自适应样本量设计的效率研究
该研究比较自适应样本量设计与成组序贯设计的效率,证明在相同最大样本量和拒绝边界下可构建一致更高效的自适应设计,其期望样本量在真实治疗差异范围内一致小于成组序贯设计,挑战了成组序贯设计更高效的常见观点。
要点
- 基于相同最大样本量和拒绝边界,可构建一致更高效的自适应样本量设计
- 自适应设计的期望样本量在真实治疗差异范围内一致小于成组序贯设计
- 自适应设计兼具良好统计性能与易实施性,更适合现代临床试验
核心结论
- 在相同最大样本量和拒绝边界下,自适应样本量设计的期望样本量一致小于成组序贯设计。
研究设计与样本
- 方法学研究,比较自适应样本量设计与成组序贯设计的效率。
主要发现
- 给定成组序贯设计,可构建基于相同最大样本量和拒绝边界的一致更高效自适应样本量设计。
- 在维持所需统计功效稳定的同时,自适应设计的期望样本量在真实治疗差异范围内一致小于成组序贯设计。
临床怎么用
- 自适应样本量设计因良好自适应性能、易实施等操作优势,更适用于现代临床试验。
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
Adaptive sample size designs, including group sequential designs, have been used as alternatives to fixed sample size designs to achieve more robust statistical power and better trial efficiency. This work investigates the efficiency of adaptive sample size designs as compared to group sequential designs. We show that given a group sequential design, a uniformly more efficient adaptive sample size design based on the same maximum sample size and rejection boundary can be constructed. While maintaining stable statistical power at the required level, the expected sample size of the obtained adaptive sample size design is uniformly smaller than that of the group sequential design with respect to a range of the true treatment difference. The finding provides further insights into the efficiency of adaptive sample size designs and challenges the popular belief of better efficiency associated with group sequential designs. Good adaptive performance plus easy implementation and other desirable operational features make adaptive sample size designs more attractive and applicable to modern clinical trials.
引用
- Cui L, Zhang L. On the efficiency of adaptive sample size design. Stat Med. 2018. PMID: 30450621. doi:10.1002/sim.8034
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。