Introduction: Electronic databases provide unreliable access to identifying randomized controlled trials (RCTs). Hand searching journals remains a costly necessity. We can add to the value of hand searching by conducting epidemiological studies of the trials we find. We report a study, assessing the accessibility and methodological quality of the RCTs found in the British Journal of Surgery (Br J Surg).

Objective: To asess the profile of RCTs found by handsearching Br J Surg 1947-1993.

Methods: The Br J Surg was hand searched in 1994 and each RCT coded as to whether:

1. it contained the term "random" (randomly allocated, randomization etc.) in the written title or abstract
2. it was tagged on MEDLINE as a Randomized Controlled Trial in the database's Publication Type field (PT=RCT)
3. it was reported in full or as an abstract
4. the method of randomization was described
5. all the subjects randomized had been adequately accounted for in the analysis
6. the outcomes had been blindly assessed


Results: 552 RCTs were identified by handsearching, of which only 83 had been tagged on MEDLINE PT=RCT in 1994, although 426 had used the term "random" in the title or abstract. Of the 552, 417 were full length articles. Of these 417, 179 included some description of the randomization procedure, 191 accounted for all the subjects in the analyses and 138 reported that the outcomes had been blindly assessed.

Discussion: Cochrane entities, such as Fields, are ideally placed to undertake reviews of key journals, identifying those areas where the quality of reported research is poor and indicating ways in which journals can be more effective in making trials identifiable for indexers and reviewers.