Juan Jorge Manriquez. Searching the LILACS Database Could Improve Systematic Reviews in Dermatology. Arch Dermatol. 2009;145(8):947–968. doi:10.1001/archdermatol.2009.153
Well-performed systematic reviews should analyze as many articles as possible to provide the best evidence available. However, some reviews limit their searches only to a few numbers of databases, mainly literature in English, published in journals fundamentally edited in developed countries.1
The LILACS database is an underused source of trials that indexes journals mainly from Latin American and Caribbean countries.2 In the present study, I sought to assess whether including a LILACS search improved the quality of systematic reviews in dermatology.
I evaluated reviews from the Cochrane Skin Group and a sample of non-Cochrane reviews published from 2002 to 2007 in the 4 dermatologic journals with the highest impact factors (Journal of the American Academy of Dermatology, Archives of Dermatology, Journal of Investigative Dermatology, and British Journal of Dermatology). The first group of reviews was retrieved from the Cochrane Skin Group Web site,3 and the non-Cochrane reviews were found through a search in Medline, according to the strategy described by Montori et al.4 A systematic review was defined as a study that used any systematic way of searching the literature using explicit criteria for article selection.
For the included reviews, we used LILACS to locate randomized clinical trials that matched each review's inclusion criteria using a validated, highly sensitive LILACS search strategy described elsewhere.1 The LILACS search result was classified positive when at least 1 randomized controlled trial that fit the inclusion criteria was found and negative when no such trials were located. The search results were considered inconclusive when at least 1 trial that fit the inclusion criteria was found but it could not be sorted as a randomized one. For the positive LILACS search results, the references listed in the review were checked to determine if the identified articles had already been located.
A total of 44 reviews (25 Cochrane and 19 non-Cochrane) were analyzed. Three of the Cochrane reviews included a LILACS search and were excluded from further analysis (Table and eTable). Twenty percent of the reviews (8 of 41) were restricted to English-language articles and 51% (21 of 41) explicitly had no language restriction.
The LILACS search results were positive in 29% of all reviews (12 of 41), inconclusive in 34% (14 of 41), and negative in 37% (15 of 41). In 5 of the 14 inconclusive cases, the original review allowed any kind of trial as inclusion criteria. In Cochrane reviews, 18% of the search results were positive (4 of 22), whereas in the non-Cochrane reviews 42% were positive (8 of 19). On the other hand, 27% (6 of 22) and 42% (8 of 19), respectively, produced inconclusive results (references available from the author).
Among the 12 reviews with positive LILACS search results (4 Cochrane and 8 non-Cochrane), only 1 Cochrane review had located the article identified on LILACS by other methods. Therefore, in 27% of the systematic reviews (11 of 41), a LILACS search was effective in identifying new articles suitable for inclusion and not located by the authors.
In conclusion, using LILACS can increase the number of trials potentially suited for inclusion in systematic reviews.
Correspondence: Dr Manriquez, Department of Dermatology, Unit of Evidence-Based Medicine, School of Medicine, Pontificia Universidad Catolica de Chile, 4686 Vicuña Mackenna St Macul, Santiago, Chile (email@example.com).
Financial Disclosure: None reported.