Presentation on theme: "Systematic Searching of the Literature Canadian Partnership Against Cancer: Atlantic Clinical Practice Guideline Workshop October 16 and 17, 2008 Halifax,"— Presentation transcript:
Systematic Searching of the Literature Canadian Partnership Against Cancer: Atlantic Clinical Practice Guideline Workshop October 16 and 17, 2008 Halifax, Nova Scotia Seana Collins MA, MLIS email@example.com Librarian Educator Capital District Health Authority
Systematic Searching The systematic identification of evidence is an essential step in clinical guideline development. (NICE Guidelines Manual, April 2008, p. 48)
Systematic Searching There are several steps involved in developing and executing a literature search: 1. Formulate an “answerable” question - PICO 2. Identify key search concepts 3. Structure the search strategy using Boolean Logic 4. Consider the question domain: therapy? diagnosis? prognosis? etiology? 5. Consider the type of information required to answer the question: background information? foreground information? 6. Consider the sources of evidence: (core databases, subject databases etc.)
7. Run a preliminary or “scoping” search (What are the existing guidelines?) 8. Execute a comprehensive literature search (database searching, sources of guidelines, subject specific databases) 9. Evaluate the search results: too broad? too narrow? relevant? 10. Refine the search results using appropriate LIMITERS: age group, gender, publication date range, study design etc. 11. Document search results 12. Manage and capture search results (RefWorks, EndNote, ProCite, Reference Manager, etc.) Systematic Searching
1. Formulate the Question Begin by composing a clear, well stated question. Ambiguous questions will often lead to ambiguous search results. Write down your research topic or question Underline or circle the most important concepts These will be your search terms Example Question: Use of marijuana for treating cancer. Formulate the research topic into an “answerable” question using PICO.
PICO PICO is a tool used in evidence base research methods to help identify key concepts and articulate an “answerable question”. P - Population (patient) What is the patient group or condition being assessed? I - Intervention What is (are) the interventions being considered? C – Comparison Intervention Is there a comparison being considered? O - Outcome What is the expected outcome?
PICO Our Research Topic: Use of marijuana for treating cancer. Population = cancer patients Who are the patients of interest? Is there a particular age group, gender or population? What is the health concern? Intervention = marijuana What therapeutic, diagnostic, preventive or other health care interventions are you interested in knowing more about? What health care management strategies are you interested in comparing? Comparison = none Is there a comparison to be evaluated against the intervention? Only used if more than one intervention, or if no intervention is a factor. Outcome = improved quality of life What is the desired outcome to be evaluated? How will the patient or population be affected, or not affected, by the intervention?
1. Formulate the Question Our Question: Use of marijuana for treating cancer. “Answerable” Question: Is marijuana an effective treatment in improving the quality of life for cancer patients? Using PICO to formulate a question can help to identify concepts that can be used to structure your search strategy.
2. Identify Key Concepts Identify each specific concept in the question to be searched: Using our example our search concepts would be: treatment and cancer and marijuana One could simply enter these terms into a database and search: treatment AND cancer AND marijuana; however, this would mean: any papers where the term neoplasms is used in place of cancer would be missed any papers where the the terms: treat or treatments or therapy is used in place of treatment would be missed You would also retrieve papers that have a passing reference to the topic and are not useful (not specific or relevant)
2. Identify Key Concepts Use appropriate thesauri or controlled vocabularies, such as MeSH (PubMed/Cochrane) or other subject listings to translate individual concepts into valid subject headings. (EMTREE Headings, CINAHL Headings) Verify if any of your subject terms can be translated to MeSH subheadings. If an appropriate subject heading can not be found, consider doing a keyword search. When keyword searching, remember to use truncation, when keyword searching for compound words or phrases use quotations i.e. “health services”. Truncation: Symbols which instruct the database to find all word variants and endings (*,$,#). Example: surg$ will retrieve: surgery, surgeon and surgical.
3. Construct the Search Strategy Search terms or “search statements” are organized by concepts or terms which are then combined using Boolean Operators: AND, OR, NOT. It is very helpful to write out your search strategy with the relevant Boolean Operators (OR, AND, NOT) before executing the search in the database. Example: cancer OR neoplasms AND treatment OR therapy AND marijuana Identify similar terms or synonyms Connect search terms using Boolean Operators Consider phrase searching, some databases such as PubMed and Cochrane Library will search for phrases. Put the terms in “quotation marks” to indicate a phrase: example: “sleeping sickness”
Boolean Logic ‘ Boolean Logic’ uses the logical relationships OR, AND, NOT, to create relationships between concepts or subject headings. Understanding and application of Boolean logic correctly has become fundamental to all online searching. OR AND NOT
3. Construct the Search Strategy It is often very helpful to write out your concepts using a table to assist in organizing your search: Your Conceptmarijuanacancertreatment MeSHcannabis OR neoplasms OR therapeutics OR keywordsweedcarcinoma*therap* Possible search strategy: (neoplasms OR carcinoma*) AND (therapeutics OR therap*) AND (cannibis OR weed)
4. Consider Question Domain Consider the question domain or the clinical study category of your question, this will assist you in assessing the results of your search (applicability). therapy question? Clinical studies that discuss the treatment of diseases diagnosis question? Clinical studies addressing disease diagnosis prognosis question? Clinical studies addressing disease prognosis etiology question? Clinical studies addressing causation/harm in disease and diagnostics
4. Consider Question Domain The consideration of question domain will assist you in identifying the type of clinical studies that would best answer your question. (RCT, meta-analysis, systematic review, cohort study, practice guideline, etc.) In addition it will assist you to identify key appropriate resources to search for literature on your question. Some information resources are stronger in terms of therapeutics (Cochrane Systematic Reviews) others may be stronger in diagnostic questions (Health Technology Assessment Database).
5. Type of Information Consider the type of information or evidence you are looking for. Is it background information? If so, it is likely that you will need to search/consult a textbook. (Harrison’s Principles of Internal Medicine, MD Consult, 5 Minute Clinical Consult, or a Library Catalogue) Is it foreground information? If so, you will need to search the primary literature, it is likely that you will have to conduct a literature search in a bibliographic database. (PubMed, Cochrane, EMBASE etc.)
6. Resources What are considered to be the core resources for your subject area? are these resources available to you? Identify the subject database most relevant to your research question. What type of database is it? bibliographic? textual? (fulltext) Review how each database works, does it use controlled vocabulary? Consult a Librarian and/or your institution’s Library for a list of resources Guidelines: Are there established guidelines for specific resources that must be consulted for the type of study you are conducting? i.e. Cochrane Manual (systematic reviews), NICE Guidelines Manual, CMA Guidelines Manual, ADAPTE Manual (guideline adaptation)
6. Sources of Evidence Core Databases: Cochrane Database of Systematic Reviews – CDSR (Cochrane reviews) Database of Abstracts of Reviews of Effects – DARE (other reviews) Cochrane Central Register of Controlled Trials – CENTRAL (clinical trials) Health Technology Assessment (HTA) database MEDLINE/MEDLINE In-Process (PubMed) EMBASE CINAHL (Cumulative Index to Nursing and Allied Health Literature)
6. Sources of Evidence Subject Specific Databases: PsycINFO AMED ERIC PEDro AGELine Other Sources of Evidence: Guidelines: i.e. National Guidelines Clearinghouse Controlled Trial Registries UK Clinical Research Network National Research Register (NRR) Archive Web of Science / Scopus Conference Papers and Dissertations Indices
7. Preliminary Search It is essential to run a preliminary search to determine if your topic is well published and to determine the level of published evidence available on your topic. If you are conducting a systematic review of RCTs for example, and only one RCT has been published on your topic, it may impact on the validity of your study. Run the preliminary search in the most comprehensive and relevant database to your topic. (Within biomedicine, and for English speaking researchers, this will likely be PubMed.) Most importantly a preliminary literature search will allow you to review search results and determine whether or not your initial search strategy needs to be adjusted.
7. Preliminary Search Based on the preliminary or scoping search you may decide to limit your search results to a specific population group, or Intervention. Consider: Checking the indexing of the relevant articles retrieved in your search results for additional subject headings or terms that can be incorporated into your search strategy. Scanning the abstracts of the results to determine whether you have used the appropriate subject term(s) and relevant key words. A preliminary search will also save you time and effort in terms of sensitivity versus precision.
Sensitivity versus Precision It is important that searches for systematic reviews attempt to identify all the relevant literature. However, there needs to be a trade-off between conducting an exhaustive search that will require additional resources versus undertaking a more modest search that may miss some studies. The preliminary / scoping search can help to identify key studies for a review question that can assist in balancing the search sensitivity.
8. Execute Full Search Start by executing a search using broad subject terms. Use key words, including all relevant synonymous terms and word variants in combination with your subject terms. The goal is to be as comprehensive as possible. You can refine and focus the search after you have a set of broad results to limit. Use limiters and inclusion / exclusion criteria to focus and refine your search.
Refine the Search To decrease search result numbers, consider the use of limiters. Most databases allow users to limit or refine search results to specific criteria. Some examples include: Publication date(s) range Language Publication type (i.e. Review, Letter, Comment etc.) Age group Gender (i.e. Male / Female) Study-design (i.e. systematic review, guideline, randomised controlled trial) Using limiters to refine your search can provide you with a method to review a manageable number of abstracts, especially if you are searching a very large database such as PubMed.
9. Evaluate Search Results Evaluate your search results, are the results what you expected based on your research question? Is the evidence reliable? valid? Are the results applicable to your patient population? If your search results are satisfactory utilize the same concepts to expand and build on your search results using other subject specific databases. Basic literature searching skills are transferable from one resource to another. The basic principles are the same it is the controlled vocabulary and the resource search interface that may differ from database to database.
9. Evaluate Search Results If your search results are not satisfactory consider the following: Re-assess your search question: Have you used the best subject terms to represent the search concepts? For example: perhaps a broader subject term should be used to capture more results? Have you used a keyword search in combination with your subject search to represent concepts? Are your results too narrow? Are your limiters appropriate? For example, have you excluded useful search results by applying too restrictive limits: publication date range? age group? study-design? etc. Are your results too broad? for example should you exclude similar subject terms that are often associated with one another. For example: glucosamine NOT condroitin dementia NOT Alzheimer’s Disease
9. Evaluate Search Results Have you used the most appropriate databases or resources for your search? Have you searched all available sources of guidelines? Consult an expert, ask a Librarian or members of your Guidelines Group to review your search strategy.
10. Document the Search Literature searches must be thorough, transparent and reproducible. It is essential that literature search methods be documented. Provide the search strategies used for each database so that they can be replicated.
10. Document the Search Provide details of which databases you searched (e.g. MEDLINE, EMBASE, CENTRAL, etc.) Ensure that you record when you searched the database this might include: The name of the database searched The name of the host/system/vendor/interface used The date when the search was run The years covered by the search (all Limiters used) Filename for search strategy (e.g. Embasestrategy.txt) Filename for search results (e.g. Embaseresults.txt)
11. Manage Search Results Once you have completed the literature search, ensure that all of your search results have been saved, emailed, printed, or exported. Each resource will provide this functionality. Note: If the databases you have used provide user accounts or registration, it is a good practice to take advantage of this option. It will allow you to save search strategies and results for future use, This will also allow you to view your search results at any time to refine, adjust, or update and your search.
11. Managing Search Results Capture your research results using bibliographic management software such as RefWorks,RefWorks EndNote, ProCite, Reference Manager. This will allow you to archive your references, produce a formatted bibliography or list of references in any citation style you may require.
11. Managing Search Results In addition to archiving references, reference management software can be used for: Coding the references with additional information Providing links to the full text of articles Tracking the ordering and/or receipt of documents Linking to word processing software output styles to automatically cite while writing. It is essential to document and manage all search results in order to maintain transparency and the ability to replicate search results.
Exercise #2: Clinical Scenario: A middle aged woman presents in family clinic complaining of severe joint pain, stiffness, and swollen hands and feet. She has a family history of osteoarthritis and wants to know if taking glucosamine supplements will improve her pain and stiffness. What is the question?
“Answerable” Question: In adult patients with osteoarthritis, is glucosamine effective in reducing pain and improving function? Exclusion Criteria: NOT chondroitin Possible Limiters: middle aged women (age group), systematic review, guideline, RCTs