完全顺序约束的多臂多阶段临床试验设计
针对含两个或以上活性治疗组与一个对照组的试验,在假定活性组疗效相对于对照存在顺序时,提出利用该顺序计算检验统计量并设定检验顺序的设计;该设计在保持1型错误率强控制下所需样本量小于各活性组独立与对照比较的设计,单阶段和两阶段试验样本量至少降低20%。
要点
- 利用活性组疗效顺序可减少多臂多阶段试验所需样本量。
- 顺序约束设计在保持1型错误率强控制下,单阶段和两阶段试验样本量至少降低20%。
- 该设计适用于以识别所有有前景治疗为把握度的情形。
核心结论
- 当研究以识别所有有前景治疗为把握度时,利用疗效顺序计算检验统计量并设定检验顺序的设计所需样本量小于各活性组独立与对照比较的设计。
研究设计与样本
- 考虑含两个或以上活性治疗组加一个对照组的的多臂试验,假定活性组疗效相对于对照存在顺序。
主要发现
- 在所考虑的设定下,单阶段试验和两阶段试验的样本量至少降低20%。
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
We consider a multi-arm trial with two or more active treatments plus a control where it is reasonable to assume an order for the treatment effects of the active arms compared to control. For example, the arms could be a high dose and low dose of a new drug and a placebo. The objective of the trial is to compare each active arm to control while maintaining strong control of the type 1 error rate. We show that when the study is powered to identify all promising treatments, a design that uses the order of the treatment effects to calculate the test statistic and to set the order of testing requires a smaller sample size than a design where each active arm is tested against the control arm independently. Under the considered settings, the sample size for a single-stage trial and a two-stage trial was reduced by at least 20%.
引用
- Kanapka L, Ivanova A. Fully order restricted multi-arm multi-stage clinical trial design. Stat Med. 2023. PMID: 37190881. doi:10.1002/sim.9767
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。