Reviewing the meaning of ‘systematic review’
Reviewing the meaning of ‘systematic review’
复制标题
回顾“系统审查”的含义
DOI：
--
复制
发表时间：
2007
影响因子：
3.5
通讯作者：
J. Daly
中科院分区：
文献类型：
--
作者：
J. Lumley;J. Daly
在公共卫生这样的多学科领域，我们必须小心单词或短语的含义范围。在《华尔街日报》给我们造成困难的一个术语是“系统性综述”。对该术语的解释范围从常识性使用（作者在文献综述中或多或少地进行了系统性分析）到对文献中的信息进行相当技术性和形式化的综合。我们在此的目的是审查这些分歧，并提出一个折衷办法，其中强调审查的系统性。在意义范围的技术端是元分析。这就是它是怎么来的。在过去的二三十年里，进行的随机对照试验数量大幅增加。并非所有这些试验都提供了统计学上显著的结果，有时是因为样本量不足，并且同一程序的不同试验的结果可能相互矛盾。那么有两个选择。第一个是进行更大、更严格、因此也是最后的试验。另一种是使用统计技术进行荟萃分析，将各种随机对照试验的结果联合收割机结合起来，从而加强结论。累积荟萃分析提供了及时和可信的证据。进行这种荟萃分析的方法被清楚地阐明，并形成了科克伦协作组发表的大部分工作的基础。这些荟萃分析的一个重要方面是应该对文献进行全面检索，以便识别所有相关研究。根据研究方法的严谨性，必须有明确的标准来确定哪些内容被纳入和排除。由于荟萃分析涉及对原始研究数据的重新分析，因此可能需要联系作者以确认或提供其他数据。如果一些研究因为结果是负面的而没有发表，那么就引入了一种偏见，需要通过跟踪一个领域中所有已发表和未发表的研究来解决。显然，这些都是要求很高的研究，他们被认为是作出了重要的研究贡献。科克伦协作网的创始人Iain查尔默斯更喜欢“系统性综述”一词，不仅包括统计学荟萃分析，还包括通过系统收集、分类和分析一组解决同一问题的研究来减少偏倚的其他技术。当然，研究人员必须在方法论和创造性上开发出严格的方法来结合一组观察性研究的结果，或者结合使用不同方法的研究结果。然而，这是可以做到的，如果做得好，它是从业人员和决策者的良好证据来源。那么，从文献中选择关键论文进行描述并得出结论的文献综述的命运是什么？这些评论对于为研究论文提供背景的几个段落来说可能仍然足够。然而，它们不足以作为报告文献概述的独立论文，特别是如果文章引用了系统综述的概念。随着数据库检索相关文章的兴起，有时会产生大量相关研究的积累，但并非所有研究都是严格或完善的，作者必须说明并证明其检索方法和标准，以选择或排除综述中的文章。然后，这些独立的审查需要阐明用于综合文献并得出总体结论的过程。我们意识到，这种系统评价的观点提高了许多不习惯荟萃分析程序的研究人员的门槛。然而，这种更明确、更严格的方法的优点远远超过缺点。
In a multi-disciplinary field such as public health, we have to be careful about the range of meanings that can be attached to words or phrases. One term that is causing us difficulty at the Journal is ‘systematic review’. The interpretation of the term ranges from a common sense use, where authors have been more or less systematic in their review of the literature, to the use of a quite technical and formalised synthesis of information from the literature. Our aim here is to examine these differences and to suggest a compromise in which the emphasis is on the systematic nature of the review. At the technical end of the spectrum of meaning is meta-analysis. That is, how it came about. In the past two or three decades there has been a substantial increase in the number of randomised controlled trials that have been conducted. Not all of these trials have delivered statistically significant results, sometimes because the sample sizes are inadequate, and the results of different trials of the same procedure may be contradictory. There are then two choices. The first is to conduct a larger, more rigorous and, therefore, definitive trial. The other is to conduct a meta-analysis using statistical techniques to combine the results from various randomised controlled trials as they emerge, thus strengthening the conclusions. Cumulative meta-analysis provides timely and trustworthy evidence. The methods for conducting such meta-analyses are clearly spelled out and form the basis for much of the work published by the Cochrane Collaboration. An important facet of these metaanalyses is that there should be a comprehensive search of the literature, so that all relevant studies are identified. There have to be explicit criteria for what is included and excluded, based on the methodological rigour of the study. Since the meta-analysis involves a re-analysis of data from the original studies, authors may have to be contacted to confirm or provide additional data. If some studies are unpublished because the result is negative, that introduces a bias to be addressed by keeping track of all studies in a field, published and unpublished. Clearly these are demanding studies and they are recognised as making an important research contribution. Iain Chalmers, founder of the Cochrane Collaboration, preferred the term ‘systematic review’ to cover not only statistical meta-analyses but also other techniques for reducing bias by the systematic collection, categorisation and analysis of a set of studies addressing the same problem. Of course, researchers have to be methodologically and inventive to develop rigorous methods for combining the results of, say, a set of observational studies, or for combining the results of studies using different methods. Nevertheless, it can be done and when it is well done it is a source of good evidence for practitioners and policy makers. What then is the fate of the literature review that describes and draws conclusions from a selection of key papers from the literature? These reviews may still be adequate for the few paragraphs that provide background for a research paper. However, they are inadequate as stand-alone papers reporting an overview of the literature, especially if the article invokes the notion of a systematic review. With the rise of database searching for relevant articles, generating what is sometimes a vast accumulation of relevant studies, not all of them rigorous or well-developed, authors must specify and justify their search methods and criteria for selecting or excluding articles from the review. These stand-alone reviews then need to spell out the process used to synthesise the literature and to draw overall conclusions. We are aware that this view of the systematic review raises the bar for many researchers not accustomed to meta-analytic procedures. However, the advantages of this more explicit and, hopefully, rigorous approach well outweighs the disadvantages.