期中分析的条件把握度及样本含量再估计  

The Conditional Power of Interim Analysis and Sample Size Re-estimation

作　　者：陈建平[1] 魏永越[1] 陈峰[1] 于浩[1] 

机构地区：[1]南京医科大学流行病与卫生统计学系,210029

出　　处：《中国卫生统计》2010年第4期361-363,共3页Chinese Journal of Health Statistics

基　　金：江苏省自然科学基金(BK2008449)

摘　　要：目的介绍临床试验期中分析时条件把握度的计算原理及样本含量再估计方法。方法根据信息时间,构造B统计量,运用布朗运动理论,估计条件把握度,并据此采用二分搜索法估计后一阶段所需样本含量。以正态分布资料为例,介绍计算过程。结果如果试验期初对参数的估计不准确,可以根据试验中期数据,计算条件把握度。如果条件把握度较低,则通过调整样本含量可以达到期望把握度;如果条件把握度较高,则可以提前结束临床试验。结论该方法可以在试验过程中有效监测检验效能,在不违背临床试验方案的前提下,通过调整样本含量,保证临床试验达到预期的检验效能。Objective To introduce the principle of conditional power of interim analysis and the method of re-estimating sample size in response to the data being collected in clinical trials.Methods The conditional power of rejecting a null hypotheses,defined over information time,is a Brownian motion,and can be calculated as a conditional probability under the Brownian motion model.The sample size required in the second stage can also be re-estimated based on the conditional power.The method was illustrated by a real normally distributed data.Results If we have limited information about the underlying parameters before clinical trial,the calculated sample size would be inaccurate.We can use the data in the existing study,to reassess the parameters and,of course,the sample size required in next stage.Conclusion The method introduced in the paper can guarantee the expected power of test by re-estimating sample size based on the conditional power.

关 键 词：临床试验 期中分析 条件把握度 样本含量 B值 

分 类 号：R195[医药卫生—卫生统计学]