6.3.3.2 Searching MEDLINE and EMBASE: specific issues

Despite the fact that both MEDLINE and EMBASE have been searched systematically for reports of trials and that these reports of trials have been included in CENTRAL, as described in Sections 6.3.2.1 and 6.3.2.2, supplementary searches of both MEDLINE and EMBASE are recommended. Any such searches, however, should be undertaken in the knowledge of what searching has already been conducted to avoid duplication of effort.

 

Searching MEDLINE

There is a delay of some months between records being indexed in MEDLINE and appearing indexed as reports of trials in CENTRAL, since CENTRAL is only updated quarterly. For example, for the issue of The Cochrane Library published in January 2007, the MEDLINE records were downloaded by Wiley-Blackwell staff in November 2006. The January 2007 publication of The Cochrane Library was the current issue until April 2007, so the MEDLINE records range between being two to five months out of date. The most recent months of MEDLINE should, therefore, be searched, at least for records indexed as either ‘Randomized Controlled Trial’ or ‘Controlled Clinical Trial’ in the Publication Type, to identify those records recently indexed as RCTs or CCTs in MEDLINE.

 

Additionally, the most recent year to be searched under the project to identify reports of trials in MEDLINE and send them back to the US National Library of Medicine for re-tagging was 2004, so records added to MEDLINE during and since 2005 should be searched using one of the search strategies described in Section 6.4.11.1.

 

Finally, for extra sensitivity, or where the use of a randomized trial ‘filter’ is not appropriate, review authors should search MEDLINE for all years using subject terms only.

 

It should be remembered that the MEDLINE re-tagging project described in Section 6.3.2.1 assessed whether the records identified were reports of trials on the basis of the title and abstract only, so any supplementary search of MEDLINE that is followed up by accessing the full text of the articles will identify additional reports of trials, most likely through the methods sections, that were not identified through the titles or abstracts alone.

 

For guidance on running separate search strategies in the MEDLINE-indexed versions of MEDLINE and the versions of MEDLINE containing ‘in process’ and other non-indexed records please refer to Section 6.4.11.1.

 

Any reports of trials identified by the review author can be submitted to the Trials Search Co-ordinator who can ensure that they are added to CENTRAL. Any errors, in respect of records indexed as trials in MEDLINE that on the basis of the full article are definitely not reports of trials according to the definitions used by the National Library of Medicine (NLM) (see Section 6.3.2.1), should also be reported to the Trials Search Co-ordinator, so they can be referred to the NLM and corrected.

 

For general information about searching, which is relevant to searching MEDLINE, see Section 6.4.

 

Searching EMBASE

The project to identify reports of trials in EMBASE for inclusion in CENTRAL, described in Section 6.3.2.2, is carried out on an annual basis, so there is a time lag of approximately one to two years with respect to EMBASE records appearing in CENTRAL. The last two years of EMBASE should, therefore, be searched to cover work still in progress.  Some suggested search terms are listed in Section 6.3.2.2. A search filter designed by the McMaster Hedges Team is also available (Wong 2006).

 

Finally, for extra sensitivity, or where the use of a randomized trial ‘filter’ is not appropriate, review authors should search EMBASE for all years using subject terms only, as described under similar circumstances for MEDLINE above. It should be remembered that the EMBASE project described above assessed whether the records identified were reports of trials on the basis of the title and abstract only, in the same way as the MEDLINE project described above. Therefore, any supplementary search of EMBASE that is followed up by accessing the full text of the articles will identify additional reports of trials, most likely through the methods sections, that were not identified through the titles or abstracts alone.

 

For general information about searching, which is relevant to searching EMBASE, see Section 6.4.