文献速递脑血管病

减轻与量化急性卒中试验中的挑选患者问题

Mitigating and Quantifying Cherry-Picking in Acute Stroke Trials.
PubMed 文献 · Stroke · 2025-04-28

本文讨论急性卒中随机对照试验中因选择性招募患者(挑选患者)而损害结果普遍性的问题,分析其成因(如研究者激励问题),评述识别与减轻策略,并提出一种量化挑选患者并区别于单纯招募不足的方法。

要点

  • 连续入组合格患者是随机对照试验内部与外部真实性的基础。
  • 选择性招募(挑选患者)会严重削弱试验结果的普遍性。
  • 提出量化挑选患者的方法,以区别于单纯招募不足。

核心结论

  • 只有持续量化挑选患者问题,才能在解决该问题并增强随机对照试验有效性方面取得有意义的进展。

研究设计与样本

  • 本文为综述类文章,讨论急性卒中试验中挑选患者的问题、成因、识别策略、减轻策略及量化方法。

主要发现

  • 当入组患者不能代表更广泛目标人群时,试验结果的普遍性会大大削弱,若大部分本合格个体在试验外接受治疗则尤其可能。
  • 文章探讨了导致挑选患者的因素,如次级研究者的「What's In It For Us」问题。
  • 文章讨论了识别试验中正在发生挑选患者的策略,并批判性审查了多种减轻挑选患者的策略。
  • 文章提出了一种在急性卒中试验中量化挑选患者同时区别于单纯招募不足的方法。

局限性

  • 原文未述局限性

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

Consecutive enrollment of eligible patients is fundamental to the internal and external validities of randomized controlled trials. The generalizability of trial results is greatly undermined when enrolled patients are not representative of the broader target population, which is especially likely if a large proportion of otherwise-eligible individuals receive treatment outside the trial. In this article, we discuss the problem that such selective recruitment or cherry-picking of patients poses to clinical trials. We explore factors contributing to such cherry-picking, such as the What's In It For Us problem for subinvestigators, and discuss strategies to identify when cherry-picking is going on in a trial. We also critically examine various strategies to mitigate cherry-picking. Moreover, we propose a method to quantify cherry-picking within an acute stroke trial while distinguishing it from simple underrecruitment. It is only when we seek to consistently quantify the problem of cherry-picking, that we will make meaningful strides toward resolving this issue and further strengthening the validity of our randomized controlled trials.

来源:Stroke,查看原文。版权归原作者所有。

引用

  1. Goyal M, Ganesh A. Mitigating and Quantifying Cherry-Picking in Acute Stroke Trials. Stroke. 2025. PMID: 40294172. doi:10.1161/STROKEAHA.125.051062

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。