About (DRAFT ver 0.9)

Overview

PICO.vet is designed to help veterinary clinicians, researchers, librarians and others build a well-structured and focused clinical question that interactively searches the biomedical literature. Search terms are refined to support veterinary medicine when compared with the more widely known PICO models (see below) developed for human medicine.

What's a focused clinical question or PICO?

PICO stands for Patient, Population and/or Problem, Intervention, Comparison Intervention and Outcome. PICO.vet is actually a PICOTT+ builder. The "T's" represent "Question Type" (e.g., questions about etiology, diagnosis, prognosis, therapy, etc.) and "Publication Type" (e.g. randomized controlled trial, systematic review, etc.). The "+" represents additional query parameters such as patient signalment (e.g. species, breed, sex) and a filter to restrict results to full-text articles from open access journals. Another approach to PICO is the educational prescription (PDF)

Asking a focused and answerable question is an important initial step in the development of a Clinical Knowledge Summary (CKS) such as a Critically Appraised Topic (CAT), a BestBET (Best Evidence Topic) or a POEM (Patient-Oriented Evidence That Matters). See the excellent "How to Write a Critically Appraised Topic (CAT)" (574KB PDF) by Sadigh G. et al., McMaster University, 2012.

Why this tool?

The PICO methodology has been around for some time. While it's an important evidence-based medicine paradigm, putting PICO into action, especially at the point-of-care, is challenging. Answering clinical questions requires an array of knowledge acquisition skills such as building relatively complex queries within a biomedical literature resource such as PubMed or CABI's VetMed Resource, an understanding of the limitations of various publication types (from case reports to systematic reviews), effective critical appraisal of the literature and of course the time required to plan and execute such a query.

PICO tutorials or text-based "toolkits" are good introductions, but it's uncertain how well they transform clinician behavior towards using PICO-based questions to perform literature searches when they arise directly during patient care. The National Library of Medicine has developed a PICO for handhelds site, but it does not specialize the search sufficiently for veterinary medicine. A PICO query tool ostensibly reduces some of the adoption barriers by making it easier to perform a search at the point-of-care — whether from a mobile, laptop or desktop device.

The author considers this only an intermediate solution. It would be desirable to build queries directly from semantically rich patient information within the electronic health record as anticipatory background tasks, using the Context-Aware Knowledge Retrieval (Infobutton) standard.

See the companion help page to learn more about the structure of a PICO query and guidance on each section of the query form.

Status

This is a developmental βeta release to evaluate user experience and interface design, form content and logic, query rules or parameters (e.g. controlled vocabulary terms and conditional operators), technical requirements and bugs before testing with a wider audience. Integration with other resources, such as CABI's VetMed Resource and Google Scholar is in development.

The responsive UI should work with any device — from smartphone, tablet to desktop browsers. We'll prioritize work on overall functionality before making more detailed refinements to the design.

We'd love your help and expertise. Please help by sending any comments or suggestions to the author or submit a issue or request ticket here.

Query design

The following are some existing design features. These are expected to change - perhaps substantially. Unless stated otherwise, all comments relate to PubMed queries.

  1. The default foundational query (no user input) attempts to constrain the search space to the veterinary literature:
    (veterinary[sb] OR "veterinary"[Subheading] OR "veterinary"[All Fields])
  2. No species are activated by default; one or more may be selected. Currently, the boolean operator for two or more species is limited to "AND". More complex query logic may be included later (e.g. "dogs AND cats" OR horses) if there is justification and demand. For PubMed queries, species are searched using their MeSH term. For example:
    "Dogs"[MeSH] AND "Cats"[MeSH]
  3. There are currently no required fields and no terse form validation. There is a need to resolve which fields should be required (e.g. problem and primary intervention). This is also a training resource, therefore suggestions are sought regarding how required field or inline help text or conditional logic may educate the user. For example, by emphasizing the need to include not only a primary intervention, but a comparison intervention as well. Additional form structure may be helpful to disambiguate the types of problems or interventions for more precise vocabulary encoding to PubMed (e.g. findings vs. diseases vs. conditions).
  4. Question types constrain the query string using prebuilt search terms developed by the NLM (Brian Haynes et. al.) and used in PubMed's Clinical Queries feature. These are broken down by question context (e.g. diagnosis, prognosis, etiology, harm) and by question relevance via recall/precision using sensitive/broad and specific/narrow boundaries.

    Other than the addition of "Quality of Life", these terms are currently exactly as those published here. These have not been thoroughly validated for veterinary medicine. One goal of this project is to refine and publish prebuilt queries more appropriate to this domain.

    For example, for "Diagnosis (sensitive/broad)", the query string is:

    (sensitiv*[Title/Abstract] OR sensitivity and specificity[MeSH Terms] OR diagnose[Title/Abstract] OR diagnosed[Title/Abstract] OR diagnoses[Title/Abstract] OR diagnosing[Title/Abstract] OR diagnosis[Title/Abstract] OR diagnostic[Title/Abstract] OR diagnosis[MeSH:noexp] OR diagnostic * [MeSH:noexp] OR diagnosis,differential[MeSH:noexp] OR diagnosis[Subheading:noexp])
  5. Problem, Primary Intervention and Secondary Intervention are blank and rely on the user to understand their usage. A help/tutorial page is in development. A typical query for an RCT comparing NSAIDs to surgery for chronic pain in dogs over the last 3 years might look like the following:

    ("veterinary"[Subheading] OR "veterinary"[All Fields]) AND "Dogs"[Mesh] AND osteoarthritis AND nsaid AND surgery AND (randomized controlled trial[Publication Type] OR (randomized[Title/Abstract] AND controlled[Title/Abstract] AND trial[Title/Abstract])) AND "Clinical Trial" [Publication Type] AND "last 3 years"[PDat]
  6. The user may constrain the query to publications from the last 1 year, 3 years or 5 years. Question: Is this sufficiently retrospective and granular?
  7. Filtering for open access articles uses the following:
    pmc cc license[filter]

Development Roadmap

The following are some items on the development roadmap:
  1. Toggle constraint to the veterinary literature (on or off, default is on)
  2. Toggle to a "Full" or "Mini" view (default is "Full")
  3. Saved search histories
  4. Meta-search: Searching across multiple resources, as individual literature resource API's permit
  5. Better support for targetted search of open access journals.
For those asking about Google Scholar, Google has yet to provide an Open API and discourages 3rd party front-ends to their service. Regarding Scopus and CABI's VetMedResource, these are commercial, proprietary services without an OpenAPI. PICO.vet currently supports primarily the open source/content community, including open access journals.

Acknowledgements

I'd like to thank Erik Fausak for his inspiration to build a veterinary PICO tool similar to the PubMed for Handhelds PICO search tool developed by the National Library of Medicine. Dr. Virginia Fajt, Heather Moberly, Dr. Moe Milstein, Dr. Alex Sherer and others have provided valuable feedback so far as well. Thank you!

About the Author

Dr. Stuart Turner | I'm a biomedical informaticist (a healthcare information architect) and veterinarian and the past-president of the Evidence-based Veterinary Medicine Association (EBVMA) and the Association of Veterinary Informatics (AVI).

References

  1. Asking an Answerable Clinical Question, from the EBVM Toolkit, RCVS Knowledge.
  2. Asking Focused Questions, Centre for Evidence-based Medicine, University of Oxford
  3. PICO Worksheet (PDF), Centre for Evidence-based Medicine, University of Oxford
  4. Using PICO to Formulate Clinical Questions by Kate Anderson, Zalk Veterinary Medical Library, University of Missouri