Introduction

To find answers to clinical inquiries, it is necessary to conduct an evidence-based literature search. This chapter describes the most commonly used evidence-based resources, recommends search strategies that will help enable the oral health practitioner to retrieve the most useful clinical information, and suggests a workflow for navigating evidence-based information.

The practice of evidence-based dentistry includes articulating a clinical question, conducting a search for relevant literature, evaluating the retrieved evidence, applying that evidence to answer the clinical question at hand, and putting that answer into effect in a clinical setting.

The amount of health-related information available via the internet is staggering. Even just the published results of clinical trials and systematic reviews are impossible to keep up with. It is estimated, for example, that an average of 11 systematic reviews and 75 results of clinical trials are published in the medical literature daily.¹ In addition, health information found within the peer-reviewed literature can be contradictory. The exponential growth of complex health information is why it is essential to learn not only how to search for the best evidence-based literature, but also how to evaluate that literature.

Effective searching of medical databases is a very valuable skill these days. Below, we outline effective and efficient ways for busy clinicians to search the most important medical resources. For more complex searches and the development of expert searching skills, such as conducting systematic reviews, we recommend working with a librarian who has been trained in extensive searching of the scientific literature.

General Tips for Searching Electronic Resources

A general search tip is to keep things simple when first starting a search. Begin with a broad search and use only a few terms without applying any search filters; take a look at the results to determine if the search is too broad or too narrow and whether it’s displaying relevant results. Try using nouns as search terms rather than adjectives, as adjectives are difficult to quantify and can muddy the results. Beginning a search with too many terms or multiple filters applied may result in the inadvertent exclusion of relevant literature (Table 2.1).

Table 2.1. General Searching Tips for Electronic Resources

If the search is not yielding the expected results, consider using synonyms for search terms, including variations of spelling, such as U.S. versus U.K. spelling of terms. For instance, a synonym for “pregnant women” is “pregnancy.” Searching different terms for the same concept can yield different results. Databases will often have a controlled vocabulary, which is preferred terminology that is found in the database’s thesaurus and can be very helpful when searching for relevant literature, as database thesauri typically include all synonyms for a search term. A team of indexers is responsible for reading all included materials in any given database and identifying the main concepts represented in each individual work; once the main topics are identified, the indexer assigns subject headings to the work using the preferred terminology. Using the preferred terminology therefore standardizes the search process, making it easier and quicker to locate relevant literature. The searcher should consider looking up search terms in the database’s thesaurus and using the suggested term, which will yield results in which the term is one of the main subjects of the retrieved literature. If a database does not use a controlled vocabulary, consider searching by a specific field such as title, abstract, or author to further limit results. Keyword searches will look for the terms to appear anywhere in the record, while searching in a specific field limits where the term will appear. The majority of the resources described below allow searching by phrase by placing the phrase within quotation marks. Most also recognize common misspellings and abbreviations, and will suggest alternative search terms when no results are found. In addition, most of the resources below search plurals automatically.

Boolean searching refers to connecting search terms with one of the following three terms: AND, OR, or NOT. When combining two terms with AND, search results will contain both terms, resulting in fewer results than searching using only one of the terms. Combining terms with OR will return results that contain at least one of the terms, resulting in a higher number of search results than if only one of the terms was searched. Combining two terms with NOT will yield results in which the first term appears but not the second term, narrowing the number of results (Figure 2.1).

Figure 2.1. Examples of Boolean Searching

Searching Workflow

We suggest searching several databases; searching a combination of databases helps to ensure that information is not missed that might influence the answer to the clinical question at hand. While the exact databases searched will depend on the question/topic/need, the importance of finding results that contain high levels of evidence calls for, at a minimum, Cochrane, Trip, and PubMed to be searched. Information about many additional available databases are listed in the Resources section of this chapter. To avoid missing important results, use all three resources: there is less overlap among them than one would imagine, they cover different types of materials, and the content in each is updated on different cycles. Each resource has a unique architecture, vocabulary, and results algorithm and, therefore, will return unique results. Searching Cochrane, Trip, and PubMed is free, but the full text of results is not always available without a subscription to source material, which is typically provided when operating in an academic setting. For those in clinical practice outside academic settings, the American Dental Association (ADA) Library and Archives offers members access to full-text articles identified through these databases. All three resources are relatively easy to search and provide up-to-date information. By using these three databases, the oral health practitioner will have searched the gold standard for evidence-based information (Cochrane), the best “preappraised” resource (Trip), and the most comprehensive medical literature search engine (PubMed).

We recommend framing the search strategy using the PICO framework to articulate the clinical question. PICO assists clinicians in creating a clear, focused clinical question by breaking down the question into four components: population (Who is being treated? What specific characteristics does this population have?), intervention (What treatment or therapy is being considered?), comparison (What is the gold standard? The comparison can also be no comparison, or a placebo), and outcomes (What improvements or changes are hoped for?). Of course, PICO will not fit every clinical question as not all questions are ones of therapy; the framework can be modified depending on the type of question. For instance, if it is a question of harm, the clinician will frame the question by considering what exposure, rather than what intervention, to search for. We recommend that clinicians refer to the framework as a starting point and modify as necessary for individual research needs. Using the PICO framework is likely to result in the return of relevant results because the framework encourages clarity and specificity and it breaks down the clinical question into easily searchable components. Using the PICO framework to articulate the question will save the clinician time when conducting the search and better enable the clinician to review the results of the search for relevance and applicability to clinical practice (or to a particular patient) (Table 2.2).

Use keywords, indexing terms, and synonyms related to the components of the PICO question for the search. We suggest beginning with two PICO components as search terms rather than all four. Begin with search terms based on the population and intervention components of the PICO question, as the results will most likely include the comparison term or concept. Adding the outcome terms will further narrow the results, and these terms should be added if the initial search yields far too many results. Using all four components as search terms initially will narrow a search too early and will miss results that may be important. Use the Boolean operator OR to account for synonyms of the same concept. Conversely, use the Boolean operator AND to require different components to be present in the results. See Figure 2.1 for examples of Boolean searching.

Table 2.2. Main Components and Examples of Clinical Questions, According to Their Natures

Source: Brignardello-Petersen R, Carrasco-Labra A, Booth HA, Glick M, Guyatt GH, Azarpazhooh A, Agoritsas T. “A practical approach to evidence-based dentistry: II: how to search for evidence to inform clinical decisions.” J Am Dent Assoc. 2014 Dec;145(12):1262-7. doi: 10.14219/jada.2014.113. PubMed PMID: 25429040.

Most databases have a thesaurus that lists the preferred terminology or subject headings used to index articles as they are added to the database. Searching by subject headings is a simple way to retrieve a smaller pool of relevant results, as subject headings standardize the search process and return results where the search term is considered a main topic (that is, subject) of the manuscript. When not using subject headings to search, limit the search to the title and abstract. If the initial search retrieves too many results, limit the results to secondary and preappraised literature such as clinical practice guidelines, summaries, critical summaries, and systematic reviews. If secondary literature does not exist, it will be necessary to consult primary studies; we suggest initially limiting the search to randomized controlled trials as they are typically less prone to bias (Figure 2.2).

Figure 2.2. Hierarchy of Evidence for Primary Studies, Level of Processing, and Link to EBD Resources

Initially, set aside results that are more than a decade old. If the search still retrieves too many items, do a search using all four PICO components, starting with adding the outcome terms to the search. If the results of the search include materials from the highest levels of evidence (for example, guidelines or systematic reviews), consider stopping the search and declaring it “good enough.” To conduct a more comprehensive search, consult the additional databases listed below. Note that this workflow is intended to be used as a roadmap for an evidence-based search for a clinical question in order to decide action in a clinical setting—it is not a substitute for a search that would be used to conduct a systematic review or literature review.

There is always an element of creativity in developing a search strategy around a particular question and in determining the appropriate balance among relevance, breadth, and depth of results. Developing a search strategy is an iterative process; that is, as the oral health practitioner searches, he or she will modify, narrow, or expand the search terms used as well as the combinations of terms and the overall search strategy based on results. The effectiveness and efficiency of the search strategy will improve as the search progresses. If too few results are retrieved in a particular database, it may be necessary to modify the search by removing some search terms, search filters, or both. Please note that simply adding more terms to the search may not increase the number of relevant results as the terms may retrieve redundant results. Once a search is begun, there is a certain amount of trial and error in terms of refining the search terms to be used in the end. Adding and removing search terms and synonyms will allow the oral health practitioner to retrieve varying amounts, breadth, and quality of levels of evidence.

We suggest beginning the search process with Cochrane, then searching Trip if not enough relevant results are found, and then PubMed to fill in any remaining gaps in information. Since Cochrane is a database consisting solely of systematic reviews, the results presented are considered to be of a higher quality according to the evidence pyramid hierarchy. Due to the type of content, it is also a smaller database, so it is possible that the evidence needed might not be available in Cochrane. PubMed will almost always retrieve the most results from the search because of the breadth and scope of the resource. As noted below, even when using the Systematic Review filter or Clinical Queries search, which limits search results to clinical study categories, systematic reviews, and medical genetics, PubMed retrieves narrative reviews as well as descriptive articles. (For more information about Clinical Queries, please see the PubMed section of this chapter. Note that while Cochrane systematic reviews are indexed in PubMed, we recommend a separate search of the Cochrane Library because the results of a search of the Cochrane Library database and a search of Cochrane systematic reviews in PubMed will not retrieve identical results.)

We suggest conducting a search using both (that is, a combination of) keywords and subject headings. Note that Cochrane and PubMed share index terms—both databases use MeSH (Medical Subject Headings) as a controlled vocabulary or thesaurus for subject headings (see the PubMed section of this chapter for a detailed explanation of MeSH). Trip does not use a controlled vocabulary. Do not rely solely on index terms: many new articles and updates appear in databases before they are indexed, and these materials may prove to be the most useful. As the initial search is conducted, note related terms and at the end, go back and do a search using all relevant search terms. Including related terms or synonyms is crucial to conducting an effective search (Table 2.3).

Table 2.3. Workflow for Searching for Evidence

** Indicates that this type of evidence may not be available for this type of question.

For this chapter, our search examples will refer to the following PICO framework question: Are pregnant women with periodontal disease at greater risk of preterm birth or low birth weight than pregnant women without periodontal disease? In this instance, the population identified is pregnant women. Because this is not a question of therapy but of harm, the exposure is periodontal disease, with the comparison as the absence of periodontal disease. Outcomes of interest are preterm birth and low birth weight.

Resources

Comprehensive Resources

Epistemonikos

Website: www.epistemonikos.org

Cost: Free, with some limited full-text availability

Specialty of resource: Clinical practice guidelines, synopses, systematic reviews, primary studies, other

Epistemonikos is a publicly available resource whose objective is to “gather scientific information (i.e., evidence) that might be relevant for health decision-making and to provide rapid access to the best available evidence for real-life questions.” It was created primarily for health professionals, researchers, and individuals who make decisions in health care.² Epistemonikos, which translated from Greek means “what is worth knowing,” includes evidence-based policy briefs, systematic reviews, primary studies, and synopses of systematic reviews as well as primary studies; although all of these different levels of evidence are available in this resource, there is a heavy emphasis on primarily identifying systematic reviews. Epistemonikos acts as an aggregator and pulls information from 10 different databases: PubMed, the Cochrane Database of Systematic Reviews, Embase, CINAHL, PsycINFO, LILACS, the Database of Abstracts of Reviews of Effects (DARE), the Campbell Collaboration online library, the JBI Database of Systematic Reviews and Implementation Reports, and the EPPI-Centre Evidence Library. Individuals are able to search all of these resources simultaneously with one search, saving time, which may make this resource one of the best places to begin a search.

Individuals can perform a basic search from the main page or select the “advanced search” option. We recommend going straight to the advanced search option as the basic search is meant to search only one term at a time and does not support Boolean searching. For instance, entering our search terms “periodontal disease” and “pregnant women” into the basic search function returns fewer than 20 results, but entering those same terms into the advanced search returns more than 60 results. The advanced search builder allows for individuals to enter in PICO elements and combine them as needed. Filters are available on the main page of search results to further narrow down the original search results if needed.

Results are identified by level of evidence and include a basic citation, as well as an abstract, for each result. As previously mentioned, Epistemonikos is an aggregate database, so the full text of an identified article is available at the original resource (linked in the database). Availability of the full text depends on the individual’s institutional subscriptions; ADA members may also check with the librarians at the ADA Library. A helpful feature of this database is the “related evidence” tab, which is available on the main search results page. When a primary study is selected, the related evidence tab includes citations of systematic reviews that include the primary study and links to these systematic reviews. When a systematic review is selected, the related evidence tab includes citations of the included primary studies of the review, as well as a statement that identifies the total number of systematic reviews that include the individual primary study. This feature provides a quick and easy way to find more evidence on a topic.

Epistemonikos allows for the option to create a free personal account. Oral health care practitioners can save searches as well as individual article citations, marking them as favorites. Creating a personal account also allows individuals to create and save matrices of evidence. Matrixes of evidence pull the included articles from a selected systematic review and then search for other systematic reviews that include the original review’s primary studies; the end result is a grid that demonstrates what systematic reviews on the topic include which primary studies and where they overlap. The option to create a matrix is available within the record of a systematic review. The created matrix can then be saved to the personal account (Figure 2.3).

Figure 2.3. Example of an Epistemonikos Matrix of Evidence

Source: Epistemonikos.org, accessed on March 18, 2019.

Trip

Website: www.tripdatabase.com

Cost: The Trip database, previously only available via a subscription, is now freely accessible. A premium version of Trip is available to individuals via a subscription.

Specialty of resource: Clinical practice guidelines, synopses, systematic reviews, primary studies, other

Trip (which originally stood for Translating Research into Practice) is a database/meta-search engine that covers a wide range of sources, including MEDLINE, the Cochrane Library, guidelines, and more. Trip is a useful resource to consult initially because it covers a wide range of material and its search technology allows individuals to find high-quality evidence quickly and easily. Trip results provide information from multiple sources and are color-coded by level of evidence as well as sorted by category. Results are displayed according to an algorithm that provides relevant, recent, and higher quality information first. “Higher quality” information in Trip means types of information that are more highly ranked in the hierarchy of evidence-based information represented in the evidence pyramid (Figure 2.4). Results can be filtered using the “refine by” function on the right-hand side of the screen to filter by level of evidence. Results may be refined by systematic reviews, evidence-based synopses, guidelines, primary research from a core set of more than 300 journals, controlled trials, electronic textbooks, and more. Each of these categories contains information from multiple resources. “Guidelines,” for example, includes guidelines issued within North America, Europe, and other countries, while “evidence-based synopses” includes material from Bandolier, BestBets, POEMs (Patient-Oriented Evidence that Matters), Clinical Evidence, the journal Evidence-Based Complementary and Alternative Medicine, and more. The type of information being provided by any given result is indicated by an accompanying image locating the type of information on the evidence pyramid.

To retrieve the highest-level evidence from Trip, begin the search with just one or two terms and then move to more complex searches. Use the synonyms tab at the top of the search results if no results are retrieved. Special features of Trip include an ability to filter for evidence relevant to the developing world. A search using “pregnancy and periodontal disease,” for instance, results in “Trip’s best suggested answer”: “Although periodontal disease treatment is recommended during pregnancy, the results of a systematic review of 13 randomized controlled clinical trials, for which a meta-analysis was conducted in 11 of the trials, do not indicate that there is a lower risk of preterm birth or low birth weight” and links to the 2014 critical summary “Periodontal Disease Treatment Does Not Affect Pregnancy Outcomes” in the Journal of the American Dental Association³ as well as systematic reviews and more.

Trip also provides additional useful ways to search, including a “PICO search,” which allows the oral health practitioner to search by PICO categories. All four categories of the PICO search do not need to be entered in order to use the tool. The more categories that are supplied, however, the more focused the search results will be. Trip also provides a useful image and video search and includes information leaflets for patients.

The full text of documents, when available, can be accessed by clicking on the title of the document. Citations from Trip may be exported to a citation manager. Trip allows individuals to share results via Facebook or Twitter. Trip’s awareness tool, My Trip, allows individuals to select and register keywords that are automatically searched whenever the database is updated. If material is retrieved, results may be sent to the oral health practitioner’s e-mail address (Figure 2.4).

Tutorials:

Figure 2.4. Trip Database

Source: Tripdatabase.com, accessed on March 18, 2019.

Summaries and Guidelines: Point-of-Care Resources

Point-of-care resources aim to deliver the best available evidence to support decisions made by health care providers in clinical settings. These tools are also excellent resources for gathering information on both foreground and background questions. Although we highlight several resources below, there is a wide variety of point-of-care tools available.

UpToDate

Website: www.uptodate.com

Cost: Only available via a subscription. Access to the patient content of UpToDate is free.

Specialty of resource: Summaries

UpToDate is a peer-reviewed database of point-of-care information that provides a summary of research on a range of medical topics. The database is aimed at clinicians: It is very useful not only for staying current with the latest research, but also for providing information to patients. It includes access to more than 10,000 topics that are divided into more than 20 specialties (for example, drug information and pediatrics). The information included in UpToDate is provided by more than 4,000 expert clinicians who review content from more than 460 medical journals. It includes information focused on specific clinical issues (diagnosis and treatment), descriptions of diseases and conditions, and analysis of medical treatments and recommendations. It is important to note that UptoDate is not a general literature database like PubMed. (One cannot search by author name or the title of an article or journal.) Information in UptoDate that is judged to be of pressing importance is updated promptly after the peer review process (these are marked as “practice changing updates”); all topics are updated every four months. Each recommendation is graded for strength using a modified GRADE (Grading of Recommendations, Assessment, Development and Evaluations) classification scheme. Grades 1 and 2 are based on the strength of the recommendation, and grades A through C are based on the certainty in the evidence (see Chapter 14).

UpToDate includes the drug information database Lexicomp, so it is particularly helpful in looking for information concerning drug dosing, interactions, and adverse effects. It is possible to conduct a search for a drug name just like any other search. UpToDate also includes medical calculators, accessible via browsing or by medical specialty. UpToDate contains a great deal of patient-oriented information, written in plain language, divided by topic and then further subdivided into two categories: “The Basics” and “Beyond the Basics.” “Basics” information is generally one to three pages long, while “Beyond the Basics” information is generally five to 10 pages long.

Searching UpToDate is quite easy. Searches are automatically conducted across all specialties. Searches can be narrowed using the drop-down menu to the right of the search box by selecting one of the following facets: all topics, adult, pediatric, patient, or graphics (that is, pictures, algorithms, graphs, tables, movies, and figures). All results are displayed, but the results more closely associated with the chosen facet are displayed first.

Hovering over a topic title provides a preview of the topic outline for that topic, allowing the individual to jump to specific parts of the topic when necessary.

A “What’s New” section, organized by specialty, lists important updates in each field from the past three to six months, and a separate section, “Practice Changing Updates,” notes updates that may significantly change clinical practice.

Tutorials:

Stay updated, free dental videos. Join our Telegram channel

Join