Publication type: Article
Author(s): PC Burkner, P Doebler
Publication: Statistics in Medicine
Year published: 2014
Why this Article: Very often once numerous studies have been published on a particular topic a meta-analysis is conducted in order to generate a summary effect estimate from all the published evidence. However, not all studies conducted are published, and if the non-published studies are not published for systematic reasons, e.g. non-significant results, publication bias may arise. Publication bias takes on a unique twist when it comes to meta-analysis of diagnostic studies, and this article takes an in-depth look at the methods to detect publication bias in these settings.