ELDAC's Home Care Search Filter supports one-click evidence searching
A blog by Professor Jennifer Tieman, Flinders University Lead at ELDAC
We know many health professionals and managers struggle to find research literature. They can find it difficult to produce effective searches to use in bibliographic databases, as they are not familiar with search processes and the syntax of the database. This can be even more challenging when the topic of interest is relatively new, and where there is no universal definition. Home care is one such topic.
In Australia, home care is a coordinated package of care and services to help the aged person with needs live independently in their own home for as long as possible. However, a search just using the words ‘home care’ retrieves a large volume of papers, many of which are not relevant to the topic. This happens because individual countries have different care systems and use other terms to describe their arrangements. It also makes it hard to search the bibliographic databases to see what research has been done and what works in this setting. As part of our work program in 2020-2023, the Flinders researchers contributing to the ELDAC Technology and Innovation work stream were responsible for developing a validated search filter for use in PubMed.
Having a validated search that exists as a hyperlink which offers access to relevant abstracts in an open access database would save time and effort and connect the user to relevant research with one click. Being able to limit it to free full text could further support users by focusing retrievals on those they can read in full for free. Our study experimentally developed a search filter which will focus on retrieving papers from the PubMed database which are relevant to home care within an Australian context. This is now available for use on the ELDAC website.
Search filters are experimentally designed and tested search strings created by expert searchers. They can considerably reduce the time taken to find relevant articles. The performance of a search filter is determined and reported according to metrics such as the filter’s recall or precision, which denotes how many papers will be relevant amongst those retrieved, allowing the user to assess the filter’s suitability for their needs.
The first step in the process was creating a gold standard set of articles that are relevant to the topic ‘home care’. There are different ways of constructing a gold standard set. For this search filter we included papers in systematic reviews relevant to the topic of home care. With the help of the members of our Expert Advisory Group, we screened 146 systematic reviews for relevance to home care and then extracted the included papers from each review, screened out those which weren’t indexed in PubMed, and did a final dual screen for relevance. This created the gold standard test set.
We then divided the gold standard set of articles into a term identification set and a test set. The titles, abstracts, and Medical Subject Headings (MeSH) terms associated with each citation in the term identification set were examined for common concepts that might individually or collectively describe ‘home care for older people’ and the terms used to convey these concepts. This enabled us to identify potential MeSH and text words for use in building the search filter. We then checked the retrieval performance by combining each term systematically to determine the most effective search with its known level of retrieval performance.
We then completed a further set of other activities to confirm the retrieval performance in the remaining citations in the test set. Finally, we ran the search in PubMed and asked the Expert Advisory Group to each review 200 retrievals to check their relevance to the topic. This provides a ‘real world’ test of the search filter’s performance.
This seems a lot of effort to write a search. But we want to make sure the base search filter is as strong as it can be in retrieving what is relevant and reducing the retrieval of non-relevant items. We then combined the Home Care Search Filter with other searches to provide a shortcut access to evidence. Some of these searches include topics such as dementia, workforce, quality of life, and costs and economics.
Fortunately, as a user, all that you need to do is go to the ELDAC Homecare Search Filter page, click on the search you want and press the hyperlink. A new page will load in the PubMed database with relevant articles. We have taken the hard work out of searching and you are now just one click away from evidence.
Professor Jennifer Tieman
Flinders University Lead, ELDAC