The usefulness of different information sources for retrieving adverse effects data for a systematic review

Article type
Authors
Golder S1, Loke Y1
1Adverse Effects Methods Group, UK
Abstract
Background: Authors of systematic reviews tend to focus on searching MEDLINE and reference checking for identifying information on adverse effects. However, MEDLINE may not yield the most information on adverse effects.

Objectives: To determine the relative value and contribution of searching different sources to identify adverse effects data.

Methods: A systematic review of thiazolidinedione-related fractures in patients with type 2 diabetes mellitus was used as a case study. Searches were conducted in over 20 different sources including MEDLINE, EMBASE, Derwent Drug File, selected Internet sources and reference texts. A record was made for each relevant reference noting whether it was retrieved with a search for thiazolidinediones and fractures and whether it was indexed/available in each source at the time of searching. A record was made of the sensitivity, precision, and number needed to read from searching each source and from different combinations of sources.

Results: The review included 58 relevant references. The highest number of relevant references (35) were retrieved from Science Citation Index (SCI), followed by BIOSIS Previews (27) and EMBASE (23). The precision of the searches varied from 0.88% (Scirus) to 41.67% (CENTRAL). With the search strategies used, the minimum combination of sources required to retrieve all references was: the GSK website, Science Citation Index (SCI), EMBASE, BIOSIS Previews, British Library Direct, Medscape DrugInfo, handsearching and reference checking, AHFS First, and Thomson Reuters Integrity or Conference Papers Index (CPI). However, high numbers needed to read were required for BIOSIS Previews and Medscape DrugInfo even when searched after other sources with duplicate records removed.

Conclusions: In order to identify all the relevant references for this case study a number of different sources needed to be searched. The minimum combination of sources required to identify all the relevant references did not include MEDLINE.