临床试验析因设计不同研究目标及其样本量估计方法
Sample size estimation methods for different study goals of factorial de-sign in clinical trials
虞茜惠 1张曼婷 2刘玉秀 3巩浩雯 1许敏怡 2熊殷2
作者信息
- 1. 南方医科大学公共卫生学院生物统计学系 广州 510515;东部战区总医院重症医学科数据与统计室 南京 210002
- 2. 东部战区总医院重症医学科数据与统计室 南京 210002;南京医科大学公共卫生学院生物统计学系 南京 211166
- 3. 南方医科大学公共卫生学院生物统计学系 广州 510515;东部战区总医院重症医学科数据与统计室 南京 210002;南京医科大学公共卫生学院生物统计学系 南京 211166
- 折叠
摘要
目的:分析针对2×2 析因设计的临床试验不同研究目标及其样本量的估计方法.方法:析因设计的临床试验可以设定5 种研究目标,分析了针对各研究目标的样本量估计方法.选择一项 2×2 析因设计的临床试验实例,利用其假设、参数设定和数据结果,报告针对不同研究目标的样本量估计方法和估算结果.结果:临床试验析因设计针对不同的研究目标和分析策略(析因分析法、多臂分析法、Bonferroni校正法、三步分析法、Hochberg法、Hommel法),选择样本量估计方法.实例应用分析结果说明,如果在析因设计中仅以主效应推断为研究目标估计的样本量,不适用于其他研究目标,需要采用相适应的样本量估计方法,否则会存在把握度不足的问题.结论:对于析因设计,应结合研究目标选择适合的样本量估计方法.
Abstract
Aim:To analyze sample size estimation methods for different study goals of 2×2 factorial design in clinical trials.Methods:Factorial clinical trials can be set with 5 study goals,and the goal-oriented sample size estimation methods were analyzed.An actual clinical trial employing 2×2 factorial design were subjected,its assumptions,parameter settings,and data results were used to demonstrate the application of sample size estimation method for different study goals.Re-sults:Sample size estimation methods for factorial design should be chosen based on different study goals and analysis strat-egies,such as factorial analysis,multiarm analysis,Bonferroni correction procedure,three-step procedure,Hochberg proce-dure,and Hommel procedure.The analysis of the example application showed that estimating sample size solely based on the inference of main effects in a factorial trial was inadequate for other study goals.It is necessary to employ appropriate sample size estimation methods tailored to specific goals to avoid issues of insufficient statistical power.Conclusion:For factorial design,appropriate sample size estimation methods should be selected in accordance with the study goals.
关键词
临床试验/析因设计/研究目标/样本量估计/交互作用Key words
clinical trial/factorial design/study goal/sample size estimation/interaction引用本文复制引用
出版年
2024
国家科技期刊平台
NSTL
万方数据