基于β分解的等效性试验样本含量估计

Estimation of sample size for equivalence clinical trials by decomposing β

目的:两组平行阳性对照的等效性试验中时常会遇到两总体均数之差不为0(δ≠0)的情形,而目前的样本含量估计公式并未很好考虑不同δ时β的分解问题,因而造成样本含量估计的偏差,无法达到理想的把握度。本文探讨服从正态分布连续性变量为主要指标的等效性试验样本含量估计的通用方法并验证其正确性。方法:根据等效性试验的统计推断原理,基于β分解和非中心t分布理论,对样本含量计算公式进行了理论推导,并计算出不同参数设置下的样本含量,借助Monte-Carlo模拟方法对相应的样本含量逐个进行把握度模拟验证。结果:不同参数设置下模拟获得的把握度与事先设定的目标把握度水平均能很好地吻合,验证了样本含量估计方法的正确性。结论:本文给出的样本含量估计方法可以通用于主要指标为正态分布连续性变量的等效性试验。

AIM： Because the sample size estimations currently used have considered inappropriately the influence of the population mean differences of two groups in equivalence clinical trials,the actual powers could not reach the objective level of power predetermined. This paper is to explore the sample size estimation of equivalence clinical trials which the primary variable follow normal distribution. METHODS： The formulae of sample size estimation were derived based on the statistical inference principle of equivalence clinical trials incorporated with the decomposition of β and non-central t distribution theory. According to the formula,the sample sizes were estimated under different parameter settings. Monte-Carlo simulations were performed to obtain the corresponding powers respectively.RESULTS： The achieved power of Monte-Carlo simulation could coincide with the pre-determined level of power well. It showed that the method possessed correctness to estimate the sample size. CONCLUSION： The method of sample size estimation could be used commonly in equivalence clinical trials with quantitative normal distribution variable.