Common errors in meta-analysis – lessons and learning from the Cochrane review-screening programme

Article type
Authors
Bickerdike L1, Cumpston M2, Livingstone N1, Opiyo N3, Sambunjak D4
1Cochrane Editorial Unit
2Cochrane Central Executive Team
3Cochrane
4Learning & Support Department, Cochrane Central Executive Team
Abstract
Objectives: The objectives of this workshop are to highlight common statistical errors made in Cochrane systematic reviews, and to provide practical, hands-on guidance to help authors and editors address these errors.

Description: The Cochrane Editorial Unit (CEU) quality assurance team has been screening new reviews against key MECIR standards since September 2013. During the course of this work, it has become notable that many of the same errors frequently occur in the Data and Analysis section of the review. This can have a serious impact, as even the smallest statistical error can change the interpretation of the results. Examples of these common errors include data extraction errors, Unit of Analyses errors, and inappropriate methods of performing Subgroup Analyses. In addition, Cochrane’s Learning and Support Department has worked in partnership with CEU to develop a suite of online learning resources to support editors in identifying and addressing common errors.
The purpose of this workshop is to improve review authors and editor’s awareness of these errors, thus helping them to identify, rectify, and ultimately avoid making these errors. The workshop will begin with a brief PowerPoint presentation, providing an overview of common errors and introducing the accompanying online materials. Following this, the attendees will work in small groups with the facilitators to identify errors in some real-life examples, and discuss the best way to rectify the issues.