Streiner D L
Department of Psychiatry, McMaster University, Hamilton, Ontario.
Can J Psychiatry. 1991 Jun;36(5):357-62. doi: 10.1177/070674379103600508.
Meta-analysis is a technique for summarizing the results of many studies which can allow reviewers in a particular field to arrive at a synthesis of the various findings in a more objective manner than relying on expert opinion and combine the results of these studies to arrive at an overall estimate of the magnitude of the effect of a treatment, to determine if the evidence favours its effectiveness. The technique consists of determining an effect size or odds ratio for each study and combining them, taking into account the fact that studies with large samples should exert more influence than those with small samples. However, the procedure is not mechanical. The reviewer must make a series of decisions regarding the studies to be included or excluded, the weight to place on the methodological rigour of the trials, the results from each study to be used, how to handle differences in the magnitude of the results across studies, how to derive the best overall estimate of the effect across studies, and how to determine the precision of that estimate. Each of these decisions can affect both the quality of the review and the nature of the findings.
荟萃分析是一种汇总多项研究结果的技术，它能让特定领域的综述者比单纯依靠专家意见更客观地综合各种研究结果，并将这些研究的结果结合起来，以得出对某种治疗效果大小的总体估计，从而确定证据是否支持其有效性。该技术包括为每项研究确定效应量或比值比，并在考虑到样本量大的研究应比样本量小的研究具有更大影响力这一事实的情况下，将它们合并起来。然而，这个过程并非机械的。综述者必须就纳入或排除的研究、给予试验方法严谨性的权重、要使用的每项研究的结果、如何处理不同研究结果大小的差异、如何得出各项研究效应的最佳总体估计以及如何确定该估计的精确度等做出一系列决策。这些决策中的每一个都可能影响综述的质量和研究结果的性质。
