NIHR CLAHRC South West Peninsula (PenCLAHRC), University of Exeter Medical School, University of Exeter, Exeter, UK.
Res Synth Methods. 2018 Dec;9(4):579-586. doi: 10.1002/jrsm.1280. Epub 2017 Nov 27.
Qualitative research in dementia improves understanding of the experience of people affected by dementia. Searching databases for qualitative studies is problematic. Qualitative-specific search strategies might help with locating studies.
To examine the effectiveness (sensitivity and precision) of 5 qualitative strategies on locating qualitative research studies in dementia in 4 major databases (MEDLINE, EMBASE, PsycINFO, and CINAHL).
Qualitative dementia studies were checked for inclusion on MEDLINE, EMBASE, PsycINFO, and CINAHL. Five qualitative search strategies (subject headings, simple free-text terms, complex free-text terms, and 2 broad-based strategies) were tested for study retrieval. Specificity, precision and number needed to read were calculated.
Two hundred fourteen qualitative studies in dementia were included. PsycINFO and CINAHL held the most qualitative studies out the 4 databases studied (N = 171 and 166, respectively) and both held unique records (N = 14 and 7, respectively). The controlled vocabulary strategy in CINAHL returned 96% (N = 192) of studies held; by contrast, controlled vocabulary in PsycINFO returned 7% (N = 13) of studies held. The broad-based strategies returned more studies (93-99%) than the other free-text strategies (22-82%). Precision ranged from 0.061 to 0.004 resulting in a number needed to read to obtain 1 relevant study ranging from 16 (simple free-text search in CINAHL) to 239 (broad-based search in EMBASE).
Qualitative search strategies using 3 broad terms were more sensitive than long complex searches. The controlled vocabulary for qualitative research in CINAHL was particularly effective. Furthermore, results indicate that MEDLINE and EMBASE offer little benefit for locating qualitative dementia research if CINAHL and PSYCINFO are also searched.
痴呆症的定性研究可以提高对受痴呆症影响的人的体验的理解。 在数据库中搜索定性研究存在问题。 定性专用搜索策略可能有助于找到研究。
检查 5 种定性策略在 4 个主要数据库（MEDLINE、EMBASE、PsycINFO 和 CINAHL）中定位痴呆症定性研究的有效性（敏感性和精度）。
在 MEDLINE、EMBASE、PsycINFO 和 CINAHL 上检查定性痴呆症研究是否符合纳入标准。 测试了 5 种定性搜索策略（主题词、简单自由文本词、复杂自由文本词和 2 种广泛的策略）来检索研究。 计算了特异性、精度和需要阅读的数量。
共纳入 214 项痴呆症定性研究。 PsycINFO 和 CINAHL 是这 4 个数据库中包含最多定性研究的数据库（分别为 171 和 166 项），并且都包含独特的记录（分别为 14 和 7 项）。 CINAHL 中的受控词汇策略返回了 96%（192 项）的研究；相比之下，PsycINFO 中的受控词汇策略仅返回了 7%（13 项）的研究。 与其他自由文本策略（22-82%）相比，广泛的策略返回了更多的研究（93-99%）。 精度范围从 0.061 到 0.004，导致获得 1 篇相关研究的阅读次数从 16 次（CINAHL 中的简单自由文本搜索）到 239 次（EMBASE 中的广泛搜索）不等。
使用 3 个广泛术语的定性搜索策略比长而复杂的搜索更敏感。 CINAHL 中用于定性研究的受控词汇特别有效。 此外，如果还搜索了 MEDLINE 和 EMBASE，则表明 CINAHL 和 PSYCINFO 对于定位定性痴呆症研究几乎没有好处。
