Developing a sensitive search strategy in Farsi for retrieving reports of randomized trials in Iranmedex

Tags: Poster
Shahiri M, Nasser M, Javaheri H

Background: Some Iranian healthcare journals are indexed in MEDLINE

and EMBASE, all of them are in English. Iranmedex indexes 130 Iranian

healthcare journals of which 101 are in Farsi language and provides links

to the full text of most articles. This database is searched using simple

words. We had previously developed a search strategy in English for this

database. Objectives: To develop a sensitive search strategy in Farsi to

retrieve reports of randomized controlled trials (RCTs) in Iranmedex, an

Iranian healthcare database. Methods: Five Iranian healthcare journals will

be handsearched to identify reports of RCTs and quasi-randomized trials

(CCTs) according to Cochrane criteria. We will match half of the reports

identified by handsearching with their corresponding electronic record in

Iranmedex, where these exist. The electronic records will be examined for

study design terms in the keywords and full text. These terms will be used

to compile a search strategy in Iranmedex. The sensitivity and precision of

this search strategy will be tested against the second half of reports found

by the initial handsearch. The electronic records of any reports of trials

missed by this strategy will be examined for additional relevant search

terms to improve sensitivity. Results: The initial phase of the project was

completed in March 2008. The handsearching yielded a range of terms

derived from 31 reports of trials, both RCTs and CCTs, found in the

JODENT (2001-2005) and the Iranian Journal of Paediatrics (2002-2007).

These terms are the Farsi terms for: randomized (19/31, 1/31 (two

different translations for the term)), clinical trial (12/31), interventional

study (4/31), clinical trial (English term) (1/31), control (1/31, 4/31),

controlled (1/31), control group (1/31, 8/31), placebo (2/31, 2/31), single

blind (1/31), double blind (3/31), cross-over (1/31), clinical trial (1/31),

randomized clinical trial (4/31) in the abstract and title. Conclusions: Full

results of this electronic search strategy for identifying reports of RCTs in

Iranmedex, validated against handsearch results of known reports of trials

in Iranian healthcare journals, will be completed by October 2008.