Designing searches for higher stakes information retrieval: an introduction to ‘search filter’ methodology


By Raechel Damarell, PhD candidate, Research Centre for Palliative Care, Death and Dying

What is ‘higher stakes’ information retrieval?

With the evolution of powerful search engines and user-friendly interfaces, online searching for health information has never seemed easier for health professionals and lay consumers alike (just think ‘Dr Google’). In exchange for the time savings and convenience on offer, we might even be willing to accept that some of the information we get online may not be the best available. However, there remain circumstances where finding the right information—not just some or any information—should be important. Here are three such scenarios.

  1. You are a clinician needing to find the most clinically relevant, high quality evidence available that supports or refutes the benefits and safety profile of a particular therapy.
  2. As a health policymaker, you require evidence of effectiveness to ensure organisational policies and guidance endorse practices shown to maximise health benefits and reduce patient risk and unnecessary costs to the health system.
  3. You are researcher preparing a research grant submission.

When finding the right information is important with a minimum of time and resources, searching can often seem a frustrating hit-or-miss affair.

Searching and higher order thinking 

Belying the ease with which we’ve become accustomed to interrogating the internet, effective search construction should engage higher order thinking. Our degree of success relies on our ability to perform several tasks:

  1. Pinpoint our specific information need
  2. Convert concepts into the words and terms used by experts in the field to describe them
  3. Be technically correct in how we enter search terms in a particular database (and each database is different!)
  4. Critically review the outcome of search efforts and modify our approach if necessary. This means knowing how to broaden the search if we get too few results or focus it further if we are inundated with results that don’t look so relevant.

For clinicians, however, there is the very real problem of finding time to invest in searching efficiently and effectively. One study observed that GPs only look for information to help them with areas of clinical uncertainty when they believe the process will be quick and easy. This meant only half their questions were ever pursued and only two in 1000 questions involved a formal database search. (1)

People also come to searching with different levels of experience and confidence with the technical aspects, or they may be new to a discipline and as yet unfamiliar with the full range of its terminology. Search filters have emerged as one solution to these problems.

Search filters

At its most basic, a search filter is a search strategy that is generally created by specialist librarians using empirical, objective methods, optimised to perform a certain task. This task might be finding studies in a database based on a specific search design such as the randomised controlled trial (i.e. a ‘methodological’ search filter) (2) or for retrieving the most relevant citations on a specific topic to inform clinical practice (i.e. a ‘topic’ search filter). (3) ‘Objective’ methods include processes that ensure search terms are not selected because they seem likely to be relevant, but because there is evidence that they are highly associated with the topic and capable of retrieving well on it.

Another benefit of search filters is that it’s possible to embed all those aforementioned high level cognitive skills of searching into the methodology. Users can therefore put their trust in the rigorous processes underpinning the search filter tool.

Search filters can be embedded within a database for ease of use. The best known example of this would be Clinical Queries, a range of methodological search filters integrated within PubMed.

Search filters and palliative care

The Palliative Care Search Filter, developed by the CareSearch project, was one of the first topic filters to appear. (4) Its purpose was straightforward—to ensure that clinicians in this new and growing field could get to high quality evidence as conveniently and safely as possible, circumventing the problems previously described. This evidence could then be assessed, debated, adapted, applied, and evaluated as part of the process of ‘knowledge translation.’ (5)

CareSearch soon followed this with search filters for other important topics such as heart failure, lung cancer, dementia, residential aged care, and bereavement. As other research organisations took note of what search filters could do for knowledge translation and sought CareSearch expertise, filter development became the province of a new research unit at Flinders University—Flinders Filters.

Fitting a search filter for purpose

If we are honest about what we want our searches to do, we would say we hope for two things simultaneously. That we:

  1. Find as many relevant database citations as possible
  2. Don’t retrieve any irrelevant citations in the process

In information retrieval terms, this means wanting searches with high ‘sensitivity’ (the quality of not missing anything relevant) and high ‘specificity’ (the quality of not retrieving anything irrelevant). In reality, however, searching always involves a trade-off between these two things.

This means any effort to design a search strategy should begin by asking the fundamental question: What are the consequences of missing some relevant citations if it means not having to spend time reviewing lots of irrelevant ones? How people answer this usually depends on their motivation for searching and their resources of time and money.

The cornerstone of the Flinders Filters approach is the engagement of an Expert Advisory Group (EAG) of subject specialists and representatives of the various groups expected to use the search filter. First up, the EAG answers this very question of purpose and desired level of performance. It then goes on to guide filter creation by helping developers identify a ‘gold standard’—a set of highly relevant and representative test citations. It also clarifies important concepts and terminology and advises on the relevance of what the draft filter retrieves.

The result of all of this is a fit-for-purpose filter that brings subject and literature search experts into everyone’s project.

If you are interested in making use of or knowing more about search filters and their development, go to the Flinders Filters website or follow us on Twitter (@FlindersFilters)


  1. Ely JW, Osheroff JA, Ebell MH et al (1999). Analysis of questions asked by family doctors regarding patient care. British Medical Journal 319: 358– 361.
  2. Lefebvre C, Manheimer E, Glanville J. Chapter 6: Searching for Studies. In Higgins JPT, Green S, editors. Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0 [updated March 2011]. The Cochrane Collaboration; 2011. Available at [Accessed 2 July 2020].
  3. Damarell RA, May N, Hammond S, Sladek RM, Tieman JJ. Topic search filters: a systematic scoping review. Health Info Libr J. 2019;36(1):4-40. doi:10.1111/hir.12244
  4. Sladek R, Tieman J, Fazekas BS, Abernethy AP, Currow DC. Development of a subject search filter to find information relevant to palliative care in the general medical literature.J Med Libr Assoc. 2006;94(4):394-401.
  5. Straus SE, Tetroe J, Graham ID. Knowledge in action: what it is and what it isn’t. In: Straus SE, Tetroe J, Graham ID, editors. Knowledge translation in health care: moving from evidence to practice. 2nd Chichester, UK: Wiley; 2013.



Posted in
Digital Health Methodology Research

Leave a Reply