Wang Dan, Mou Zhen-yun, Zhai Jun-xia, Zong Hong-xia, Zhao Xiao-dong
Department of Epidemiology and Biostatistics, Hebei Medical University, Shijiazhuang 050017, China.
Zhonghua Liu Xing Bing Xue Za Zhi. 2008 Jul;29(7):726-9.
To introduce the application of Stata software to heterogeneity test in meta-analysis. A data set was set up according to the example in the study, and the corresponding commands of the methods in Stata 9 software were applied to test the example. The methods used were Q-test and I2 statistic attached to the fixed effect model forest plot, H statistic and Galbraith plot. The existence of the heterogeneity among studies could be detected by Q-test and H statistic and the degree of the heterogeneity could be detected by I2 statistic. The outliers which were the sources of the heterogeneity could be spotted from the Galbraith plot. Heterogeneity test in meta-analysis can be completed by the four methods in Stata software simply and quickly. H and I2 statistics are more robust, and the outliers of the heterogeneity can be clearly seen in the Galbraith plot among the four methods.
介绍Stata软件在Meta分析异质性检验中的应用。根据研究中的示例建立数据集,并应用Stata 9软件中各方法的相应命令对示例进行检验。使用的方法有Q检验、固定效应模型森林图附带的I²统计量、H统计量和Galbraith图。通过Q检验和H统计量可检测研究间异质性的存在,通过I²统计量可检测异质性的程度。从Galbraith图中可找出作为异质性来源的离群值。Stata软件中的这四种方法可简单快速地完成Meta分析中的异质性检验。H统计量和I²统计量更稳健,在这四种方法中,Galbraith图能清晰显示异质性的离群值。