Background: Minimising bias in reviews which results from the inadequate retrieval of relevant studies is an ongoing concern. The Australasian Medical Index (AMI) is the leading source of health literature published in Australia, and specialises in covering literature not indexed in databases such as MEDLINE. Preliminary searches of AMI found that although the database contains many reports of trials, the majority are poorly indexed and difficult to retrieve. Objectives: To identify reports of controlled trials from the Australasian Medical Index using a highly sensitive search strategy, and to ensure these reports are made accessible to those doing reviews.

Methods: I devised a highly sensitive search strategy for retrieving reports of controlled trials and ran it against the Australasian Medical Index (1966-99). I downloaded retrieved records and a group of trained searchers scanned them. Where it is not possible to tell conclusively from the title/abstract that it is a controlled trial, they will review the full article (if available). I will submit all reports of trials identified for inclusion in the Cochrane Library and to AMI for re-tagging with the correct indexing term. I will analyse the reports identified by subject, journal, country of publication and year of publication.

Results: A combination of thesaurus and free-text terms retrieved over 3500 potential reports of trials. Of these, less than 150 were indexed with methodological terms, such as randomized-controlled-trial. To date over 500 reports of trials have been identified, and additional reports are likely to be identified from reviewing the full article.

Conclusions: Identifying trials is an ongoing and important task to ensure the completeness of systematic reviews. This work is especially valuable as it identifies trials not widely available and contributes a wider perspective to international research.