Four systematic reviews per day: the epidemiology of research synthesis

Article type
Authors
Bastian H, Glasziou P
Abstract
Background: Systematic reviewing converts results from trials into reliable and usable knowledge. The three major publishing platforms for systematic reviews are the academic literature, Cochrane Collaboration and health technology assessment (HTA) agencies.
Objectives:

1. To identify major sources and trends in the production and publication of systematic reviews.
2. To determine what proportion of systematic reviews are published by the Cochrane Collaboration.

Methods: From 1995 the Centre for Reviews and Dissemination (CRD) has assessed all published reviews to identify those which meet minimum standards. We studied systematic reviews published between 1996 and 2003. We determined (i) the numbers of new and updated reviews (Cochrane Secretariat) (ii) the numbers of HTA reports and 'other' quality systematic reviews (CRD), and the country of origin of published meta-analyses (PubMed).

Results: From 1996 to 2003, 8,633 reviews were identified, reaching four reviews per day in 2003. About one in five systematic reviews are published by the Cochrane Collaboration, with the remainder split about evenly between HTA reports and the academic literature (see Figure). HTA production has risen twice as rapidly as the production of Cochrane reviews. Cochrane reviews are on unique subjects, while there is an unmeasured but potentially large duplication of topics within the other two sources. The United Kingdom is by far the greatest producer of Cochrane reviews (60%), while the USA is by far the greatest producer of meta-analyses published in the academic literature (at least 30%).
Only Cochrane reviews are intended to be regularly updated. Since 2001 the rate of substantively updated reviews has plateaued at around around 200 per year, compared to a needed rate of around 1,000 per year to achieve a 2-year update time.

Conclusions: Duplication of effort in systematic reviews is widespread. Cochrane currently produces around one fifth of all systematic reviews, with more than half coming from the UK. The Cochrane Collaboration's current system for maintaining reviews is insufficient to meet its goal of 2-year updates. Strategies to streamline the updating process are needed, and some suggestions will be proposed.