中文摘要：

目的比较11种K层四格表OR值的一致性检验方法的优劣，并为其应用提供依据。方法采用MonteCarlo模拟方法，考虑第i层样本量Ni、ORi值、对照组暴露概率π0i和层数K等四种参数的不同组合，使用SAS9．2 软件编程，检验水准取值为0．05，比较各个检验方法的I型错误以及检验效能。结果与结论Zelen精确概率法过于保守；Mantel—Haenszel方法由于计算的便捷性，在样本量以及层数较大时，可以视为最佳的选择；Breslow—Day—Tarone法比Breslow—Day法更为保守，因此本研究不建议对Breslow—Day法进行校正；近似非条件方法与近似条件方法相比，前者优于后者；加权最小二乘方法相对于Mantel—Haenszel以及近似非条件方法来说偏于保守，但是当样本量很大时，由于计算相对简单，因此，被广泛使用于Meta分析中；稀疏数据检验方法不提倡使用。

英文摘要：

Objective To compare the 11 homogeneity test methods of the odds ratio in K× 2 × 2 tables. Methods Data with differ- ent setting of parameters N （ sample size）, OR, π0 （ exposure rate in con- trol） of the ith stratum and K （the stratum number） were generated using Monte Carlo in SAS9.2OR. The type one error and power of the 11 meth- ods were compared. Results ＆Conclusion Zelen exact test is too conservative;Mantel-Haenszel methods should be viewed as the best choice when the sample size and K are large because of its facility in computation; In the study Breslow-Day test should not be corrected because Breslow- Day-Tarone test is more conservative than Breslow-Day test;The asymptot- ic unconditional methods perform better than the asymptotic conditional methods;Although the weighted least square method is more conservative than Mantel-Haenszel methods and asymptotic unconditional methods, it is widely used in meta-analysis because of its facility in computation when the sample size is large;The sparse data test methods should be discarded.