How to Craft a Search Strategy to Answer Clinical Questions
This is part two of two posts to foster your ability to systematically frame your questions to craft a search strategy that will offer you the best chance at finding the relevant and valid evidence you need to answer those questions, in the most time-efficient manner possible.
The first post was about identifying clinical questions and knowledge needs, creating a well-built clinical question, and identifying key terms. This post builds on the information provided in the first post and offers you search strategy tips for building a search strategy plan.
Where We Left Off: Key Terms
Key terms of the PICOT statement we used in the previous post are bolded:
In adult patients with arthritis (P), does treatment with acupuncture (I) decrease the amount of pain (O) during flare-ups (T) as compared to standard medical therapy? (C). Randomized controlled trials (S)
As you can see, the PICO above is modified with a search strategy aimed at a particular study design – RCTs in this case. Acupuncture is a therapeutic intervention, so a filter of therapy would also be an appropriate study design search strategy, giving you a broader sample of all kinds of designs for therapeutic intervention studies. You could change the S to Therapy which could then be qualified with high sensitivity or high specificity or best balance (of the two). Therapy studies would include RCTs, but also quasi-experimental, correlational, and descriptive studies. If you want only RCTs you can check that box in the Advanced Search query in CINAHL or use PubMed’s study filters.
Search Strategy “Formulas”
Different types of questions require different study designs. Since the wording of your PICOT question will provide the key terms for your search strategy, it’s a good idea to get used to how a question for a therapeutic intervention would be worded as compared to a question about patient experiences. Luckily, there are “formulas” for identifying the key PICOT terms to place in your search strategy statements.
One of the most helpful resources, when I was first learning how to create PICOT questions and identify key terms for my search strategies, was Melynk and Fineout-Overholt’s Question Templates in their textbook titled Evidence-Based Practice in Nursing & Healthcare: A Guide to Best Practice. Now in its third edition, this text continues to be one of the premier EBP texts for nursing students and practicing nurses. I highly recommend it as a good all-around EBP text.
The Question Templates (2015, p. 30) were helpful in getting my mind to adjust to different ways to ask a question about therapy or etiology or patient meaning without forgetting all of the PICOT components. The P, I, and O components are required or necessary for a good search plan; the C component is assumed for many search strategies and the T and S components are not always used, either. So the search strategy formulas I’m presenting here are based on these templates. See the table below – click on it to save it or enlarge it. Fill in the blanks with PICOTS components, as appropriate.
PICO examples for questions about:
Therapy: In patients with isolated (single vessel) proximal left anterior descending coronary artery disease (P) does percutaneous coronary interventions with drug-eluting stents (I) decrease morbidity, mortality, MI, and stroke (O) within 30 days (T) as compared to coronary artery bypass graft (CABG) surgery(C).
Etiology: Are adolescents (P) who use electronic cigarettes (I) at increased risk for COPD (O)? (The C is assumed and would be “compared to those who don’t use e-cigarettes.”)
Diagnosis: In patients presenting with dyspnea in the ED (P), is impedance cardiography (ICG) (I) more accurate in diagnosing HF (O) as compared to B-type natriuretic peptide assay (BNP) (C)?
Prognosis: In patients presenting with dyspnea in the ED (P), how do BNP levels (I) predict subsequent adverse cardiac outcomes? (O)
Meaning: How do family members (P) of patients in the intensive care unit (I) perceive the experience (O) of having a family member admitted to ICU?
Tips for Forming Search Strategy Statements
Now, of course, you can always ask your healthcare librarian for assistance creating search strategies and searching for the literature. But I’ve given you some explanations of typical search terms and tips for when you do this yourself. (And you should also start with pre-appraised evidence and not use search engines until you have to – see the end of this post for that reminder.)
When you have decided on the key variables to search for, you can use the key terms in the database search box with the word “AND.” For example, arthritis AND acupuncture AND pain AND flare-up. The word “AND” is a way to create a relationship between the keywords and is called a Boolean operator.
What are Boolean Operators?
“Boolean Operators are simple words (AND, OR, NOT or AND NOT) used as conjunctions to combine or exclude keywords in a search, resulting in more focused and productive results” (Alliant Libraries, n.d.).
Using the word “AND” will reduce the number of documents returned in your results because it will look for the literature that contains ALL of the keywords; if one is missing, then articles with the missing keyword won’t be in the search results.
Conversely, using the word “OR” will expand the number of results because it will retrieve articles with either one of your keywords (Alliant Libraries, n.d.).
The use of the term “NOT” will limit your search and narrow your results, but perhaps missing relevant articles. Likewise, using parentheses will “customize your results to more accurately reflect your topic” because the terms within the parentheses will be searched first and then those articles will be searched for the terms outside of the parentheses (Alliant Libraries, n.d.). For example, searching on (arthritis OR joint disease) AND acupuncture will give you results for arthritis and acupuncture; joint disease and acupuncture; and arthritis and joint disease and acupuncture. But the results will not have any articles without the term acupuncture in the results list.
But remember, making a search too narrow may drastically decrease the results list — perhaps finding no articles. So it’s okay to narrow your search, but review the results list and see if you might need to adjust your strategy because of too few (or no) results.
Also, limit or filter your results, to adult patients in this case (e.g., In CINAHL look under Age Groups: All Adult). I got no results with the initial search strategy with all of the keywords, so I removed the Timeframe keyword from the search terms and got 31 results.
Once you get the results list, you’ll have to decide if all of the articles returned are relevant to your question, of course.
More Search Strategy Tips: MeSH Terms, Filters, and Limiters
You could also take your keywords and identify terms that are called standardized or controlled vocabulary and used by the search database. CINAHL uses major and minor subject headings as its standardized vocabulary (Adorno et al., 2016).
Medical Subject Headings or MeSH terms were developed by the National Library of Medicine (NLM) for indexing biomedical MEDLINE records (PubMed Help, 2017, May). MeSH terms are alternative descriptors of your key search terms that are standardized, controlled, and searchable. You can use these terms to search PubMed MEDLINE or any database that uses MeSH terms. For example, MeSH descriptors for the key term “arthritis” include Infectious Arthritis, Reactive Arthritis, Juvenile Arthritis, Psoriatic Arthritis, Rheumatoid Arthritis, Gouty Arthritis, and Osteoarthritis.
The point is MeSH terms or other standardized vocabularies to make it easier to index the millions of biomedical citations. Each article in MEDLINE or CINAHL has MeSH terms or subject headings (respectively) associated with it, which allows you to create a specific search strategy. The more specific you can be, the greater likelihood of finding articles relevant to your question.
When I review my search results in CINAHL via EBSCOhost, I’ll pull up the complete reference for articles that are relevant to my question. The detailed record will list either MeSH terms or major and minor subjects terms that I can then use to substitute terms in my general search strategy. To do this, click on the hypertexted title of the article. Among other information about the authors and article, you’ll see Major Subjects and Minor Subjects specific to the article listed.
CINAHL also has an option called SmartText Searching. “Enter as much text for your search as you want – a phrase, a sentence, paragraph, or even whole pages.” SmartText Searching is only available for EBSCO Databases.
To search for a specific concept or key term in PubMed, click Clinical Queries or Topic-Specific Queries from the PubMed homepage under PubMed Tools or click on the MeSH Database listed under the More Resources tab.
Other tools to narrow your search include the use of filters (PubMed MEDLINE) and limiters (CINAHL) (Adorno et al., 2016). Both of these tools do the same thing — filter or limit the search results to categories and subheadings of your choice. Search results can be narrowed by language, age group, gender, human or animal, geography, publisher, journal, full text available, publication date, publication type, research article vs continuing education (CE), major subjects, minor subjects, and clinical queries to name a few!
When you click Advanced Search in the CINAHL database (EBSCOhost) you will find the categories for which you can filter or limit the results. Clinical queries can help you narrow your search. Clinical queries are the same as the S part of the PICOTS mnemonic. If you can identify the type of study design likely to answer your clinical question, you can filter or limit your results to that particular research design. These filters are further delineated by qualifying the study design with either a broad search (sensitivity) or a specific and narrow search (specificity).
In CINAHL, you can choose from the following clinical queries to limit your search: Therapy-High Sensitivity,
Prognosis-High Sensitivity, Prognosis-High Specificity, Prognosis-Best Balance,
Review-High Sensitivity, Review-High Specificity, Review-Best Balance,
Qualitative-High Sensitivity, Qualitative-High Specificity, Qualitative-Best Balance,
Causation (Etiology)-High Sensitivity, Causation (Etiology)-High Specificity, Causation (Etiology)-Best Balance.
ou can also request only Randomized Controlled Trials (RCTs).
In PubMed, clinical queries can be filtered by clinical study categories: therapy, diagnosis, etiology, prognosis and clinical prediction guides.
In PubMed, results can also be filtered by systematic reviews: “This strategy is intended to retrieve citations identified as systematic reviews, meta-analyses, reviews of clinical trials, evidence-based medicine, consensus development conferences, guidelines, and citations to articles from journals specializing in review studies of value to clinicians. This filter can be used in a search as systematic [sb]. Example: exercise hypertension AND systematic [sb]”(PubMed, 2017, February). Medical genetics filters contain the categories of diagnosis, differential diagnosis, clinical description, management, genetic counseling, molecular genetics, and genetic testing.
Directions for how to formulate your search using these different filters are found in detail at the PubMed site.
Don’t Forget to Start Your Search at the Top!
Once you have your clinical question formulated you can start your search. Textbooks are good informational sources to answer background questions, while the sources to answer foreground questions are research and literature databases. Foreground questions use the PICOT format.
Remember that to search for answers to clinical questions go FIRST to the 6S pyramid to start your search at the top with pre-appraised evidence. The Systems level is at the top of the pyramid, but that level refers to evidence-integrated institutional systems and most nurses don’t have access to this level and it’s not appropriate to use for school, research, or scholarly research.
Therefore, start with the Summaries level and look for clinical practice guidelines (see my post on where to find CPGs). If you can’t find valid evidence for your question at the top level of Summaries, then move down to Synopses of Syntheses and search data abstraction journals (e.g., ACP Journal Club, Evidence-Based Nursing. Move on to Syntheses, etc. CINAHL and PubMed and other search engines are lower on the 6S hierarchy because these databases do not provide pre-appraised evidence — YOU have to critique the articles you retrieve.
Adorno, M., Garbee, D., & Marix, M. L. (2016). Improving literature searches. Clinical Nurse Specialist, 30(2), 74-80.
PubMed Help. Search strategy used to create the systematic reviews subset on PubMed [Internet]. [Updated 2017 February]. Bethesda (MD): National Center for Biotechnology Information (US). Retrieved from: https://www.nlm.nih.gov/bsd/pubmed_subsets/sysreviews_strategy.html
PubMed Help [Internet]. [Updated 2017 May 25]. Bethesda (MD): National Center for Biotechnology Information (US). Retrieved from: https://www.ncbi.nlm.nih.gov/books/NBK3827/