Background:

Completeness of information is considered to be essential for conducting a meta-analysis and the use of multiple databases might be necessary. However, in meta-analysis of treatment studies, the contribution of databases beyond MEDLINE, such as EMBASE has been questioned.

Objectives:

The objective of this review was to evaluate the incremental contribution of searching EMBASE in published meta-analyses of diagnostic studies.

Methods:

We searched MEDLINE for published English language diagnostic test performance meta-analyses. We reviewed each qualifying meta-analysis and extracted information about databases used in the literature search. For each of the meta-analyses that searched EMBASE, we examined whether the primary studies used were identifiable in MEDLINE. We searched EMBASE for studies not found in MEDLINE.

Results:

We identified 236 meta-analyses of diagnostic tests that included 5855 primary studies. Forty meta-analyses (16.9%) that reported searching EMBASE included 1176 primary studies. Five of these 40 meta-analyses did not provide a list of the included studies. Eight meta-analyses included 14 primary studies (1.2% of all primary studies) that were not indexed on MEDLINE. Four of these 8 meta-analyses included a total of 5 primary studies (0.4% of all primary studies) uniquely identified in EMBASE. None of the meta-analysis included more than 2 primary studies not indexed in MEDLINE.

Conclusions:

EMBASE contributed less than 1% of studies used in published meta-analyses we examined. Search of EMBASE does not appear to be important in the meta-analyses of diagnostic test performance.