Objective: To provide an introduction to the different forms of reporting biases and the methods that might be used to detect these biases.

Description: The results of a meta-analysis can be distorted if bias has been introduced into the review process when component studies are being located and selected. Accordingly, investigation for the presence, degree, and nature of bias is recommended as a routine part of the systematic review process. The 'Reporting Bias' Methods Group has now been established to undertake continuing research in this field. The workshop will largely draw upon the work of members of this Group. We will discuss potential biases that might be introduced into reviews, with an emphasis on reporting biases. Particular attention will be paid to publication bias (the publication or non-publication of research findings, depending on the nature and direction of the results), other reporting biases and the effects of trial quality. We will demonstrate and discuss graphical and statistical methods that can be used to detect or investigate bias. These methods include the funnel plot, 'trim and fill', the adjusted rank correlation method, the regression test, and meta-regression. The workshop will consist of Powerpoint presentations, and discussions with and among participants.

Target audience: Anyone within the Collaboration who conducts, assesses or reads systematic reviews containing meta-analyses. Although categorised as for beginners, people at all levels are welcome to update their skills or to raise issues with the presenter.

Style: Training workshop