【摘要】 目的　提出用于多中心两组临床试验的样本量测定方法。方法　由Cochran检验或Mantel Haenszel检验反推得出所需总样本量 ,按层样本分数和组样本分数进行样本量的分配。结果　如此获得的样本量测定方法与检验方法一一匹配。该方法具有同质性 ,当只有一层时还原为同质假设下的简化正态法或同质假设下的简化正态法加 1。针对Cochran检验的设计可为等层或非等层设计 ,而针对Mantel Haenszel检验的只能是等层设计。结论　该方法可用于多中心两组临床试验方案的设计。附有工作实例描述设计过程更多还原

【Abstract】 Objective This paper concerns the determination of the required sample size for multi-center two-group clinical trials. Methods The required total sample size is derived inversely from the Cochran test or the Mantel-Haenszel test and allocated into each stratum and each group in the light of the relevant sample fractions. Results The procedures are certainly matched with the tests respectively. They possess the property of homogeneity and reduce to the procedure of simple normal assuming homogeneity or that plus one when there is only one stratum. The procedure derived from the Cochran test can be used for the designs with equal-or unequal-sized strata but the procedure derived from the Mantel-Haenszel test only for the designs with equal-sized strata. Conclusion The procedures proposed can be applied to planning multi-center two-group clinical trials. A worked example illustrates the planning process.更多还原