Statistics from Altmetric.com
The acquisition of information is an important part of scientific work. In recent years, available information sources have multiplied exponentially, and as a result, we have come to find ourselves immersed in an “information overload”. Although we feel committed to keeping up to date with the advances in neurology by using our information seeking skills, paradoxically the new, widely expanded information technologies have made it more and more difficult to gain knowledge about relevant literature sources and to retrieve information. Under the pressure of this situation, many of us have resorted to the passiveness of searching only the most commonly used databases such as Medline and Current Contents to acquire the information we need.
In the frequent practice of retrieving information needed for patient assistance, it is accepted that the use of Medline gives good results for European and North American countries. However, in searches directed towards investigation, we suspected that other, perhaps less accessible databases, could provide valuable additional information. To determine to what extent this might be true, we performed a literature search in the field of multiple sclerosis.
The analysis began by connecting to the 179 Science and Technology databases included in the host Dialog (Knight-Ridder Inc), using the commands for multiple database searching. To avoid classification problems due to variations in the internal structure of the databases, we requested that the search term “multiple sclerosis” appear in any part of the document. As the databases cover different periods and have different indexing policies, we restricted the search to the complete year of 1995 and, at the beginning, to the title field. However, we realised that using only the title would lead to poor results (for example, 613 titles in Biosis instead of the 1100 we used later), so we used the entire reference fields instead. Finally, we excluded datatabases devoted to medical Industries and ones that are contained in more general information sources (for example, the New England Journal Full Text Database). The first search yielded 24 usable databases with a total of 5350 references.
It was found that most references retrieved (4615) were concentrated within the following databases: Current Contents (1239), Biosis (1100), Embase (892), Medline (739), Pascal (406 ), and IAC Health and Wellness (378). The titles of the articles recovered were extracted and two main points were developed: duplication of references among the databases and the suitability of the databases for each information requirement. To determine suitability, the documents were classified according to their titles into four groups, representing the major headings inMcAlpines’ multiple sclerosis handbook 1 : (1) epidemiology, (2) clinical aspects (signs and symptoms, course, and prognosis, natural history, neuropsychology, diagnosis, laboratory diagnosis, and therapy), (3) pathogenesis (including genetics, immunology, and animal models), and (4) pathology. After carrying out the classification, we found that an important subset of documents did not correspond to any of the groups; these were then categorised into: (5) health promotion (including quality of life and social aspects), (6) general aspects (particularly review articles dealing with several of the former topics), and, (7) noise (documents with no apparent relation with multiple sclerosis). All the documents retrieved were classified by members of the multiple sclerosis unit (clinicians and basic researchers) after training sessions to establish the classification criteria, and the final review was performed by a neurologist expert in multiple sclerosis.
A high percentage of duplication of references was found in five of the databases studied (average overlap 22.4%). Health and Wellness showed no overlapping and, moreover, indexed most of the articles on Social Aspects. Considering only Medline, Embase, and Current Contents, average overlap was 26.6%.
The table shows the number of references retrieved and the percentage of references in relation to the total for each aspect of multiple sclerosis on each database.
For our purposes, quantity was more valuable than percentage, as the larger the number of references obtained the more the information that was gathered.
Thus according to the categories established, we found that the most suitable database for epidemiology, clinical aspects, and pathogenesis was Biosis, whereas the most appropriate for health promotion was Health and Wellness; pathology was best covered by Current Contents, and general aspects by Embase. The highest rate of noise was found in Current Contents.
These results evidence that use of Medline alone for multiple sclerosis information searches will not provide optimum returns and can have economic implications resulting from duplication of scientific efforts.
However, Medline is the best known and most available database around the world and it is unusual to find other information sources integrated within institutional information systems.
A way out of this problem is the use of On-line Search Services, a common element of library services. They can provide access to a wide range of databases and are managed by specialised information professionals.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.