Residential Aged Care Search Filter project

CareSearch investigators: Jennifer Tieman and Sarah Hayman

Project status: Completed in October 2012.

Link to: Residential Aged Care Search Filter and associated topic searches

The 2011-2014 CareSearch Agreement included the development of a Residential Aged Care Hub as a CareSearch project activity.

Health care reform and the growing impact of Australia's ageing population as well as feedback to the CareSearch project had highlighted the needs of, and contribution made by, residential aged care in caring for residents at the end-of-life.

The Residential Aged Care (RAC) Hub is a collection of information, tools and resources developed to form a hub of knowledge. It is designed to support staff of RAC to provide practical support to residents who are dying, and to provide quality information for residents and their families.

The RAC Search Filter was developed as part of the RAC Hub project, to provide reliable and ready access to quality evidence by enabling one-click access to PubMed search results about residential aged care in a palliative care context.

The general RAC Search Filter was developed and validated using established methodology as outlined below. This filter has been published on the CareSearch site as a general RAC search filter and in combination with the existing palliative care filter.

Overview of methodology:

The RAC Search Filter was developed using the classic methodology outlined below. It can therefore be designated a ‘validated’ search filter.

The RAC Search Filter Expert Reference Group ratified a set of systematic reviews taken from a range of sources for evidence-based literature across medicine, nursing and allied health disciplines. This provided a set of credible residential aged care systematic reviews that covered a broad range of topics relevant to residential aged care, while focusing exclusively on a residential aged care population. The sources were: the Cochrane Library (focus: medicine), Joanna Briggs Connect+ (focus: nursing), PEDro (focus: physiotherapy), PsycBITE (focus: psychology), SpeechBITE (focus: speech therapy), OTSeeker (focus: occupational therapy) and the CareSearch Review Collection (focus: palliative care). Some additional systematic reviews were added following advice from the expert reference group.

The references were collected from the set of systematic reviews and stored in an EndNote library. References were added to the gold standard set if they:

  • were included studies
  • were excluded studies explicitly excluded for reasons other than relevance to RAC (i.e. they were analysed by the systematic review authors as relevant to RAC)
  • were in English
  • were indexed in Medline
  • contained an abstract.

It became clear on scrutiny of the titles, abstracts and keywords by the project team that a significant number of the remaining candidate references appeared of doubtful relevance to residential aged care, although meeting the inclusion criteria for the systematic reviews. Where full text could be obtained for these doubtful references, they were provided to members of the expert review group, working in pairs, who determined which of these were relevant. The remainder were either classified not relevant on review of the full text or the full text had not been able to be obtained. This produced a gold standard set. The gold standard set was randomly divided into three subsets: one for term identification; one for filter development; and one reserved for final filter validation.

Frequency analysis was applied to the Term identification set to identify high frequency subject headings (MeSH terms) and textwords (title/abstract natural language terms). Other potential search terms were drawn from the search strategies of the relevant systematic reviews themselves, from consultation with the expert working group, and by exploring all relevant branches of the Medline’s MeSH thesaurus concept tree.

All candidate search terms identified in step three were tested for their ability to retrieve relevant citations, both singly and in combination with other candidate search terms, in the filter development set.

The combination of search terms best able to retrieve the largest proportion of citations from the development set was then tested in the filter validation set. The draft search filter was run in open Medline and the first 200 references retrieved were added to an Endnote Library. Of these, 181 were in English and were sent to pairs of reviewers for assessment of relevance, with a third reviewer assessing relevance where there was not agreement.

Results of post hoc relevance assessment by reviewers indicated that the high sensitivity of the draft search filter had resulted in a lower than desired specificity.

Further versions of the search filter were developed and tested in three sets (the Filter development set, the Filter validation set and the Post hoc relevance set (with known relevance assessment by the reviewers) to ascertain the combination of terms that would maximise both sensitivity and specificity. We also bore in mind the clearly expressed view of the Expert Reference Group that greater specificity (fewest irrelevant items retrieved) would be of the greatest clinical benefit in practice on the RAC Hub.

The final Medline search filter was:

Nursing homes/ or Nursing home$.ti. or Long term care.ti. or longterm care.ti or Homes for the aged/ or Residential aged care.mp. or LTCF.mp. or Aged care facilit$.mp. or skilled nursing facilities/

Further post hoc relevance testing was carried out with 100 records retrieved from open Medline by this final draft filter. Records were reviewed by 2 different reviewers and the same third reviewer made a final decision where there had not been agreement.

The Medline RAC Search Filter was then translated for PubMed using a validated experimental methodology. This included adding a supplementary ‘textword only’ component to ensure adequate retrieval of PubMed’s non-indexed content (additional and usually more recent literature).

The final PubMed translated RAC Search Filter created as a result of this exercise is shown below, with the ‘textword only’ component indicated in bold:

Nursing homes[mh] OR Nursing home*[ti] OR Long term care[ti] OR Longterm care[ti] OR Homes for the aged[mh] OR Residential aged care[tw] OR LTCF[tw] OR Aged care facilit*[tw] OR Skilled nursing facilities[mh] OR ((Nursing home[tw] OR care home*[tw] OR long term care[tw] OR care facilit*[tw] OR residential care[tw] OR institutionalised elders[tw] OR Institutionalised elderly[tw] OR institutionalized elder*[tw] OR skilled nursing facilit*[tw] OR (resident*[tw] AND dementia[tw])) NOT Medline[sb]) AND English[la]
 

Residential Aged Care Hub Search Filter Advisory Group

  • Ms Marg Adams
  • Mr Peter Jenkin
  • Ms Angie Lo-Faro
  • Ms Trish McReynolds
  • Associate Professor Deb Parker
  • Ms Vicky Yardley

Search filter plus topics

A set of topic searches has also been created using the RAC Search Filter, and where appropriate the Palliative Care Search Filter, to provide access to evidence about specific topics of interest in RAC at the end of life (such as: Assessment tools; Spiritual care; Weight loss; Nurses and care staff).

Contact information: caresearch@flinders.edu.au

How to cite
Tieman, JJ, Hayman, SL (2012). Residential Aged Care Search Filter. Adelaide: CareSearch, Flinders University. Available at:http://www.flinders.edu.au/clinical-change/research/flinders-filters/. [Last Accessed DD Month YYYY]