随机临床试验中固定P值显著性阈值的应用不确定性
本文讨论随机临床试验中长期使用固定P值显著性阈值(如P<0.05)的问题,指出固定α水平可能不符合受益原则,增加假阳性与假阴性错误风险,降低阈值至0.005不能解决可重复性危机,建议根据研究设计、样本量、先验证据和效应量等因素采用灵活的α水平。
要点
- 固定P值阈值可能不符合受益原则,增加假阳性和假阴性错误风险。
- 降低阈值至0.005可减少假阳性但增加假阴性,未解决可重复性危机。
- 显著性阈值的选择应视情境而定,受研究设计、样本量、先验证据和效应量影响。
核心结论
- 僵化遵守固定α水平,无论多严格,可能在方法学上受限并在伦理上存疑。
研究设计与样本
- 本文为针对随机临床试验中固定P值显著性阈值使用的讨论性文章,未报告系统检索范围或样本量。
主要发现
- 依赖单一固定α水平增加假阳性(I型)和假阴性(II型)错误率,可能使患者暴露于无效或有害治疗,或错失有益治疗。
- 降低阈值至0.005可减少假阳性率,但增加假阴性率,且未解决可重复性危机。
- 显著性阈值的选择应视情境而定,受研究设计、样本量、先验证据和效应量等因素影响。
- 采用平衡I型和II型错误的灵活α水平可能提升科学有效性和伦理完整性。
临床怎么用
- 原文未据此给出推荐。
局限性
- 原文未报告局限性。
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
The continued use of a fixed p value significance threshold, such as p < 0.05, in randomized clinical trials has long been debated. Although such thresholds are conventional, they may not always align with the ethical principle of beneficence, which emphasizes minimizing harm and maximizing benefit. Reliance on a single, fixed α level increases the risk of both false-positive (type I) and false-negative (type II) error rates, thereby either exposing patients to ineffective or harmful treatments or withholding beneficial therapies. While lowering the threshold to values like 0.005 may reduce the false-positive rate, it also increases the false-negative rate and does not resolve the reproducibility crisis. Similar to diagnostic testing, the selection of a significance threshold should be context-dependent, influenced by study design, sample size, prior evidence, and effect size, among other things. Employing a flexible α level that balances type I and type II errors may enhance both scientific validity and ethical integrity. Thus, rigid adherence to fixed α levels, no matter how stringent, may be methodologically limiting and ethically questionable.
引用
- Habibzadeh F. Uncertain use of a fixed p value significance threshold in randomized clinical trials. Trials. 2026. PMID: 41501859. doi:10.1186/s13063-026-09438-4
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。