Home > RNJ > 2008 > July/August > Searching the Literature for Evidence

Searching the Literature for Evidence
Linda Ehrlich-Jones, PhD RN Linda O'Dwyer, MSLIS MA Kathleen Stevens, PhD RN CRRN Anne Deutsch, PhD RN CRRN

Nurses are expected to use evidence-based knowledge when planning and caring for patients and their families. For nurses who practice evidence-based care, knowing how to effectively search the literature for evidence is an important skill. The objectives of this article are to provide an overview of key sources of evidence and offer instruction on how to do an effective MEDLINE search in PubMed and how to critically review research articles. A sample literature search is performed using the example case study question, “How effective are compression stockings in preventing deep vein thrombosis or pulmonary embolism for thrombotic stroke patients who are on an oral anticoagulant regimen?”

Nurses are expected to use evidence-based knowledge when planning and providing care to patients and their families. Evidence-based practice emphasizes conscientious and judicious decision making based on current research and consideration of the patient’s goals and wishes (Sackett, Straus, Richardson, Rosenberg, & Haynes, 2000). Stettler and colleagues (1998) further specified that “evidence-based nursing deemphasizes ritual, isolated, and unsystematic clinical experiences, ungrounded opinions, and tradition as a basis for nursing practices, and stresses instead the use of research findings, and, as appropriate quality improvement data, and other operational and evaluation data, the consensus of recognized experts, and affirmed experience to substantiate practice” (pp. 48–49). For nurses who practice evidence-based care, knowing how to effectively search the literature for evidence is an important skill.

The objectives of this article are to provide an overview of key sources of evidence and offer instruction on how to do an effective MEDLINE search in PubMed and how to critically review research articles.

Patient Case Study

DB is a 57-year-old African-American man with a 10-year history of hypertension who was admitted to the inpatient rehabilitation unit with left-sided weakness following a thrombotic stroke. He takes enoxaparin (Lovenox®) 40 mg subcutaneous daily and has worn thigh-high compression stockings to prevent deep vein thrombosis for 3 weeks since the onset of his stroke. DB is progressing with his rehabilitation; however, his ability to independently change the lower extremity dressing is hampered by difficulty applying and removing the compression stockings. It is unclear whether compression stockings in addition to oral anticoagulant therapy are necessary for deep vein thrombosis prevention poststroke.

Sources of Evidence

Key sources of evidence include original research and reviews published in journals, which can be identified through databases such as PubMed, Cumulative Index to Nursing and Allied Health Literature (CINAHL), and the Web of Science. PubMed (http://pubmed.gov) is a Web-based search and retrieval system that provides free access to MEDLINE, the National Library of Medicine’s database of citations and abstracts in the fields of medicine, nursing, dentistry, veterinary medicine, healthcare systems, and preclinical sciences. PubMed contains links to some full-text articles. The database contains more than 17 million citations dating back to the 1950s. Coverage is worldwide, but most records are from English-language sources or have English abstracts.

CINAHL (www.cinahl.com) is another database that provides citations and abstracts (e.g., journals, books, pamphlets, dissertations, audiovisuals, software) for nursing and allied health disciplines and also covers consumer health, biomedicine, and alternative therapy. The database has full-text articles from selected state nursing journals, standards of practice, practice acts, critical paths, research instruments, and government publications. CINAHL is updated monthly and contains more than 800,000 records and coverage from 1982 to the present.

Web of Science (http://scientific.thomson.com/products/wos) is a multidisciplinary database that permits users to navigate forward and backward through the literature. For a given reference, a user is provided with links to the abstracts of the manuscripts referenced by the authors as well as the manuscripts that cite the original reference. This database contains 36 million citations from more than 9,000 journals.

Both CINAHL and Web of Science may be accessed through the Internet with an institutional subscription. Nurses may have access through a hospital’s or medical center library’s subscription.

Not all questions can be answered by research studies; expert consensus may be the best evidence available. Many government agencies, specialty practices, and academic organizations have made best-practice guidelines available online. Ideally, guidelines should be well researched, peer reviewed, and endorsed by clinical experts or organizations that have experience with the patient population of interest. The National Guidelines Clearinghouse (http://guideline.gov), an initiative of the Agency for Healthcare Research and Quality, is a public resource created to provide physicians, nurses, and other healthcare professionals access to objective, detailed information on clinical practice guidelines.

Key international sources for guidelines include the evidence-based reviews reported by the Cochrane Collaboration (www.cochrane.org) and the Joanna Briggs Institute (www.joannabriggs.edu.au). Although health systems and language may differ, the clinical problem is often a universal concern. Searching international sources may alert healthcare providers to interventions not commonly considered in the United States.

Practice guidelines that specifically focus on the prevention and treatment of deep vein thrombosis are available from a number of sources. When evaluating literature reviews, consider the peer-review process and the clinical focus of the organization. For example, the American College of Chest Physicians’ guidelines focus on the prevention and treatment of venous thromboembolism (Geerts et al., 2004), whereas, the guideline Best Practices: Preventing Deep Vein Thrombosis and Pulmonary Embolism addresses a broader scope (Center for Outcomes Research, University of Massachusetts Medical Center, 1998). Medical reviews may focus more on diagnostic, surgical, and pharmacological interventions. In contrast, the evidence-based review published by the Joanna Briggs Institute limits the review to the use of “graduated compression stockings for the prevention of postoperative venous thromboembolism” (Joanna Briggs Institute, 2001, p. 1). Although this review addresses the use of compression stockings in the prevention of venous thromboembolism, the patient population is broader and may present with different risk factors than a rehabilitation patient. In addition to the quality of the publication, it is equally important to consider the relevance to the clinical question of interest and to the patient population.

Searching the Literature Using PubMed

As noted previously, PubMed is a free resource that may help answer clinical questions. The following describes the process of searching PubMed.

Choosing Your Search Terms

Often, novice users begin a search by typing in the first thing that comes to mind. However, the most successful literature searches contain at most two or three key search concepts, and most failed or incomplete literature searches are a result of using inappropriate search terms or failing to specify alternate search terms (Doig & Simpson, 2003). Before embarking on a search, take the time to make a list of some relevant search terms. You may not use all the search terms, but knowing them in advance can help focus the search.

Writing a focused clinical question based on a specific clinical problem may facilitate searching the literature (Richardson, Wilson, Nishikawa, & Hayward, 1995) and can be developed using the PICO model (Stone, 2002). Table 1 shows how the PICO model can be applied to DB’s case study.

The PICO model helps identify a focused clinical question: For thrombotic stroke patients who are on an oral anticoagulant regimen, how effective are compression stockings in preventing deep vein thrombosis or pulmonary embolism? This exercise leads to a list of relevant keywords: thrombotic, stroke, oral anticoagulant regimen, effective, compression stockings, deep vein thrombosis, and pulmonary embolism.

Next, construct a likely search strategy based on some or all of the search terms. It is usually better to begin with broad terms. Here are some possible examples:

  1. (deep vein thrombosis OR pulmonary embolism) AND compression stockings AND anticoagulants
  2. (deep vein thrombosis OR pulmonary embolism) AND compression stockings AND rehabilitation
  3. (deep vein thrombosis OR pulmonary embolism) AND compression stockings AND stroke

Note that we are searching “deep vein thrombosis OR pulmonary embolism” because we would like to find articles on either term. It is possible that an article may discuss deep vein thrombosis but not mention pulmonary embolism and vice versa.

On the PubMed home page enter each of these search strategies into the query box at the top of the page and click “Go.” These search strategies produced 176, 29, and 13 articles respectively (searched July 31, 2007). One hundred seventy-six is quite a lot of articles, but 29 and 13 are more manageable. Scanning the set of 29 articles (that are restricted to rehabilitation), some seem relevant, but there are a few articles that are off topic. The set of 13 articles (that are restricted to stroke) are most relevant to DB’s case study, but how can users be assured to see the full picture?

Searching with Medical Subject Headings

Precision searching is facilitated by using medical subject headings (MeSH) terms, the official language of MEDLINE. MeSH terms provide a consistent way to search all articles on a topic where different terms may have been used to describe similar concepts. For example, the MeSH term for “stroke” is “cerebrovascular accident.” By using this term to search MEDLINE, you don’t have to search for alternative terms or synonyms such as “stroke”; “cerebral stroke”; “vascular accident, brain”; or “cerebrovascular accident.” Usually the PubMed search system will translate the search term (e.g., “stroke”) to the MeSH term (e.g., “cerebrovascular accident”); however, a working knowledge of the MeSH database is helpful.

Performing a Search in PubMed

Depending on the literature database, the search term will be searched exactly as is or will be translated into a preferred search terminology, such as MeSH. Searches in PubMed actually do both. When searching on the PubMed home page for “(deep vein thrombosis OR pulmonary embolism) AND compression stockings AND anticoagulants,” PubMed attempts to translate the search terms into MeSH terms. PubMed translates the first search statement into (venous thrombosis [MeSH terms] OR deep vein thrombosis [text word]) OR (pulmonary embolism [MeSH terms] OR pulmonary embolism [text word]) AND (stockings, compression [MeSH terms] OR compression stockings [text word]) AND (anticoagulants [MeSH terms] OR anticoagulants [pharmacological action] OR anticoagulants [text word]).

Deep vein thrombosis is translated into “venous thrombosis” as a MeSH term, but the phrase “deep vein thrombosis” is also searched as a text word phrase in the title or abstract. This automatic term-mapping feature of PubMed tries to find the official term used by the indexers and search the term used in the title or abstract. To see how the search is performed click “Details” on the PubMed results page.

Using the MeSH Database

On the PubMed home page, click MeSH Database on the left side of the screen. The MeSH Database allows users to identify likely search terms and combine them to produce more relevant results. Subheadings can be applied to focus a search for further precision. To use the MeSH Database effectively, search for one term at a time. If “deep vein thrombosis” is entered as the first term, MeSH produces a number of possible terms. The correct term is “venous thrombosis,” which is defined as “the formation or presence of a thrombus within a vein.” Notice that the term itself is a blue hyperlink. When this hyperlink is clicked, a more detailed treatment of the subject heading “venous thrombosis” appears. From here users can choose subheadings, restrict the search to Major Topic Headings only, or check a box so that the search won’t be exploded. Scrolling further, users will see a list of Entry Terms and finally a hierarchical arrangement of terms that includes “venous thrombosis.”

Subheadings. By choosing one or more relevant subheadings, users can exclude information that is not pertinent to their current search. In searching terms for DB’s case study, the specific topic of interest is the prevention of deep vein thrombosis, so the “prevention and control” subheading would be an appropriate choice. Other choices could be “therapy,” “nursing,” or “rehabilitation.” Although it is tempting to use the “rehabilitation” or “nursing” subheadings to focus this search, it is possible to miss out on relevant literature by doing this. Try not to use these subheadings alone but in conjunction with other subheadings, such as “prevention and control” or “therapy.”

Restrict search to major topic headings only. The PubMed indexers apply up to 40 subject headings to each article, but to help searchers achieve more precision, indexers determine that some of the terms are major topic headings (approximately 5–10 terms). These major terms will be assigned an asterisk. Checking this box will cut retrieval considerably, but, again, this may result in missing useful articles. Users should avoid beginning a broad search by restricting to major topic headings; later, if users have too many results, searches can be restricted.

Do not explode this term. PubMed automatically “explodes” MeSH terms. This means that if the MeSH term has more specific terms beneath it in the MeSH tree, the search will be expanded to include those terms (Figure 1). However, if users do not want to search for more narrow terms, they can check the box beside “Do not explode this term.” For this search, the default settings include leaving the “Do not explode this term” box unchecked. All narrow terms will be included by default.

Entry terms. This is a list of terms that, when entered, will automatically map or translate to the preferred MeSH term. Entering “phlebothrombosis” in the MeSH Database or in the PubMed query box will translate to “venous thrombosis.”

The MeSH tree. The MeSH tree is a hierarchical arrangement of terms. It starts broadly with “All MeSH Categories” and then becomes more specific until it reaches the term being searched. Looking at the MeSH tree can identify other possible terms for a search. Clicking on any of the terms takes users to that section of the MeSH tree (see Figure 1).

Constructing a Search in the MeSH Database

  • When users find a likely term, they can click on the term itself to choose subheadings or to restrict their search to major topic headings only.
  • Build a search by clicking the Send to Search Box with AND/OR.
  • When users are satisfied with their query, click Search PubMed.

DB's case study search

  • Type “deep vein thrombosis.” Click on “venous thrombosis,” check the subheading “prevention and control,” and select Send to Search Box with AND (see Figure 2).
  • Type “pulmonary embolism.” Check the subheading “prevention and control,” and select Send to Search Box with OR.
  • Type “compression stockings.” Click on “stockings, compression.” Note the date this was added to the database (2007) and the previous indexing for this term (“bandages”). If, as in the initial search, the only term searched for was “compression stockings,” the results would be limited to very recent articles, probably only those published since late 2006. This means a search will also have to be done for “bandages” to ensure coverage of the earlier literature.
  • Add “stockings, compression” and “bandages” to the search but don’t choose any subheadings for these search terms.
  • Type “anticoagulants.” On the results page, check the boxes beside “anticoagulants and anticoagulants” [pharmacological action], and select Send to Search Box with AND.
  • Change any AND connectors to OR, if necessary, and add parentheses to structure the search.

See Figure 3 for an example of this search.

  • Click Search PubMed beneath the search box.
  • Apply Limits: Published in the last 5 years and English (see Figure 4).

As of July 31, 2007, this search strategy resulted in 87 articles.

Although using MeSH will result in more precise searches, there are times when MeSH won’t work. If users are looking for the name of a particular study (e.g., ALLHAT [Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial]) or a particular type of compression stocking (e.g., T.E.D.™ hose or knee-high), it may help to add these search terms to the search strategy as text words. After a MeSH Database search is completed, it’s easy to just add the new search terms to the end of the search strategy, either in the PubMed query box or by clicking on Details.

Think of search terms that relate to a patient’s condition, but be aware that some terms may be too specific. One strategy here would be to add the term “stroke” as text word or MeSH term to help focus the search further. Adding stroke to the search limits the results to seven articles, all of which seem to be right on topic. However, adding too many terms may lead to few or no articles. Ideally, users want to find an article that discusses their particular patient population; however, a high-quality general systematic review may be more useful from an evidence standpoint than a case study that specifically mentions a patient population. See Table 2 for additional search strategies.

When the articles have been found, they need to be critically reviewed. This review should consider whether the research results will help answer the current clinical question. Questions to consider are presented in Table 3 (Craig & Smyth, 2002; Greenberg, Daniels, Flanders, Eley, & Boring, 2005).

For example, questions about the effectiveness of an intervention are best answered by a randomized controlled trial or prospective or retrospective cohort studies that compare the outcomes of patients who did and did not receive the intervention. The study participants and the setting should be reviewed to determine whether they match the current question (e.g., rehabilitation setting, patients with stroke). The outcome under study should also be considered (e.g., deep vein thrombosis or pulmonary embolism).

Implications for Practice

Knowing how to search the literature is an important skill for nurses. Information retrieved can help support the clinical decisions related to patient care. As in the clinical scenario of DB, it can provide information regarding the treatment plan and provide nurses data that can be shared with the rehabilitation team to ensure the most beneficial care for the patient.

Training clinicians in the techniques needed to search the evidence will enhance the quality of the evidence that is retrieved (Haynes, Johnston, McKibbon, Walker, & Willan, 1993). Further education in critical appraisal of what is found will increase clinicians’ abilities to apply the information to the clinical setting (Taylor et al., 2000). The literature search may provide answers to users’ questions or may lead to additional questions. The literature could show that the most efficacious treatment is not the current standard of care. Determining whether the literature has a sufficient research base to make recommendations for a change in practice is essential (Kleinpell, Gawlinski, & Burns, 2006). If that is the case, the evidence would then lead to changes in policy and procedure for the institution.

About the Authors

Linda Ehrlich-Jones, PhD RN, is a clinical research scientist at the Rehabilitation Institute of Chicago, Chicago, IL. Address correspondence to her at lehrlich@ric.org.

Linda O’Dwyer, MSLIS MA, is an education librarian at Northwestern University, Evanston, IL.

Kathleen Stevens, PhD RN CRRN, is director of nursing eduction at the Rehabilitation Institute of Chicago, Chicago, IL.

Anne Deutsch, PhD RN CRRN, is a clinical research scientist at the Rehabilitation Institute of Chicago, Chicago, IL.


Center for Outcomes Research, University of Massachusetts Medical Center. (1998). Best practices: Preventing deep vein thrombosis and pulmonary embolism. Amherst, MA: Author.

Craig, J. V., & Smyth, R. L. (2002). The evidence-based practice manual for nurses. Edinburgh, NY: Churchill Livingstone.

Doig, G. S., & Simpson, F. (2003). Efficient literature searching: A core skill for the practice of evidence-based medicine. Intensive Care Medicine, 29(12), 2119–2127.

Geerts, W. H., Pineo, G. R., Heit, J. A., Bergqvist, D., Lassen, M. R., Colwell, C. W., et al. (2004) Prevention of venous thromboembolism: The seventh ACCP Conference on antithrombotic and thrombolytic therapy. Chest, 126, 338–400.

Greenberg, R. S., Daniels, S. R., Flanders, W. D., Eley, J. W., & Boring, J. R. (2005). Medical epidemiology (4th ed.). New York: McGraw-Hill.

Haynes, R. B., Johnston, M. E., McKibbon, D. A., Walker, C. J., & Willan, A. R. (1993). A program to enhance clinical use of MEDLINE: A randomized controlled trial. Online Journal of Current Clinical Trials, May 11, 1993, Doc No 56.

Joanna Briggs Institute. (2001). Graduated compression stockings for the prevention of post-operative venous thromboembolism. Retrieved April 3, 2008, from www.joannabriggs.edu.au/pdf/BPISstock.pdf.

Kleinpell, R. M., Gawlinski, A., & Burns, S. M. (2006). Searching and critiquing literature essential for acute care NPs. Nurse Practitioner, 31(8), 12–13.

Richardson, W. S., Wilson, M. C., Nishikawa, J., & Hayward, R. S. (1995). The well-built clinical question: A key to evidence-based decisions. ACP Journal Club, 123(3), A12–A13.

Sackett, D. L., Straus, E. E., Richardson, W. S., Rosenberg, W., & Haynes, R. B. (2000). Evidence-based medicine: How to practice and teach evidence-based medicine (2nd ed.). St. Louis: Churchill Livingstone.

Stettler, C. B., Brunell, M., Giuliano, K. K., Morsi, D., Prince, L., & Newell-Stokes, V. (1998). Evidence-based practice and the role of nursing leadership. Journal of Nursing Administration, 28, 45–53.

Stone, P. W. (2002). Popping the (PICO) question in research and evidence-based practice. Applied Nursing Research, 16, 197–198.

Taylor, R., Reeves, B., Ewing, P., Binns, S., Keast, J., & Mears, R. (2000). A systematic review of the effectiveness of critical appraisal skills training for clinicians. Medical Education, 34(2), 120–125.