Searches for randomized controlled trials of drugs in MEDLINE and EMBASE using only generic drug names compared with searches applied in current practice in systematic reviews.

BACKGROUND

It is unclear which terms should be included in bibliographic searches for randomized controlled trials (RCTs) of drugs, and identifying relevant drug terms can be extremely laborious. The aim of our analysis was to determine whether a bibliographic search using only the generic drug name produces sufficient results for the generation of informative systematic reviews (SRs).

METHODS

We conducted a retrospective analysis of relevant references included in SRs of drugs. We determined the proportion of references identified by a simplified technique consisting of a systematic search for RCTs in MEDLINE and EMBASE via the search interface Ovid and using only the truncated generic drug name in all search fields. We calculated aggregated sensitivity and also evaluated the unidentified references.

RESULTS

Forty-eight SRs contained 873 primary publications, of which we found 829 in MEDLINE and 757 in EMBASE ("gold standard"). The simplified search identified 823 of the 829 MEDLINE references (sensitivity 99.3%) and 754 of the 757 EMBASE references (99.6%). Ultimately, only three references could not be found by additional searches.

CONCLUSION

Our findings indicate that when searching for RCTs of drugs in MEDLINE and EMBASE, a search using the truncated generic drug name in all fields produces sufficient results.