Article type
Year
Abstract
Background: A Cochrane methodology review found that handsearching journals not indexed in MEDLINE, in particular those published in languages other than English, can be a valuable way to identify randomized controlled trials (RCTs) of healthcare interventions.
Objectives: We sought medical journals published in Africa in 1985 - 2003 and handsearched these to identify RCTs.
Methods: We identified African medical journals by using country and disease terms to search the journal catalogues of the National Library of Medicine, and of the libraries of the London School of Hygiene and Tropical Medicine, the World Health Organization and the Kenyan Medical Research Institute. We also searched electronic journal databases relevant to Africa (including ZETOC and ExtraMED) and communicated with the Federation of African Medical Journal Editors. To obtain a broadly representative sample from the identified journals, we handsearched at least one journal per country. One of three handsearchers (one a first-language French speaker) checked the tables of contents of each journal, obtaining the article if the title suggested an RCT. We recorded the number of original articles and RCT reports per issue per journal.
Results: We identified 67 relevant journals from 20 countries. Forty were not MEDLINE-indexed. We handsearched 35 of these, from 16 countries. Twenty-eight journals were English, six French and one Portuguese. The library holdings we had access to were incomplete for many journals and some journals did not have regular publication periods. Overall, there were 6378 original articles of which 14 described RCTs. These 14 reports were published in seven different journals.
Conclusions: The yield of RCTs per original article is low compared with similar studies done elsewhere. Our results suggest that few non-indexed African medical journals publish RCTs and that handsearching these journals might not be cost-beneficial. Consecutive numbering of issues, rather than volume/issue numbering would facilitate the identification of missing issues.
Objectives: We sought medical journals published in Africa in 1985 - 2003 and handsearched these to identify RCTs.
Methods: We identified African medical journals by using country and disease terms to search the journal catalogues of the National Library of Medicine, and of the libraries of the London School of Hygiene and Tropical Medicine, the World Health Organization and the Kenyan Medical Research Institute. We also searched electronic journal databases relevant to Africa (including ZETOC and ExtraMED) and communicated with the Federation of African Medical Journal Editors. To obtain a broadly representative sample from the identified journals, we handsearched at least one journal per country. One of three handsearchers (one a first-language French speaker) checked the tables of contents of each journal, obtaining the article if the title suggested an RCT. We recorded the number of original articles and RCT reports per issue per journal.
Results: We identified 67 relevant journals from 20 countries. Forty were not MEDLINE-indexed. We handsearched 35 of these, from 16 countries. Twenty-eight journals were English, six French and one Portuguese. The library holdings we had access to were incomplete for many journals and some journals did not have regular publication periods. Overall, there were 6378 original articles of which 14 described RCTs. These 14 reports were published in seven different journals.
Conclusions: The yield of RCTs per original article is low compared with similar studies done elsewhere. Our results suggest that few non-indexed African medical journals publish RCTs and that handsearching these journals might not be cost-beneficial. Consecutive numbering of issues, rather than volume/issue numbering would facilitate the identification of missing issues.