CMH检验和Meta分析在临床试验中心效应分析的应用探讨

目的探讨比较多中心临床随机对照试验中心效应的统计分析方法。方法以一项多中心临床随机对照试验数据为例，运用CMH检验及Meta分析。结果CMH检验显示各中心间效应值一致性检验差异有统计学意义（P〈0．05）而Meta分析的异质性检验结果显示差异无统计学意义（χ^2=12．81，P=0．12），采用固定效应模型，合并后效应值组问差异有统计学意义（Z=3．35，P〈0．05）。结论多中心临床随机对照试验研究中，如果存在分中心组间疗效差异趋势不一致时，可选择Meta分析，再根据Meta分析异质性检验结果，选择适合的模型进行合并后效应值的组问比较，并结合CMH检验给出结论。

Application of CMH Test and Meta-analysis in Multi-center Randomized Controlled Trials

Objective To seek the statistical solution in the comparison of different effects from multi-center randomized controlled trials (RCTs). Methods The data collected from a multi-center RCT were used as the examples and processed by CMH test and meta-analysis. Results The result of CMH test indicated that the significant difference of the effect values existed among centers (P<0.05). While meta-analysis showed no significant difference (P>0.05) by heterogeneity test. However, when using fixed effect model, inter-group significant difference of merged effect values was observed (P<0.05). Conclusions In the clinical research based on the method of multi-center RCT, meta-analysis can be applied if the difference of inclination of the inter-group therapeutic effect is found among different centers. The proper mathematical model should be selected based on the result of heterogeneity test to merge and compare the effect values. The conclusions should be drawn from the results of both meta-analysis and CMH test.