Understanding the performance of geographic limits on Web of Science Core Collection databases, using the United Kingdom as an example.

Abstract

Objective: To consider the approaches within Web of Science Core Collection (WoSCC) databases for limiting geographically. To compare the limits to an adaptation of NICE's UK MEDLINE filter for use on WoSCC databases.

Methods: We tested and appraised the inbuilt functions and search field options that support identification by countries/regions and affiliations. We compared these with an adapted filter to identify healthcare research on or about the UK. We calculated the recall of the inbuilt limits and filter using 177 studies and investigated why records were missed. We also calculated the percentage reduction of the overall number-needed-to-screen (ONNS).

Results: Inbuilt limits within WoSCC enable identification of research from specific countries/regions or affiliations if there is data in the address field. Refining by affiliations allows retrieval of research where affiliations are in the 200 or 500 most frequent for a set of results. An adaptation of the UK MEDLINE filter achieved an average of 97% recall. ONNS was significantly reduced using the filter. However, studies where the countries or regions are only mentioned within the full text or other non-searchable fields will be missed.

Conclusion: Information specialists should consider how inbuilt geographic limits operate on WoSCC and whether these are suitable for their research. The adapted filter can sensitively limit to the UK and could be useful for systematic reviews due to its high recall and ability to significantly reduce ONNS. Geographic filters can be feasible to adapt for use on WoSCC databases (where similar search fields are used between platforms).