网状Meta分析中敏感性分析与一致性分析在R软件中的实现.pdfVIP

2015, 15(1): 99102 MMetaeta R 1,2 1,2 1,2 1,2* 1. 1100012. 110001 Meta R R metafor combinat Meta Meta Meta R Meta Meta R WinBUGS Sensitivity and Homogeneity Analysis in Network Meta-analysis Using R Sot ware YANG Bo-wen1,2, CHEN Xin1,2, SUN Hao1,2, SHI Jing-pu1,2* 1. Department of Clinical Epidemiology, First Alienated Hospital, China Medical University, Shenyang 110001, China; 2. Evidence-Based Medicine Center of Liaoning Province, Shenyang 110001, China Abstract Objective To introduce sensitivity and homogeneity tests in network meta-analysis and its implementation in R software. Methods Using an example, we performed sensitivity analysis by comparing the random effect model with the fixed effect model. Homogeneity analysis was performed using metaphor package and combinat package in R software. Results The results of the two models were similar, and the data was steady. The results of homogeneity analysis showed that the confidential intervals in all loops were crossed over with blank value; and direct and indirect estimates of the effects in network meta-analysis were not significantly different, with good homogeneity. Conclusion Network meta-analysis is a kind of indirect comparison analysis method, and its sensitivity is especially important. The introduction of homogeneity makes network meta-analysis more accurate. Using R software for sensitivity and homogeneity analysis in network meta-analysis is a feasible method. Key words Network meta-analysis; R software; WinBUGS; Sensitivity; Homogeneity [1] Meta RCT Meta Meta Meta