The assessment of an electronic searching strategy with a Spanish medical journal

Article type
Year
Authors
Lacalle JR, Fernandez A, Polo J, Moreno A, Marin I
Abstract
Introduction/Objective: A searching strategy has been proposed in the Collaboration Handbook for locating published randomized clinical trials; this tool merits a proper evaluation in terms of sensitivity and specificity. The case of papers not published in English needs a particular assessment. In this case, there is an additional problem: coding of papers is based on translations and the process is not completely structured, and consistency is not assured. The objective of this work is to assess the searching strategy (measuring the precision and recall indexes), comparing results of the electronic searching and the hand searching of a Spanish general medical journal, Medicina Clinica.

Methods: A hand searching of papers reporting results of randomized clinical trials, and published in that journal, is undertaking by the authors. Medicina Clinica has been published since 1943, and nowadays it is one of the most respected Spanish medical journals. Its impact factor in 1993 is 0.909, and it is ranked 32 among 115 journals included in the bibliometric study of Science Citation Index/Journal Citation Report. It has been indexed in the Index Medicus since 1983. The methodology for hand searching, developed at the Baltimore Centre, is adopted. The electronic searching is made with the strategy developed at the Oxford Centre and published in the Cochrane Collaboration Handbook, and adapted to the specific aim of this work, by restricting the searching to those records from Medicina Clinica. Considering results of the hand searching as a gold standard, two indexes are estimated. The recall index is defined as the number of relevant citations retrieved divided by the number of relevant citations published; it is equivalent to the sensitivity of the search. The precision index is defined as the number of relevant citations retrieved divided by the total number of citations retrieved; it is equivalent to the specificity of the search.

Results/Discussion: The comparison of the two searches provides some keys to improve the electronic searching strategy. Several papers, indexed as randomized clinical trials, are considered false positive ones because they deal with methodological aspects of design and analysis of clinical trials, but they do not report original results.