Bias against European journals in medical publication databases

one of the international databases. The Medline search retrieved 222 (69·4%) articles, Embase 242 (75·6%) articles, and PsycLIT 257 (80·3%) articles. The overall breakdown between Medline and Embase was: non-indexed 18·4%; Medline uniquely 5·9%, overlap 63·4%; Embase uniquely 12·2%. The breakdown between Medline and PsycLIT was: non-indexed 4·1%; Medline uniquely 15·6%, overlap 53·7%; PsycLIT uniquely 26·6%. Of the 98 articles that were not found in Medline, 39 (39·8%) were included in Embase, and 85 (86·7%) in PsycLIT. Most of these Medline non-indexed articles (60·2%) reported original studies based on the statistical analysis of data, 21·4% were reviews and 18·4% other narrative reports. The proportion of clinical trials was under 1 0 % . There were nine international journals that had published at least two articles but which are not indexed by Medline (table). These were all European journals. We conclude that this proportion 27·2% (87/320) estimates the coverage bias against European journals in Medline. In locating articles, researchers often use Medline as the only computerised database as it is easy to access and the costs are minimal. The probability of identifying relevant studies for references is influenced by database coverage. We have noted and empirically defined coverage bias against European journals in the Medline database. This bias can lead to lower citation counts and will not increase the visibility of European research. Articles published in non-indexed journals are rarely located, and hence rarely cited. The results also imply that a substantial portion of psychiatric research is published outside the journals indexed by Embase or Medline. Whereas there is overlap between databases, they each have a unique focus, and significant differences do exist. Although Medline is the best known database for all health professionals, PsycLIT is particularly valuable for psychiatrists and other mental health p r o f e s s i o n a l s .