1. Shortening the distance to
research evidence
The librarian as knowledge
translation activist
Raechel Damarell
2. Outline
• Search filters
– Research
– Implementation
– Dissemination
– Evaluation
• Flinders Filters: a new research initiative and
opportunity for librarians
3. What is a search filter?
• Tool to retrieve specific subset of literature
• Much more than an expert search
– Developed experimentally following a strict (and
usually transparent) research methodology
– Has a known level of performance in the database for
which it was developed
• The InterTASC Information Specialists' Sub-
Group Search Filter Resource
4. Why use a search filter?
Searching without a filter
5. Why use a search filter?
Searching with a filter
6. Types of search filters
• Methodological vs subject-based
– Cochrane HSSS; PubMed Clinical Queries
• 1st, 2nd, and 3rd generation
– subjectively derived/effectiveness unknown
– subjectively derived/effectiveness known
– objectively derived/effectiveness known
7. Subject-based filters
• Diagnostic imaging techniques
• Heart failure
• Nephrology
• Return to work
• Alcohol-impaired driving
• Sleep
• Ultrasound for DVT
• Knowledge translation
• Cardiovascular biology
• Occupational origin of disease
• Nurse staffing research
• Palliative care
• Patient safety
• Mental health
• Spanish studies
• Measurement properties of
measurement instruments
• Administrative data studies
• Primary health care
• Occupation health interventions
• Evidence based nursing
8. Methodological filters
• Adverse effects
• Diagnostic studies
• Economic evaluations
• Epidemiological studies
• Etiology
• Guidelines
• Health services research
• Observational studies
• Prognosis
• Qualitative research
• Quality of life
• RCTs and other trials
• Systematic review
• Therapy studies
9. Basic filter methodology
• Establish an Expert Advisory Group
– Scoping conceptual boundaries
– Preliminary consensus on potential search terms
– Advice on Gold Standard Set
10. The Gold Standard Set
• Set of representative citations of known
relevance indexed in the database of interest
• Target set: all citations in it should be retrieved
by final filter
• Provides means of iteratively improving a
search strategy
11. Gold Standard Set options
• Journal ‘hand search’ (e.g. Palliative Care Filter)
• Included references of:
– Clinical Practice Guidelines (Heart Failure)
– Systematic Reviews (Lung Cancer)
– All citations from range of specialty journals on topic (PHC)
• Others:
– Delphi study method (Best Supportive Care)
– Set retrieved by a search on the topic
12. Basic filter methodology
• Establish an Expert Advisory Group
• Division of Gold Standard Set
– Term Identification Set
– Filter Development Set
– Filter validation Set(s)
13. Performance measures
Sensitivity (or Recall)
The number of relevant citations retrieved by the search filter
from the gold standard as a proportion of the total number of
relevant citations (100% = all relevant citations retrieved)
Precision
The number of relevant records retrieved as a proportion of the
total number of records retrieved (100% = all citations retrieved
relevant)
Specificity
Number of irrelevant citations not retrieved (correct exclusions)
14. Basic filter methodology
• Establish an Expert Advisory Group
– Scoping conceptual boundaries
– Preliminary consensus on potential search terms
– Advice on Gold Standard Set
• Division of Gold Standard Set
– Term Identification Set; Filter Development Set; Filter
Validation Set(s)
• Implementation: making it available
15. Why PubMed?
• Accessibility: Largest freely available
biomedical database
• Currency: PubMed’s ‘in process’ and ‘as
supplied by publisher’ citations
• Immediacy: Search filter can be saved as
hyperlink for one-click searching
• Implementable: Can be embedded in any
webpage
16. The Palliative Care Filter
• Developed in 2004 based on 1999-2001
dataset
• Specificity = 99% but sensitivity <50%
• Combined with 60+ topics of clinical
importance to palliative care clinicians
• Subject of three published studies (4th
currently being written up)
17. What the users don’t see
((advance care planning[mh] OR attitude to death[mh] OR
bereavement[mh] OR terminal care[mh] OR hospices[mh] OR life support
care[mh] OR palliative care[mh] OR terminally ill[mh] OR death[mh:noexp]
OR palliat*[tw] OR hospice*[tw] OR terminal care[tw] OR 1049-9091[is] OR
1472-684X[is] OR 1357-6321[is] OR 1536-0539[is] OR 0825-8597[is] OR
1557-7740[is] OR 1552-4264[is] OR 1478-9523[is] OR 1477-030X[is] OR
0749-1565[is] OR 0742-969X[is] OR 1544-6794[is] OR 0941-4355[is] OR
1873-6513[is] OR 0145-7624[is] OR 1091-7683[is] OR 0030-2228[is]) AND
Medline[sb]) OR ((advance care plan*[tw] OR attitude to death[tw] OR
bereavement[tw] OR terminal care[tw] OR life supportive care[tw] OR
terminally ill[tw] OR palliat*[tw] OR hospice*[tw] OR 1049-9091[is] OR
1472-684X[is] OR 1357-6321[is] OR 1536-0539[is] OR 0825-8597[is] OR
1557-7740[is] OR 1552-4264[is] OR 1478-9523[is] OR 1477-030X[is] OR
0749-1565[is] OR 0742-969X[is] OR 1544-6794[is] OR 0941-4355[is] OR
1873-6513[is] OR 0145-7624[is] OR 1091-7683[is] OR 0030-2228[is]) NOT
Medline[sb]) AND English[la]
18. The Heart Failure Filter
• Developed in collaboration with Centre for
Cardiovascular & Chronic Care (UTS)
• Gold standard: Evidence based guidelines
• Sensitivity: 97%; Precision: 75%
• Methodology article published in BMC Medical
Research Methodology
• Innovative methodology for PubMed translation
(article in progress)
• Combined with 40+ topics of clinical importance to
heart failure/palliative care clinicians
19.
20. The Heart Failure Filter
• Developed in collaboration with Centre for
Cardiovascular & Chronic Care (UTS)
• Gold standard: Evidence based guidelines
• Sensitivity: 97%; Precision: 75%
• Methodology article published in BMC Medical
Research Methodology
• Innovative methodology for PubMed translation
(article in progress)
• Combined with 40+ topics of clinical importance to
heart failure/palliative care clinicians
21. What the users don’t see
((heart failure[tw] OR ventricular dysfunction,
left[mh:noexp] OR cardiomyopathy[tw] OR left
ventricular ejection fraction[tw]) AND Medline[sb]) OR
((heart failure[tw] OR left ventricular dysfunction[tw] OR
cardiomyopathy[tw] OR left ventricular ejection
fraction[tw] OR cardiac resynchronization[tw] OR LV
dysfunction[tw] OR left ventricular systolic
dysfunction[tw] OR left ventricular diastolic
dysfunction[tw] OR cardiac failure[tw]) NOT medline[sb])
AND english[la]
22. Palliative Care Filter effectiveness studies
Study 1
• Compiled more recent gold standard based on
included studies in 29 palliative care systematic
reviews
• Found that Palliative Care Filter performed better in
more recent dataset than in original one (˃ 5%)
23. Palliative Care Filter effectiveness studies
Study 2
• Real life PubMed searches of pc clinicians for pc
literature (n=37) analysed
• Each run in gold standard set and retrieval compared
with that of PCF
• Results highlighted difficulties in searching:
– Very limited number of search terms used for complex topic
– Incorrect search construction
– Negative correlation between stated confidence and actual
performance (i.e. overconfidence is issue)
– Search filter retrieved 68% of gold standard; average
clinician search retrieved 25%
24. Palliative Care Filter effectiveness studies
Study 3
• Were palliative care clinicians aware of topic
searches built on the PCF?
• Of those aware:
– Were they using them? If yes, what did they think? If no, why
not?
• If not aware:
– What did they think after using search? (intervention study)
• Follow up study
– Were they still using searches after 6 weeks?
26. Flinders Filters
• Established in 2012 as Flinders co-
funded research venture
• Committed to producing over 2 years:
– Two health filters
– Two non-health filters
– Four research papers that make a
contribution to search filter methodology
29. Flinders Filters
New projects:
• ‘Indigenous health’ and ‘Australian Aboriginal health’ filters
(Lowitja Institute)
• Best Supportive Care (Duke University)
• Dementia
• Sexual health
30. Flinders Filters
Urgent issues for FF:
• Non-health filters: What? Who with? Where
appropriate?
• Finding time to write for publication
• More librarians needed to join team