在随机对照试验出版物中可视化呈现危害:共识与建议
该共识研究通过方法学综述和专家投票,从28种可视化方法中推荐10种供研究者在主要研究结果出版物中考虑使用。可视化方法的选择取决于结局类型和场景,并提供了决策树、示例、潜在局限性及统计软件实现代码,以改善危害信息的呈现与解读。
要点
- 从28种可视化方法中推荐10种供随机对照试验出版物考虑使用。
- 可视化选择取决于结局类型(如二分类、计数、时间至事件、连续型)和场景。
- 提供决策树辅助选择,并附示例、解读、局限性和统计软件代码。
核心结论
- 可视化是传达临床试验危害的有力工具,为传统频率表提供了替代视角。
- 增加可视化在危害结局中的使用将提供更清晰的信息呈现和更有信息量的解读。
- 统计师和临床试验团队必须最终决定最适合其数据和目标的可视化方法。
- 试验者应继续同时检查粗数字和可视化以全面理解危害特征。
研究设计与样本
- 共识研究,参与方包括15个英国临床研究协作组注册的临床试验单位、一个学术人口健康部门、Roche Products和The BMJ。
- 参与者包括20名学术统计师、1名工业统计师、1名学术卫生经济学家、1名数据图形设计师和2名临床医生。
- 通过方法学综述识别可视化方法,并在三次会议中以至少60%的可用投票达成共识。
- 得分略低于阈值(50-60%)的可视化经进一步讨论和重新投票直至达成共识。
主要发现
- 共考虑28种可视化方法,其中10种推荐供研究者在主要研究结果出版物中考虑使用。
- 可视化方法的选择取决于结局类型(如二分类、计数、时间至事件或连续型)和场景(如总结多个新发事件或一个关注事件)。
- 提供了决策树以帮助试验者决定使用哪些可视化方法。
- 每种推荐可视化方法均提供示例、示例解读、潜在局限性以及跨标准统计软件的实现代码指引。
- 临床医生反馈被纳入建议中的解释信息以帮助理解和解读。
关键数值与推荐
- 共识阈值:至少60%的可用投票。
- 得分50-60%的可视化方法经进一步讨论和重新投票直至达成共识。
临床怎么用
- 原文未据此给出推荐。
局限性
- 讨论了每种可视化方法的局限性,并给出了使用不恰当的示例。
- 尽管决策树有助于选择可视化方法,但统计师和临床试验团队必须最终决定最适合其数据和目标的可视化方法。
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
OBJECTIVE: To improve communication of harm in publications of randomised controlled trials via the development of recommendations for visually presenting harm outcomes. DESIGN: Consensus study. SETTING: 15 clinical trials units registered with the UK Clinical Research Collaboration, an academic population health department, Roche Products, and The BMJ. PARTICIPANTS: Experts in clinical trials: 20 academic statisticians, one industry statistician, one academic health economist, one data graphics designer, and two clinicians. MAIN OUTCOME: measures A methodological review of statistical methods identified visualisations along with those recommended by consensus group members. Consensus on visual recommendations was achieved (at least 60% of the available votes) over a series of three meetings with participants. The participants reviewed and critically appraised candidate visualisations against an agreed framework and voted on whether to endorse each visualisation. Scores marginally below this threshold (50-60%) were revisited for further discussions and votes retaken until consensus was reached. RESULTS: 28 visualisations were considered, of which 10 are recommended for researchers to consider in publications of main research findings. The choice of visualisations to present will depend on outcome type (eg, binary, count, time-to-event, or continuous), and the scenario (eg, summarising multiple emerging events or one event of interest). A decision tree is presented to assist trialists in deciding which visualisations to use. Examples are provided of each endorsed visualisation, along with an example interpretation, potential limitations, and signposting to code for implementation across a range of standard statistical software. Clinician feedback was incorporated into the explanatory information provided in the recommendations to aid understanding and interpretation. CONCLUSIONS: Visualisations provide a powerful tool to communicate harms in clinical trials, offering an alternative perspective to the traditional frequency tables. Increasing the use of visualisations for harm outcomes in clinical trial manuscripts and reports will provide clearer presentation of information and enable more informative interpretations. The limitations of each visualisation are discussed and examples of where their use would be inappropriate are given. Although the decision tree aids the choice of visualisation, the statistician and clinical trial team must ultimately decide the most appropriate visualisations for their data and objectives. Trialists should continue to examine crude numbers alongside visualisations to fully understand harm profiles.
引用
- Phillips R, Cro S, Wheeler G, et al. Visualising harms in publications of randomised controlled trials: consensus and recommendations. BMJ. 2022. PMID: 35577357. doi:10.1136/bmj-2021-068983
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。