多中心临床随机对照试验的Meta分析

A Meta-analysis in multi-center random controlled clinical trials

目的探讨解决多中心临床随机对照试验中心效应差异的统计分析问题。方法以两项多中心临床随机对照试验数据为例,运用协方差分析及Meta分析。结果协方差分析中心间效应值差异均有统计学意义,行Meta分析,项目1异质性检验差异无统计学意义(P>0.05),采用固定效应模型分析合并效应值组间差异有统计学意义(P<0.05)。项目2异质性检验差异有统计学意义(P<0.05),采用随机效应模型分析合并效应值组间差异无统计学意义(P>0.05)。结论多中心临床随机对照试验研究中,如果存在中心间以及中心与分组间交互效应差异有统计学意义时,可根据Meta分析异质性检验结果选择适合的模型进行合并效应值的组间比较,如果协方差与Meta分析结果不一致时,建议选择Meta分析的结果较为稳妥。

Objective To study the center effect discrepancy in the multi-center clinical trials. Methods Two groups of data collected from the multi-center clinical trials were used.Data were processed by covariance analysis and Meta-analysis. Results In the covariance analysis,the discrepancy of the center effect values indicated statistical significance.Through Meta-analysis on fixed effect model,the discrepancy in one heterogeneity test showed no statistical significance( P> 0.05 ) while the inter-group discrepancy of the merged effect values drawn from analysis based on fixed effect model having statistical significance( P< 0.05 ).In the random effect model,the discrepancy in one heterogeneity test showed statistical significance( P< 0.05 ) while the inter-group discrepancy of the merged effect values drawn from analysis based on random effect model having no statistical significance( P> 0.05 ). Conclusions Studies on multi-center random controlled clinical trials,when statistical significance was found in the interaction discrepancy between the inter-center and the center-group relation,the merged effect values should be compared and analyzed by an appropriate statistic model based on the heterogeneous test results from the Meta-analysis.However, if the result from covariance analysis and the one from Meta-analysis did not agree to each other, the results drawn from the Meta-analysis were reliable.