Hide and seek! Looking for studies included and (missing) links to other versions

Article type
Authors
Johansen M, Vist G, Jamtvedt G, von Mehren Sæterdal I, Merete Reinar L, Nordheim L
Abstract
Background: To study the 'life cycle' of Cochrane reviews, we wanted to examine those reviews going from previously including none to currently including some studies. We investigated the Cochrane Database of Systematic Reviews (CDSR), the online version of the Cochrane Library Issue 1 2007. Extracting information about the number of included studies in current and previous versions became more challenging than anticipated. The Cochrane Handbook has since version 4.0 (July 1999), given explicit recommendations on how to report included studies, and where this information should appear in reviews1. The abstract's main results section should begin with reporting the total number of trials and participants included. The recommendations give a hypothetical example: "Seventeen trials involving 689 people were included".
Objective: How valid is the Cochrane Library to identify information needed to draw the history of Cochrane reviews?
Methods: We extracted information from all the reviews in the CDSR, using the online version of the Cochrane Library Issue 1 2007 (n=3009). The Main results section of each abstract was read. If the section lacked information about included studies, or the information was phrased ambiguously, we checked the Description of studies section in the review's main body. When the description of studies gave insufficient information, the studies in the Characteristics of included studies table were counted. Links to Other Versions were followed, and the review's Cover sheet was checked to find when the review was first published. To see whether the review abstracts were in accordance with the explicit recommendation given in the Handbook on how and where to report the number of included studies, the following search was run on the library, limited to the abstract field: included near/20 "main results".
Results: The number of included studies was reported in a variety of ways, and was not always present in the abstract's Main results section. Our search retrieved 1516 of the 3009 reviews. Only 600 reviews had links to previous versions. Some links led only to the review protocol. Some Cover sheets had no information on when the review was first published.
Conclusions: Reporting of selected information, needed to draw the history of Cochrane reviews, was either hard to find, incomplete or absent in the Cochrane Database of Systematic Reviews.

References 1. Higgins JPT, Green S, editors. Guide to the contents of a protocol and review. Cochrane Handbook for Systematic Reviews of Interventions 4.2.6 [updated September 2006]; Section 3. In: The Cochrane Library, Issue 4, 2006. Chichester, UK: John Wiley & Sons, Ltd.