摘要
目的探讨两阶段整群随机化设计两组比较样本例数估计方法。方法通过估计组内相关系数,根据设计效应及群的大小估计样本例数。结果本文给出了计量和计数两种情况下的样本含量估计公式。结论在整群随机化设计中,数据由于隶属关系而存在聚集性,表现为至少两个水平的层次结构及组内相关性,因此在设计阶段需对其设计效应进行估计,进而估计其样本例数。
Objective Explore the method for the estimation of sample size in two stage cluster randomization tirals. Methods According to the design effect and the number of cluster to estimate the sample size. Results Provide the formula of sample size estimation for comparison of means and proportions. Conclusion In cluster randomization trials, there are correlation among data because of clustering, shows the intracorrelation and the hierarchical structure which should be two levels at least, so it is necessary to estimate the design effect at the design stage, based on it the get the final sample size.
出处
《中国卫生统计》
CSCD
北大核心
2008年第2期117-119,共3页
Chinese Journal of Health Statistics
基金
美国国立卫生研究院(NIH)资助
美国国立变态反应和传染病研究院(NIAID)资助
关键词
整群随机化
样本例数
组内相关性
设计效应
Cluster randomization
Sample size
Intracorre-lation
Design effect
