Some systematic review topics are context-sensitive and informed by evidence about a specific geographic setting. Smart information retrieval methods are required to identify such evidence in an efficient manner. This presentation will discuss how validated geographic search filters enable smart literature searching for context-sensitive reviews using the National Institute for Health and Care Excellence (NICE) United Kingdom (UK) filters for MEDLINE and Embase (OVID) as examples. The NICE UK filters were developed in 2016. The filters demonstrated high recall and high precision, however, further research was required to confirm these results.

In 2018, the filters’ recall of references from 100 UK-based multidisciplinary reviews was calculated. Reproducible search strategies were identified from twenty-six of the 100 reviews in MEDLINE and from nine reviews in Embase. From this, the precision and number-needed-to-read (NNR) were calculated.

The MEDLINE filter achieved 96 percent recall (1401 out of 1454 UK references), 2.1 percent precision and a NNR of forty-seven. The Embase filter achieved 97 percent recall (1520 out of 1560 UK references), 0.7 percent precision and a NNR of 146. Compared to not using a filter, the MEDLINE and Embase filters reduced the number of search results by an average of 87 percent and 80 percent, respectively.

The filters retrieve the majority of evidence for UK topics while reducing search result volumes and so enable smart literature searching for context-sensitive topics. Large literature search result volumes can increase development time-frames for systematic reviews. Using the filters can therefore save time for reviews with a UK focus. There are currently two other validated geographic search filters for Africa and Spain. It is hoped that the NICE UK filters’ successful retrieval performance will encourage the development of validated search filters for more geographic regions.