随机、对照与真实:对真实世界证据术语的批判性分析
本文批判性分析真实世界证据(RWE)术语及其含义,指出按与常规实践的接近程度而非方法学特征分类临床研究在概念上具有误导性,随机对照试验与观察性研究均在同一经验现实中开展,区别在于观察与分析的方法学结构而非真实性。
要点
- RWE标签基于与常规实践的接近程度而非方法学特征分类研究,概念上误导。
- 随机化与实验控制减少系统误差和不确定性,而非脱离真实世界背景。
- 代表性不等同于科学有效性,也非信息性推断的必要条件。
核心结论
- 放弃RWE等修辞性标签可促进概念清晰,并鼓励基于明确方法学原则评估证据。
研究设计与样本
- 本文为评论性文章,对RWE术语及其含义进行批判性分析。
主要发现
- 随机与观察性研究均在同一经验现实中开展,区别在于观察与分析的方法学结构而非真实性。
- 随机化与实验控制减少系统误差和不确定性以更好分离因果效应,而非脱离真实世界背景。
- 代表性不等同于科学有效性,也非信息性推断的必要条件,关键问题常是估计值对特定决策情境的情境适用性。
- 所有流行病学研究设计均涉及无偏性、精确性与泛化性之间的权衡,随机试验、观察性研究与准实验设计占据不同方法学维度位置而非根本不同认识论类别。
临床怎么用
- 评估证据时应基于明确方法学原则,而非依赖RWE等修辞性标签。
- 应关注估计值对特定决策情境的情境适用性,而非仅追求人群-wide泛化性。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
原文
OBJECTIVES: The growing use of the term "real-world evidence" (RWE) has introduced a classification of clinical research based on its perceived proximity to routine practice rather than its methodological characteristics. Although the term is often used operationally to refer to evidence derived from routinely collected clinical data or routine care settings, it has also been criticized for implying that such studies are inherently closer to "real life" than traditional randomized controlled trials (RCTs). In this Commentary, we argue that such a distinction is conceptually misleading. METHODS: A critical analysis of RWE terminology and its implications. RESULTS: Both randomized and observational studies are conducted within the same empirical reality; what differentiates them is not their "realness," but the methodological structures used to observe and analyze that reality. Randomization and other forms of experimental control do not detach research from real-world contexts but, rather, reduce systematic error and uncertainty to better isolate causal effects. We further examine the common claim that RWE provides greater external validity through more "representative" populations. Representativeness, however, is neither synonymous with scientific validity nor always necessary for informative inference. In many cases, the key question is not population-wide generalizability but the contextual applicability of estimates to specific decision settings. More broadly, all epidemiological study designs involve trade-offs among unbiasedness, precision, and generalizability. Randomized trials, observational studies, and quasi-experimental designs therefore occupy different positions along these methodological dimensions rather than constituting fundamentally different epistemic categories. CONCLUSION: We conclude that abandoning rhetorically charged labels such as RWE may promote greater conceptual clarity and encourage evaluation of evidence based on explicit methodological principles.
引用
- Al-Kassab-Córdova A, Burgess S, Bashir N, et al. Randomized, controlled, and real. J Clin Epidemiol. 2026. PMID: 42425232. doi:10.1016/j.jclinepi.2026.112414
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。