OBJECTIVES: Geographic-specific evidence is required to inform context-sensitive recommendations, e.g., UK evidence for NICE and US-specific content for AMCP dossiers. However, geographic details are not always adequately reported in titles/abstracts or controlled vocabularies in biomedical databases. We created two geography search filters and tested their performance against completed literature reviews.

METHODS: The geography search filters were created for the US and UK and consisted of the following search fields: emtree headings, abstract, title, keywords, author address, and author country. Search terms related to US/UK, UK countries, FDA/NHS, states, major cities, and nationalities were included. We identified completed reviews to test our search filters: UK or US-focused with a clear categorization of relevant references, didn’t utilize any geographic search filter (to prevent bias through including pre-restricted US/UK references), and their selection had a broad range of healthcare topics (e.g., epidemiology, quality of life, patterns) and study designs. The calculations used were Recall: Number of relevant citations retrieved by filter/Total number of relevant citations; Number needed to screen (NNS): Number of citations that must be screened to retrieve every relevant citation (1/precision).

RESULTS: The US filter was tested against five reviews. It demonstrated an average of 97.77% recall for the included US citations. One citation was missed because our filter didn’t have SEER as an identifier. The average NNS reduced from 200 to 97. The total citations were reduced by an average of 46.63% (32,271 to 17,222). Similarly, the output of the UK filter was compared to five reviews and resulted in an average recall of 100%. The average NNS reduced from 44 to 23. The total citations were reduced by an average of 59.77% (31,722 to 12,759).

CONCLUSIONS: The filters demonstrated high recall, reduced NNS, and decreased the screening workload of titles/abstracts by ~50%. They can be utilized to retrieve US and/or UK-specific evidence.