Searching for systematic reviews: How easy are they to identify in biomedical databases?




Poster session 2 Thursday: Evidence synthesis - methods / improving conduct and reporting


Thursday 14 September 2017 - 12:30 to 14:00


All authors in correct order:

Stirk L1, de Kock S1, Noake C1, Misso K1, Duffy S1, Deshpande S1, Kleijnen J1
1 Kleijnen Systematic Reviews Ltd, United Kingdom
Presenting author and contact person

Presenting author:

Lisa Stirk

Contact person:

Abstract text
Background: Quick and accurate identification of systematic reviews (SRs) and meta-analyses (MAs) has become increasingly important, both with the growing number of ‘umbrella’ reviews, and the dramatic increase in SR publication. With funding for medical research already at a premium, it is important that money is not spent on research where previous SRs already exist.

Objectives: PRISMA guidelines state that SRs and MAs should be identified as such in both title and abstract. We used the KSR Evidence database to establish how many SRs/MAs adhere to these guidelines and how easily they can be retrieved in primary databases of biomedical literature.

Methods: Seven databases are searched for KSR Evidence, using a sensitive SR search filter created by KSR Information Specialists. There are currently >30 000 SRs/MAs on KSR Evidence, identified from screening over 135 000 references (after deduplication). This dataset is used to analyse how SRs/MAs are identified in the literature.

Results: Numerous methodological terms are used to describe SRs/MAs, many of which do not adhere to PRISMA guidelines. 12% of the studies claiming to use PRISMA do not have SR/MA in their title. SRs/MAs often do not clearly describe their methods in the title and abstract, and some can only be found using a sensitive search incorporating a wide range of search terms, which results in a substantially increased screening burden. Cochrane reviews do not include study design information in the title field.

Conclusions: SRs and MAs are clearly identified as such in the title of around 70% of relevant studies (Figure 1). Many other methodological terms are used in title and abstract, and inadequate description of methods is employed. This leads to inconsistency in the literature, makes it hard to locate relevant studies without the use of specialist databases or search filters, and hinders comprehensive indexing by database producers.