Appraisal of: "Pereira RA, Puga MEDS, Atallah AN, Macedo EC, Macedo CR. LILACS search strategy for systematic reviews of diagnostic test accuracy studies. Health Info Libr J. 2019 Sep;36(3):223-243."

Short description: 

The aim of this study was to add to the literature on effective information retrieval methods to identify DTA studies in Latin American and Caribbean Health Sciences Literature database (LILACS). Authors first developed a six-step process for translating MEDLINE search strategies into LILACS database, then used the process to translate and run a sample of 70 published Cochrane systematic reviews of DTA MEDLINE search strategies into LILACS databases. If the Cochrane systematic review in the sample had reported running the strategy in LILACS, authors assessed and compared the number of studies retrieved between the Cochrane strategy and the strategy they developed.

Of the 29 Cochrane systematic reviews reported to have searched LILACS, 21 included their full LILACS search strategy. An analysis of the 21 Cochrane search strategies used in LILACS revealed suboptimal searching which may have reduced the number of studies retrieved: less than 50% used tools like truncation or wildcards and 86% used only English language terms (LILACS database also includes Spanish and Portuguese language publications). The study’s findings underscore the need for proper translation of search strategies into LILACS database.

Limitations stated by the author(s): 

Authors identified limiting the sample to Cochrane DTA systematic reviews only as potentially limiting the generalizability of their findings. Authors did not assess whether unique studies identified in LILACS would have met the inclusion criteria of the DTA reviews, nor if these unique studies being included would change the reviews’ conclusions.

Limitations stated by the reviewer(s): 
It is not clear if the authors using the following metric as an indicator of effective translation is correct: “We considered the translated LILACS search strategy effective if it retrieved at least one study” (p.228).
Study Type: 
Single study