Introduction: Searching MEDLINE for randomised controlled trials (RCTs) is tried and tested but identifies only about 50% (30% - 80%) of relevant reports. The most efficient way to find the other 50% is uncertain (Dickersin et al, 1994).

Objective: To assess, in the context of 6 systematic reviews related to end stage renal disease, the extra dividend, in terms of relevant RCTs, from alternative search strategies, once a MEDLINE search had been completed.

Methods:: Systematic searches of 4 additional electronic databases (EMBASE, CINAHL, BIOSIS and the Cochrane Library); searches of reference lists of relevant RCTs and possible RCTs found in MEDLINE; asking these authors to identify extra RCTs; handsearching one key journal; and other specific sources including SIGLE (System for Information on Grey Literature in Europe compiled by EAGLE), ChemAbs (Chemical Abstracts), and CRIB (Current Research in Britain). Comparison of the numbers of RCTs identified from these sources individually and in combination.

Results: Currently, a total of 65 trials have met the inclusion criteria. 45 of these were found in MEDLINE after reading 5429 abstracts; a further 9 were found amongst 6447 abstracts generated by the searches in the other electronic databases; the remaining 11 were found in other ways. Of 128 sets of authors contacted, 49 replied providing 72 references and, to date, 2 more RCTs have been identified and will be included in the final reviews. A hierarchy of RCT yields by source will be shown.

Discussion: MEDLINE identified about 70% of RCTs relevant to these reviews. This study will help those developing similar searches to decide the ordering and type of further searches to achieve maximum efficiency.

References: Dickersin K, Scherer R, Lefebvre C. Identifying relevant studies for systematic reviews. BMJ 1994;309:1286-91.