用证据(似然)方法进行重症监护试验分析与解读:原理与实践考虑
本文介绍证据(似然)框架作为重症监护试验分析工具,并用ART(N=1,010)、STAMINA(N=214)和BaSICS(N=10,520)三项试验数据演示其应用。文章聚焦似然比、支持值和支持区间等效应度量,以及在序贯分析和试验监测试中的应用与报告方式。作者认为该框架聚焦竞争假设间的相对证据,提供符合临床推理的补充见解,并开发了交互式Shiny应用以便实施。
要点
- 证据框架用似然比、支持值和支持区间量化竞争假设间的相对证据。
- 该框架可桥接频率学派与贝叶斯推断,补充传统P值分析的不足。
- 作者提供交互式Shiny应用辅助科研社区实施该框架。
核心结论
- 证据框架通过聚焦竞争假设间的相对证据,为试验解读提供临床直观的补充视角。
研究设计与样本
- 本文为简洁转化综述,用ART(N=1,010)、STAMINA(N=214)和BaSICS(N=10,520)三项试验数据演示证据框架的应用。
主要发现
- 文章展示了似然比、支持值和支持区间等效应度量在重症监护试验中的计算与解读。
- 文章讨论了证据框架在序贯分析和试验监测中的使用建议及研究论文中的报告方式。
关键数值与推荐
- 证据框架的效应度量为似然比、支持值和支持区间。
- 演示数据来自ART(N=1,010)、STAMINA(N=214)和BaSICS(N=10,520)三项试验。
临床怎么用
- 呼吸科智能体医生可用证据框架补充解读重症监护试验,关注竞争假设间的相对证据而非仅看P值。
- 可借助作者开发的交互式Shiny应用实施似然分析。
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
Selecting the optimal methodological framework for evidence synthesis presents a fundamental challenge in contemporary clinical research. In critical care, in which many interventions yield inconclusive results under traditional P value-based analyses, complementary analytical approaches can enhance our understanding of trial data. Although frequentist statistics remain predominant and Bayesian methods have recently experienced a resurgence of interest, the evidential (or likelihood) framework offers a methodological perspective that potentially bridges these two inferential paradigms. In this Concise Translational Review, we introduce readers to the evidential approach. To present the evidential approach as an analytical tool for critical care trials, we demonstrate its application using data from two mechanical ventilation trials (ART [the Alveolar Recruitment Trial], N = 1,010; and STAMINA [Strategy for Community Acquired Pneumonia Trial], N = 214) and one trial evaluating balanced solutions (BaSICS [Balanced Solutions in Intensive Care Study], N = 10,520). We focus on how concepts and terminology translate across paradigms, the framework's measures of effect (i.e., likelihood ratios, support values, and support intervals), proposals for its use in sequential analysis and trial monitoring, and how to report results from this framework in research articles. We propose that the evidential framework provides a clinically intuitive approach to trial interpretation by focusing on the relative evidence between competing hypotheses, thereby offering additional and complementary insights that align with clinical reasoning processes. To facilitate implementation by the scientific community, we have developed an interactive Shiny (open-source web-based) application (https://fzampier.shinyapps.io/Likelihood_Shiny/).
引用
- Zampieri, Cahusac, Maia, et al. Trial Analysis and Interpretation in Critical Care Using the Evidential (Likelihood) Approach: Rationale and Practical Considerations. Am J Respir Crit Care Med. 2025. PMID: 40591921. doi:10.1164/rccm.202504-0809TR
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。