Article type
Year
Abstract
Background: For systematic reviews to reach balanced conclusions they need to include information about harms as well as benefits. The Adverse Effects Subgroup of the Cochrane Non-randomised Studies Methods Group has produced preliminary guidance on including adverse effects in Cochrane reviews, but evidence is sparse. We investigated current methods used to incorporate adverse effects in published reviews. Our first step was to retrieve example reviews published from 1994 onwards.
Objectives: To identify systematic reviews in which the primary outcome was adverse effects.
Methods: Search strategies were developed to search the Cochrane Database of Systematic Reviews (CDSR) and the Database of Abstracts of Reviews of Effects (DARE). CDSR was searched via the title, MeSH subheadings and the terms for adverse effects within 20 words proximity of the word "objectives" in the abstract. DARE in the Cochrane Library was searched using floating subheadings and CRD's DARE interface was searched using synonyms for "adverse effects" in the title and outcomes assessed fields. All provisional abstracts in DARE were included.
Results: The Cochrane Library search yielded 989 CDSR records and 749 DARE records. CRD's DARE interface yielded 1095 provisional abstracts, 977 full abstracts and 52 Cochrane review records. After de-duplication two researchers independently screened 3539 unique records and 301 were potentially eligible. There were problems searching each interface. The outcomes assessed field in DARE and objectives field in Cochrane reviews cannot be searched in the Cochrane Library interface. Subheadings (e.g. adverse-effects) cannot be searched in CRD's DARE interface, and provisional records in DARE don't yet have an abstract. Less than half the relevant papers contained MeSH terms or subheadings indicating that the review included adverse effects. Reviews that examined specific adverse effects, such as nausea, could not be retrieved by textwords for synonyms of "adverse effects".
Conclusions: Searching for systematic reviews solely about adverse effects is problematic despite the existence of databases of systematic reviews. This suggests that it will be even more difficult to search for systematic reviews reporting adverse effects as a secondary outcome. Database publishers and indexers should consider ways of identifying these studies more easily.
Objectives: To identify systematic reviews in which the primary outcome was adverse effects.
Methods: Search strategies were developed to search the Cochrane Database of Systematic Reviews (CDSR) and the Database of Abstracts of Reviews of Effects (DARE). CDSR was searched via the title, MeSH subheadings and the terms for adverse effects within 20 words proximity of the word "objectives" in the abstract. DARE in the Cochrane Library was searched using floating subheadings and CRD's DARE interface was searched using synonyms for "adverse effects" in the title and outcomes assessed fields. All provisional abstracts in DARE were included.
Results: The Cochrane Library search yielded 989 CDSR records and 749 DARE records. CRD's DARE interface yielded 1095 provisional abstracts, 977 full abstracts and 52 Cochrane review records. After de-duplication two researchers independently screened 3539 unique records and 301 were potentially eligible. There were problems searching each interface. The outcomes assessed field in DARE and objectives field in Cochrane reviews cannot be searched in the Cochrane Library interface. Subheadings (e.g. adverse-effects) cannot be searched in CRD's DARE interface, and provisional records in DARE don't yet have an abstract. Less than half the relevant papers contained MeSH terms or subheadings indicating that the review included adverse effects. Reviews that examined specific adverse effects, such as nausea, could not be retrieved by textwords for synonyms of "adverse effects".
Conclusions: Searching for systematic reviews solely about adverse effects is problematic despite the existence of databases of systematic reviews. This suggests that it will be even more difficult to search for systematic reviews reporting adverse effects as a secondary outcome. Database publishers and indexers should consider ways of identifying these studies more easily.
PDF