Hand searching of Finnish medical journals

Article type
Authors
Pasternack I, Varonen H, Juva K, Lehtinen E, Jormanainen V, Mäkelä MS
Abstract
Objective: To determine the number of randomised controlled trials (RCT), controlled clinical trials (CCT) and meta-analyzes (MA) published in main medical journals, in Finland. To assess the sensitivity and precision of hand search and electronical (Medline) search.

Methods: Finnish Medical Society Duodecim's journal Duodecim was the target of our pilot hand search. We developed a congruent and comprehensive hand search method within a group of six people. Quality control of individual work was ascertained by regular meetings where problem cases were discussed. Optimal search strategy for Medline developed by Cochrane Collaboration was used to find electronically every RCT and CCT published in Duodecim regardless of subject. The findings from hand search and Medline were compared to the best possible result which was the combined result of both methods.

Results: Our pilot hand search identified 68 studies from Duodecim 1954-1996 meeting the criteria of RCT, CCT or MA. From 1966 onwards the number was 55 clinical trials hand searched and 50 from Medline. Of these 50 trials from Medline 25 were non-eligible and 6 not found in hand search. Of the 55 trials hand searched 36 were not found in Medline. Thus the sensitivity of hand search was 90% and that of Medline 41%.

Discussion: Finnish medical journals proved to be a relevant source of randomized and controlled clinical trials. Our Medline search for clinical trials appeared to be quite unspecific, giving much "irrelevant" material: editorials, follow-up studies etc. On the other hand more than half of the randomized and controlled clinical trials identified by hand search were not coded as clinical trials in Medline.