常用統合分析軟體之比較：以JAMA某文獻為例

Comparison of Commonly Used Meta-Analysis Software: An Example of JAMA Article

目的：國內常使用的統合分析（Meta-Analysis, MA）軟體，雖皆有使用手冊介紹，仍不易執行，亦少有其較簡化的使用介紹，特別是Review Manager（RevMan），其新檔案的設定頗不方便，因此我們擬設定常用軟體模版，方便研究者直接下載使用，且舉一實例做較常用軟體間的比較。方法：關鍵字為“統合分析（Meta-Analysis）軟體”或“系統回顧（Systematic Review）軟體”或“森林圖（Forest Plot）”，在Google搜尋引擎尋找較常被使用的MA軟體。並搜尋PubMed中適合的文獻實例做計算及比較。結果：可找到13種常見MA軟體，其中以Comprehensive Meta-Analysis（CMA）、RevMan及Stata較常見。以RevMan及CMA之“Events／Total＂輸入格式計算JAMA某文獻資料，二者結果完全相同，且亦同於此文獻之結果。使用CMA＂Computed Odds Ratio and Confidence Limits＂輸入格式計算，除權重稍有差異外，其餘資料亦同於文獻。然二種軟體之設定較麻煩，建議使用舊檔案或本文提供之模版直接下載使用。結論：以RevMan及CMA之“Events／Total＂輸入格式，分析JAMA某文獻，結果二者完全相同，但CMA尚可使用輸入格式為“Computed Odds Ratio＂及併用不同輸入格式以計算MA。

Purposes: Although methods for operating meta-analysis (MA) software commonly used in Taiwan have been introduced in instruction manuals, they are still difficult to implement. In addition, few simple introductory manuals exist, especially for Review Manager (RevMan), whose settings for new files are rather inconvenient. Thus, our aim is to establish a common software template that is convenient for researchers to download and use directly, as well as to provide an example comparing it to more commonly used software. Methods: Google's search engine was used to find commonly used MA software, by employing the keywords ＂meta-analysis software,＂ ＂systematic review software,＂ or ＂forest plot.＂ In addition, PubMed was used to search for appropriate literature examples for calculations and comparisons. Results: Thirteen common examples of MA software were found, with Comprehensive Meta-Analysis (CMA), RevMan, and Stata being the most common. The ＂Events/Total＂ input format of RevMan and CMA were used to calculate data from the Journal of the American Medical Association (JAMA) literature; the results were completely identical to each other and similar to the results of this paper. When the CMA's ＂computed odds ratio and confidence limits＂ input format was used for calculation, all of the data was also found to be the same as in the literature, except for a slight difference in weighting. However, the settings of the two software programs were cumbersome, so we recommend using the old files or the templates provided in this paper to download them directly for use. Conclusions: The ＂Events/Total＂ input formats of RevMan and CMA were used to analyze JAMA literature, and the results were completely identical to each other. However, the ＂computed odds ratio＂ and other input formats can still be used for MA calculations with CMA.