Department of Clinical, Educational and Health Psychology, University College London, London WC1E 7HB, UK.
Department of Clinical, Educational and Health Psychology, University College London, London WC1E 7HB, UK.
J Clin Epidemiol. 2019 Aug;112:12-19. doi: 10.1016/j.jclinepi.2019.04.002. Epub 2019 Apr 13.
The authors were becoming increasingly aware of studies reporting randomized controlled trial (RCT), which reported trial phase but did not mention study design or randomization in the title or abstract. The objective of this study was to determine if established RCT literature search filters should include terms for trial phase.
This study is a case study. A search filter for trial phase (the P3 filter) was developed, and its sensitivity, efficiency, and value were determined when compared with two established RCT literature search filters (The Cochrane Highly Sensitive Search Strategies [HSSS] and the Royle and Waugh Brief RCT Search Strategy [BRSS]) in the year 2015-improved sensitivity was determined where the P3 filter identified studies missed by either of the established filters; efficiency was determined by the number needed to read; and the Cochrane risk of bias tool was used to determine study quality as a proxy for value.
Both established filters missed studies. The HSSS missed one RCT and four follow-up RCT studies. The BRSS missed one RCT and five follow-up RCT studies. Study quality was unclear.
Established RCT literature search filters may miss studies where trial phase is reported instead of terms for study design or randomization. The P3 filter can be incorporated to improve sensitivity.
作者越来越意识到有研究报告随机对照试验（RCT），这些研究报告了试验阶段，但在标题或摘要中没有提到研究设计或随机分组。本研究的目的是确定既定的 RCT 文献检索过滤器是否应包括试验阶段的术语。
本研究为案例研究。开发了一个用于试验阶段的搜索过滤器（P3 过滤器），并在 2015 年将其与两个既定的 RCT 文献检索过滤器（Cochrane 高度敏感搜索策略[HSSS]和 Royle 和 Waugh 简短 RCT 搜索策略[BRSS]）进行比较，以确定其灵敏度、效率和价值。灵敏度通过 P3 过滤器确定了两个既定过滤器漏掉的研究来确定；效率通过需要阅读的数量来确定；Cochrane 偏倚风险工具用于确定研究质量作为价值的替代指标。
两个既定的过滤器都漏掉了一些研究。HSSS 漏掉了一个 RCT 和四个后续 RCT 研究。BRSS 漏掉了一个 RCT 和五个后续 RCT 研究。研究质量不明确。
既定的 RCT 文献检索过滤器可能会漏掉仅报告试验阶段而不是研究设计或随机分组术语的研究。可以纳入 P3 过滤器以提高灵敏度。
