Are you searching the right database? An example from mental health

Tags: Poster
Taylor L, McDonald S, Adams C

Introduction/Objective: To assess the coverage of the mental health literature within readily available bibliographic databases, and to consider the implications of this for the preparation of systematic reviews.

Methods: Using the Ulrich's Serials database, details of all journals and other publications indexed under the single term 'PSYCHIATRY AND NEUROLOGY" were downloaded and transferred into a ProCite database. The publications were then categorised as psychiatry or neurology. Most of the titles were categorised according to whether the term psych* or neuro* appeared in the title, (neuropsychiatry was grouped under psychiatry). The remainder were categorised on a title by title basis (CA). Journals and other publications where the subject matter was unclear were not assigned to either category. (The majority of these seemed to be newsletters rather than academic publications and were not indexed in any database.) Using the information in Ulrich's relating to bibliographic databases, journals assigned to the psychiatry category were then analysed to see which database (or combination of databases) indexed the most titles.

Results: A total of 1435 publications indexed under 'PSYCHIATRY AND NEUROLOGY" were identified from Ulrich's. Nine hundred and eighty were categorised as being psychiatry. A total, 223 databases were listed as indexing at least one of the 980 journals. Of the 980, 505 were listed as being indexed in one or more of the following four databases (Biological Abstracts, EMBASE, MEDLINE, Psychological Abstracts). Of the top four databases, Psychological Abstracts indexed 370 journals (73%), EMBASE 337 (67%), Biological Abstracts 242 (48%), and MEDLINE 236 (47%). One hundred and twenty two journals (24%) were indexed in all four databases. Combining the database with the highest yield (Psychological Abstracts) with any one of the other databases increased the number of journals indexed as follows: 73% (Psych Abs alone), 85% (Psych Abs and MEDLINE), 87% (Psych Abs and Biol Abs), and 91% (Psych Abs and EMBASE).

Discussion: No one database covered more than 73% (Psych Abs) of the psychiatry/mental health journals in this study, and MEDLINE alone only covered 47%. Furthermore, each of the four databases had a number of unique titles, 175 in total (35%). For those involved in the process of identifying reports of trials for inclusion in systematic reviews, this study emphasises the importance of searching more than one database to ensure optimal coverage of the literature.