LITERATURE SEARCHING: A PRIMER

LITERATURE SEARCHING: A PRIMER Contents Defining your research topic... 2 Choose databases... 3 Supplementary searching... 3 Boolean operators... 4 Gathering search terms... 5 Adding database-specific syntax... 6 Putting it all together... 7 Download results... 8 Reporting the search... 8 APPENDIX 1: Sample citation page in Ovid MEDLINE database last updated 20-nov-17 page 1 of 9

Defining your research topic Before beginning your search, it is important to know what you are searching for. One common way to break down a research question is the PICO format: Example 1 Example 2 Your Topic (P)opulation (I)ntervention (C)omparison Post-menopausal women Hormone replacement therapy No hormone replacement therapy Adults with disabilities Interventions to promote physical activity N/A (O)utcome risk of breast cancer Any What type of review are you conducting? Literature Review Scoping Review Systematic Review Reporting guidelines N/A Arskey & O'Malley framework 1 PRISMA checklist 2 Research question Searching topic-based various levels of exhaustivity "What is known about?" (focuses on breadth of studies in a particular area) systematic searches of multiple databases and grey literature e/supplementary searching "What is the effectiveness of?" (focuses on answering a specific research question) systematic searches of multiple databases and grey literature/supplementary searching Critical Appraisal N/A optional required Synthesis narrative narrative/tabular narrative/tabular, sometimes accompanied by a meta-analysis Write out your research question: 1 Arksey H, O'Malley L. Scoping studies: towards a methodological framework. International journal of social research methodology. 2005 Feb 1;8(1):19-32. 2 PRISMA checklist: http://www.prisma-statement.org/prismastatement/checklist.aspx last updated 20-nov-17 page 2 of 9

Choose databases There are many subject databases, and which ones you search will depend on your search question. Every database has a slightly different scope and includes different journals. Some common databases include: MEDLINE Embase Cochrane Library PsycINFO CINAHL AMED Run by the National Library of Medicine, MEDLINE is one of the biggest and most well organised databases in medicine. MEDLINE can be accessed through the PubMed or Ovid interfaces. Another large biomedical database with a more European focus includes a database of Cochrane systematic reviews, a register of controlled trials, as well as several other small databases Psychology database Nursing and allied health database Allied health and complementary medicine database There are many more databases that are discipline-specific. Discuss with your information specialist if there are other databases that are relevant to your research question, or browse a list of databases by subject here: http://plymouth.libguides.com/sb.php To access a list of all subscription databases at Plymouth University, navigate to the library guide titled "A-Z databases" (http://plymouth.libguides.com/c.php?g=48969). Which database(s) are relevant for your project? MEDLINE (Ovid or PubMed interface) Supplementary searching Formal reviews aim to be comprehensive that is, to find every paper that is relevant to the research question. Database searches may miss relevant articles to your topic, so it is important to conduct supplementary searches. Supplementary searches are particularly useful when non-published material (such as white papers or government reports) may add value to your review. What supplementary searching will you conduct for your project? Search the reference lists of included citations (backwards citation chaining) Search for any citations which have cited your included citations (forwards citation chaining) Targeted searches of Google and/or Google Scholar Search for relevant conference proceedings Handsearch of relevant think tank, association, or other organisational websites. Search of clinical trials registries Ask experts in the subject area for recommendations of relevant research articles Search for relevant theses or dissertations last updated 20-nov-17 page 3 of 9

Boolean operators There are two main boolean operators. These special words allow you to construct searches that will return more precise results. AND OR OR anxiety depression Using OR always returns more results than either term alone! anxiety OR depression Searching using OR returns any results that contain any of the search terms. In the diagram above, a citation containing the word anxiety but not the word depression would still be returned in your results. Search terms Results anxiety 179,934 depression 320,518 OR is used for closely aligned or synonymous terms. anxiety OR depression 433,517 (more results) AND Using AND always returns less results than either term alone! depression AND poverty Searching using AND only returns results where both terms are in the citation. In the diagram above, a citation containing the word depression but not the word poverty would not be returned in your results. Search terms Results depression 320,518 poverty 156,325 AND is used to combine different concepts in your strategy. depression AND poverty 8,521 (less results) last updated 20-nov-17 page 4 of 9

Gathering search terms Now it's time to put together all the pieces. For each of your major concepts, try to think about the different terms or synonyms that might be used in the literature. Consider the following research question: "What is known about interventions promoting physical activity in adults with disability?" Here are some examples of synonyms for each concept: Adults with disability (concept 1) Physical activity interventions (concept 2) (concept 3) Disability / disabled Visually impaired / impairment Hearing impaired / impairment Down's syndrome (etc ) Physical activity Exercise Fitness Sports Yoga (etc ) (For some questions, it may not be necessary to search for every concept! For this question, searching for concepts 1 and 2 will probably return a feasible number of results to screen, so we don t need to add anything else) What are some synonyms for each concept in your research question? (concept 1) (concept 2) (concept 3) AND AND last updated 20-nov-17 page 5 of 9

Adding database-specific syntax Each database has its own syntax that is used to tell the database where to search for the term. Searching the title and abstract of a citation EBSCO databases ProQuest databases Ovid databases (e.g. CINAHL, AMED) (e.g. PyscINFO) (e.g. Ovid MEDLINE, Embase) Title and Abstract TI(search term) OR AB(search term) TI(search term) OR AB(search term) search term.ti,ab Truncation Most databases allow truncation, generally using an asterisk (*). Truncation allows you to search for several different word endings, beginnings, or alternative spellings, at the same time, without writing them all out. visual* impair* disab* wom*n behavio*r disability visually impaired woman behavior disabled visual impairment women behaviour disablement Caution! Not all databases use the same symbols, and some databases do not support the use of truncation at the beginning of a word or in the middle of a word. Be sure to test the terms to make sure they return the results you expect, or read the database documentation. Adding Subject Headings So far, we have structured a search using free text terms that can be found by searching an article's title or abstract. It s not possible to come up with every single phrase and word variation for every concept. Well-structured searches use subject headings to retrieve more relevant results. Subject headings are chosen from a list, rather than generated by the searcher. Experts compile these lists of terms and apply them to citations based on the subject of the article. Note: every single database has its own subject heading list! To find relevant subject headings: 1. Search the list of subject headings directly. Most databases will have a tab or button that says "subject headings" (or a similar phrase), which will direct you to a searchable list. 2. Look for articles in the database that are relevant to your topic, and check to see what terms have been applied to the article some of them may be relevant to your search as well. Subject headings are hierarchical Choose the narrowest subject heading that still makes sense for your topic, and be sure to search for any narrower terms that are also relevant. last updated 20-nov-17 page 6 of 9

Subject heading syntax EBSCO databases ProQuest databases Ovid databases (e.g. CINAHL, AMED) (e.g. PyscINFO) (e.g. Ovid MEDLINE, Embase) Subject headings (these terms are chosen from a list created by the database) In CINAHL: MH(subject heading) In most others: SU(subject heading) SU(subject heading) subject heading/ Putting it all together Once you have gathered your search terms and subject headings, it's time to start structuring your search. It's best practice to run your search with one item on each line, like the one below. This formal search strategy uses Ovid MEDLINE syntax # Searches Results Explanation 1 osteoarthritis, knee/ 17446 Subject headings are chosen from a list created by the database. In Ovid MEDLINE, subject headings are denoted by the "slash" symbol. Search concept 1 2 (knee adj2 osteoarthritis).ti,ab. 8772 Adj# will search for terms within x number of one another in Ovid MEDLINE. This line will capture the phrases "knee osteoarthritis" and "osteoarthritis of the knee" in the title and abstract fields. Most databases have this function, but the syntax will vary. Search concept 2 3 1 or 2 19933 4 exp exercise therapy/ 44851 5 exercis*.ti,ab. 236379 6 physical activity.ti,ab. 88623 7 4 or 5 or 6 326439 8 3 and 7 1810 Use OR to include any article that is picked up by searches on lines 1 or 2. Exp means that the database should search for any narrower terms in the subject heading hierarchy. In some databases, this shorthand doesn't exist, and you will need to write out all narrower terms manually. In Ovid MEDLINE, use the asterisk symbol to truncate words or phrases. This line will capture terms such as "exercise", "exercises", and "exercising". Use OR to include any article that is picked up by the searches on lines 4, 5, or 6. Use AND to combine your search concepts together. Only articles that contain a search term related to concept 1 AND a search term related to concept 2 will be returned. last updated 20-nov-17 page 7 of 9

Before downloading your search results, it's best to make sure that the results are relevant. Take a look at the results in the database from your search, and make changes to your strategy. This process can take several weeks! When you looked through the results, did you see any new search terms that could be added to your strategy? Did some of the search terms produce unexpected or irrelevant results? Consider changing or removing them. Download results Every major database allows users to download the results of their search. It is very important to finalise your search strategy before downloading your results. After this point, you will not be able to make any changes to your search. Downloading full search results EBSCO databases ProQuest databases Ovid databases (e.g. CINAHL, AMED) (e.g. PyscINFO) (e.g. Ovid MEDLINE, Embase) How to download Select the "share" button, Use checkbox to select all items Select the checkbox "all", then citations then "export results" on the page, then "save" click "export" Reporting the search Record-keeping At the same time that you download your search results, it is important to carefully record the following information for each database searched and keep it in a safe place: The date that you conducted the search The person conducting the search The full name of the database (e.g. PsycINFO) and interface (e.g. ProQuest) for each search o In Ovid MEDLINE, be sure to record the exact database being used (e.g. Ovid MEDLINE(R) Epub Ahead of Print, In-Process & Other Non-Indexed Citations 1946 to present) The full search strategy used, exactly as it was conducted in the database, with the number of citations retrieved by each line. It is useful to copy-paste directly from the database itself for your records. Any additional limiters used (e.g. a date limit) The number of citations downloaded for screening (which should correspond to the number of citations on the last line of the search strategy) De-duplicating your results After all citations have been downloaded from all database searches, it is important to de-duplicate the citations before screening otherwise, you may waste time and energy screening some citations multiple times! Most citation management systems allow for automatic de-duplication of citations. Be sure to record the number of citations downloaded from all sources before and after de-duplication. Both of these numbers are required for reporting your final results. last updated 20-nov-17 page 8 of 9

Screening your results It is best practice to screen all citations by two people independently (i.e. without knowing one another's screening decisions), with a third person available to resolve any screening disputes. There are two main ways to screen your citations: EndNote The following tutorial gives a good overview about how this process works: http://medicine.exeter.ac.uk/esmi/workstreams/informationscience/is_resources,_guidance_&_advice/ Rayyan Freely available web-based software specially made for screening papers for review projects rayyan.qcri.org PRISMA flow diagram There is an expectation that any scoping review or systematic review will present the number of citations downloaded and screened using a PRISMA flow diagram. Be sure that you are collecting the numbers required for reporting throughout the research project otherwise, it can be difficult to recreate them later on. The PRISMA template can be found at http://www.prisma-statement.org. PRISMA flow diagram last updated 20-nov-17 page 9 of 9

Last updated 16-jan-17 amanda.wanner@plymouth.ac.uk Sample Ovid Medline Search Database(s): Ovid MEDLINE(R) In-Process & Other Non-Indexed Citations and Ovid MEDLINE(R) 1946 to Present Search Strategy: # Searches Results Explanation 1 osteoarthritis, knee/ 17541 Subject headings are chosen from a list and denoted by a slash ( / ) 2 (knee adj2 osteoarthritis).ti,ab. 8764 3 1 or 2 19827 4 exp exercise/ 168844 5 exp exercise therapy/ 44812 6 exp exercise movement techniques/ 7247 7 4 or 5 or 6 204441 8 exp pain/ 376423 9 pain.ti,ab. 536841 10 8 or 9 694984 Adj# will search for terms within x number of one another. For example, this line will search for "knee osteoarthritis" and "osteoarthritis of the knee" Use OR to include any article that is captured by either of the above lines. OR = more. Exp means that the database should search for any narrower terms in the subject heading hierarchy. 11 3 and 7 and 10 954 Use AND to combine all your major concepts 12 limit 11 to (english language and "all aged (65 and over)" and "review") 20 Use limiters to further narrow your results Tips: PubMed is a freely available database (though the articles themselves are not free!). However, to access Ovid Medline, you will need to log in with your university credentials. While PubMed automatically includes narrower terms in the subject heading hierarchy, in Ovid you will need to use "exp" (short for "explode")

My Account Support & Training Help Logoff Search Journals Multimedia My Workspace EBP Tools Mobile < Back to Search Results. 1 of 1 Results Keep Selected Tools Unique Identifier: 25691583 Record Owner: From MEDLINE, a database of the U.S. National Library of Medicine. Abstract Reference Find Similar Status: MEDLINE Find Citing Articles Authors: Tanaka R; Ozawa J; Kito N; Moriyama H. Authors Full Name: Tanaka, Ryo; Ozawa, Junya; Kito, Nobuhiro; Moriyama, Hideki. Institution: Tanaka, Ryo. Department of Rehabilitation, Faculty of Rehabilitation, Hiroshima International University, Japan r-tanaka@hs.hirokoku-u.ac.jp. Ozawa, Junya. Department of Rehabilitation, Faculty of Rehabilitation, Hiroshima International University, Japan. Kito, Nobuhiro. Department of Rehabilitation, Faculty of Rehabilitation, Hiroshima International University, Japan. Moriyama, Hideki. Department of Rehabilitation Science, Graduate School of Health Sciences, Kobe University, Japan. External Resolver My Projects Annotate Title: Source: Abbreviated Source: NLM Journal Name: Publishing Model: Effects of exercise therapy on walking ability in individuals with knee osteoarthritis: a systematic review and meta-analysis of randomised controlled trials. [Review] Clinical Rehabilitation. 30(1):36-52, 2016 Jan. Clin Rehabil. 30(1):36-52, 2016 Jan. Clinical rehabilitation Journal available in: Print-Electronic Citation processed from: Internet NLM Journal Code: cr7, 8802181 ISO Journal Abbreviation: Journal Subset: Country of Publication: MeSH Subject Headings: Keyword Heading: Clin Rehabil Index Medicus England *Acceleration Aged Disability Evaluation *Exercise Therapy / mt [Methods] Female Gait / ph [Physiology] Humans Male Middle Aged *Osteoarthritis, Knee / di [Diagnosis] *Osteoarthritis, Knee / rh [Rehabilitation] Prognosis Randomized Controlled Trials as Topic Severity of Illness Index Treatment Outcome *Walking / ph [Physiology] Exercise knee osteoarthritis

meta-analysis systematic review walking Abstract: OBJECTIVE: To examine the effect of exercise therapy on the walking ability of individuals with knee osteoarthritis. DATA SOURCES: Randomised clinical trials (RCTs) were identified by searching through PubMed, Cochrane Central Register of Controlled Trials, Physiotherapy Evidence Database, and Cumulative Index to Nursing and Allied Health Literature. All literature published to October 2014 were included in the search. REVIEW METHODS: Data were collected from RCTs that compared the effects of exercise therapy on walking ability with the effects of no intervention or psychoeducational intervention in participants with knee osteoarthritis. The outcome data on the total distance walked (6-minute walk test); the amount of time spent walking (the time to walk arbitrary distances); and gait velocity were obtained and analysed. Standardized mean differences (SMDs) and 95% confidence intervals (CIs) were calculated. RESULTS: Twenty-eight RCTs were identified. Meta-analysis provided very-low-quality evidence that exercise therapy increased the total distance walked in the 6-minute walk test, in comparison with the effects of the control interventions (SMD = 0.44, 95% CI 0.27 to 0.60). Meta-analysis also provided low- or moderate-quality evidence that the amount of time spent walking and gait velocity were improved more by exercise therapy than by the control interventions (the amount of time spent walking: SMD = -0.50, 95% CI -0.70 to -0.30; gait velocity: SMD = 1.78, 95% CI 0.98 to 2.58). CONCLUSION: In individuals with knee osteoarthritis, exercise therapy can improve the amount of time spent walking, gait velocity, and maybe the total distance walked. Copyright The Author(s) 2015. ISSN Electronic: 1477-0873 ISSN Linking: 0269-2155 Publisher Item Identifier: 0269215515570098 Digital Object Identifier: Publication Type: Article Identifier: Publication Status: Publication History Status: Language: https://dx-doi-org.plymouth.idm.oclc.org/10.1177/02692... Journal Article. Meta-Analysis. Research Support, Non-U.S. Gov't. Review. 25691583 [pubmed] 0269215515570098 [pii] 10.1177/0269215515570098 [doi] ppublish 2014/03/26 [received] 2014/12/28 [accepted] English Electronic Date of Publication: 20150217 Date of Publication: 2016 Jan Date Created: 20151215 Entrez Date: 2015/02/19 06:00 MeSH Date: 2016/09/20 06:00 Create Date: 2015/02/19 06:00 Year of Publication: 2016 Entry Date: 20160919 Revision Date: 20151215 Update Date: 20161215