和/或报告，因此本文仅包含少数实际示例。此外，监管机构审查的数据保密性阻止了对特定药物相关示例的深入讨论。此外，我们量化本文所述注意事项在 RCT 研究结果中遇到的程度或实际影响的能力有限。值得注意的是，遇到给定的注意事项并不意味着它将消除来自特定试验的药物安全性信息的效用。其影响程度预计将根据试验的具体情况、研究的药物、适应症和不良事件的性质而有所不同。
尽管有些局限性是 RCT 固有的，但通过仔细的 RCT 设计、计划的随访以及改善对不良事件信息的收集，可以最大限度地减少本文强调的一些偏倚来源。随着未来的研究更深入地了解相关知识空白和问题，将极大地增强最大限度地利用 RCT 和 RCT 荟萃分析来解决药物安全性问题的能力。
