Ishida Haku
Department of Medical Informatics and Decision Sciences School of Medicine, Yamaguchi University, Ube, 755-8505.
Rinsho Byori. 2003 Jul;51(7):682-90.
Searching for relevant literature is an important process when carrying out a systematic review. Although general-purpose literature databases such as MEDLINE contain a huge number of articles, only a small fraction of these report evidence applicable to a systematic review. Development of an optimal search strategy giving priority to minimizing the number of important articles not retrieved is thus desirable. The retrieval of relevant studies cited in literature databases can be substantially enhanced by selected combinations of indexing terms and textwords. In order to improve the likelihood of retrieving studies relevant to a specific clinical question, precise database search skills are required: creation of an explicit, well formulated question using PI (E) CO, and development of a list of optimal indexing terms such as MeSH and textwords. However, even a search strategy yielding maximum sensitivity could omit some relevant articles. Consequently, hand-searching for articles in key or unindexed journals is still necessary to identify all relevant studies.
在进行系统评价时，检索相关文献是一个重要的过程。虽然像MEDLINE这样的通用文献数据库包含大量文章，但其中只有一小部分报告了适用于系统评价的证据。因此，开发一种优先减少未检索到的重要文章数量的最佳检索策略是很有必要的。通过索引词和文本词的选定组合，可以显著提高文献数据库中引用的相关研究的检索率。为了提高检索与特定临床问题相关研究的可能性，需要精确的数据库搜索技巧：使用PI（E）CO创建一个明确、表述清晰的问题，并制定一份最佳索引词列表，如医学主题词（MeSH）和文本词。然而，即使是产生最大敏感性的检索策略也可能遗漏一些相关文章。因此，为了识别所有相关研究，手工检索关键或未索引期刊中的文章仍然是必要的。
