EPIB 629 Knowledge Synthesis

In-class PowerPoint presentations

Once you have a reasonably well defined research question, it is important to make sure your project has not already been recently and successfully undertaken. This means it’s important to find out if there are other research syntheses (including systematic reviews) that have been published or that are in the process of being published on your topic.

If you are submitting your systematic review for funding, for example, you may want to make a good case that your review or synthesis is needed and not duplicating work that has already been successfully and recently completed—or that is in the process of being completed. It is also important to note that what is considered “recent” will depend on your discipline and the topic.

Even if you do find another review or synthesis on your topic, it may be sufficiently out of date or you may find other defendable reasons to perform it again. In addition, looking at other research syntheses published around your topic may also help you refocus your question or redirect your research toward other gaps in the literature.

These tips may not cover every possible type of research synthesis but they provide a decent start.

To find other reviews or syntheses:

Search databases specifically focused on systematic reviews using a very broad search strategy, databases such as:

The Cochrane Library (including systematic reviews of interventions, diagnostic studies, prognostic studies, and more) is an excellent place to start.

By default, the Cochrane Library will display “Cochrane Reviews” (Cochrane Database of Systematic Reviews, aka CDSR). You can ignore the results which show up in the Trials tab when looking for systematic reviews: They are records of controlled trials relevant to Cochrane reviews. There is also an option to see Other Reviews under "More".

The example shows the number of Cochrane Reviews with hiv AND circumcision in the title, abstract, or keywords.

In addition, you can perform a sufficiently developed search strategy on MEDLINE via PubMed (be as broad in your search as is reasonable) and then use the “Article types” filter for “Systematic Reviews” to see if other reviews have been published on your topic. Systematic reviews found in PubMed and not included in The Cochrane Library may require more careful critical appraisal for quality assessment.

See Systematic Reviews Search Strategy Applied in PubMed for details.

Alternative interface to PubMed: You can also search MEDLINE on the Ovid platform. Perform a sufficiently developed search strategy (be as broad in your search as is reasonably possible) and then choose the journal subset for Systematic Reviews Pre 2019 (more sensitive, less accurate -- it is, in fact, a search filter, and not a publication checktag/limit; because it is not a checktag/limit, it is not as accurate) or Systematic Reviews.

The Systematic Reviews search strategy is based on the version used in PubMed, with significant changes made to it in December 2018.

Once you have run your search, to limit to systematic reviews: Go to Additional Limits, and under EBM-Evidence Based Medicine, choose Systematic Review

Other approaches: To search for systematic reviews in Embase on Ovid, perform a reasonably developed search strategy (be as broad in your search as is reasonably possible) and then combine your search results with a systematic review filter developed for Embase on Ovid, e.g., filters available at https://sites.google.com/a/york.ac.uk/issg-search-filters-resource/filters-to-identify-systematic-reviews

Example of a filter used to find systematic reviews for BMJ Clinical Evidence (can be copied and pasted into the Embase search box then combined with the concepts of your research question):

(exp review/ or (literature adj3 review$).ti,ab. or exp meta analysis/ or exp "Systematic Review"/) and ((medline or medlars or embase or pubmed or cinahl or amed or psychlit or psyclit or psychinfo or psycinfo or scisearch or cochrane).ti,ab. or RETRACTED ARTICLE/) or (systematic$ adj2 (review$ or overview)).ti,ab. or (meta?anal$ or meta anal$ or meta-anal$ or metaanal$ or metanal$).ti,ab.

Additional, specialized databases you can search to find existing or in-process reviews

Systematic searching

Knowledge syntheses involve systematically searching the literature. For example,

Systematic reviews of interventions require a thorough, objective and reproducible search of a range of sources to identify as many relevant studies as possible (within resource limits). This is a major factor in distinguishing systematic reviews from traditional narrative reviews (...)

Higgins JPT, Green S (editors). Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0. The Cochrane
Collaboration, 2011. Available from http://handbook-5-1.cochrane.org/.

Systematic searching is:

• Thorough: It involves searching more than one database and using a combination of textwords plus subject headings (the latter when available) to identify relevant literature. It also involves using supplementary search methods in addition to database searching
• Objective: Search terms include variations in terminology and the searcher should avoid biasing the results of the search through the selection of search terms
• Reproducible: The searcher is careful to record all the details of the search, including the database and platform with dates of coverage (may differ based on institutional subscriptions), the search strategy as executed, and the date the searches were run

Selecting databases to search

Need help choosing which database to search? This table shows you the differences between the core databases used in health sciences knowledge syntheses. 

Need help searching the databases? Start with this introduction to searching with subject headings and keywords.

Additional databases

If your question spans multiple disciplines and you would like more information on databases outside of this list, we suggest consulting the subject guides (also known as LibGuides) produced by McGill librarians.

Other high-quality knowledge syntheses published in the area of your research question can also be used for guidance.

Preliminary/exploratory searches in databases you are considering including can also give you a sense of how useful the database will be.

Developing and running the database searches

We recommend developing the search strategy in one database before translating the search strategy to the other selected databases: This will make it easier to keep track of things. If you subsequently find terms in the other selected databases, you can then go back and add them to the searches that have already been developed.

It is recommended that you have your search strategy peer reviewed. For peer review criteria and critical appraisal search strategies, consult:

• McGowan J, Sampson M, Salzwedel DM, Cogo E, Foerster V, Lefebvre C. PRESS Peer Review of Electronic Search Strategies: 2015 Guideline Statement. J Clin Epidemiol. 2016 Jul;75:40-6. doi: 10.1016/j.jclinepi.2016.01.021. Epub 2016 Mar 19. PubMed PMID: 27005575.

We also recommend running all the searches on the same day to make it easier to document the date in your manuscript. 

Once you have your searches developed and you are ready to run them, you can the export the records from each database to an EndNote library, which you will keep for your files.

Subject headings: Definition

A subject heading is an assigned word or phrase used in some databases to uniformly describe a concept. Searching using this standardized word or phrase, instead of keywords, means you do not need to worry about synonyms and spelling variations.

Example: The subject heading for cancer in MEDLINE (via PubMed) is the MeSH term Neoplasms. This means that all articles selected for indexing in MEDLINE that are about cancer at a general level will be tagged or indexed with this subject heading, or if the article is about a specfic cancer like breast cancer, with a narrower term (note: there is a time delay between the addition of records to MEDLINE and their indexing).

We recommend the following resources if you would like to explore subject headings in more detail:

• Using MeSH in Pubmed  (3 min 02)
• Advanced Search on the Ovid Platform (includes information on using subject headings and subheadings) (13 min 25)
• Using Subject Headings in CINAHL (3 min 32)

Keywords: Definition

Keyword (or textword) searching is when we search for words which we expect to find in the title, abstract, or author-defined terms of relevant articles; it is how we typically interrogate web search engines. Draw up a list of words or phrases related to each concept in your research question. When using this technique, you will need to be aware of synonyms and spelling variations.

Example: Keywords (or textwords) for cancer can include cancer / cancers / cancerous / neoplasm / neoplasms / neoplastic / tumor / tumors / tumour / tumours etc.

Subject headings versus keywords

Database and platform cheat sheet: PubMed, Ovid Online, Web of Science

Boolean operators

AND

• retrieves records that include all of your search terms
• e.g., diet therapy AND bulimia

OR

• retrieves records that contain at least one of your search terms
• e.g., (dog OR canine)

NOT

• retrieves records that contain your first term but exclude your second term
• e.g., dementia NOT alzheimer's
• we do not usually recommend that you use NOT in your searches, as you may exclude relevant results

Note: Capitalize your operators as a matter of practice. In some databases, it does not matter whether you enter them in uppercase or lowercase, but others (like PubMed or Google Scholar) require them to be in uppercase.

Boolean operators can be placed between your search terms to narrow or broaden a search, or to exclude search terms.

You will find below a brief video on how to use the boolean operators:

What is a search filter?

Search filters, sometimes also called hedges, are search strategies that usually include a series of pre-elaborated free-text terms/textwords/words/phrases plus subject headings for a given concept, idea, or study design; these search strategies have already been developed to find literature on the concept/idea/study design of interest within a particular database/platform (e.g., for the MEDLINE database on the Ovid platform). They may or may not have been validated for their sensitivity/specificity/precision, but when available, they are a useful tool to take advantage of work that others have already done to identify terms to find literature on a given concept.

Useful sources of general search filters

• InterTASC Information Specialists' Sub-Group Search Filter Resource
• McMaster Health Informatics Unit
• Search filters produced by the Scottish Intercollegiate Guideline Network (SIGN)

Additional filters not listed above or suggested at McGill

• Africa (University of Alberta Libraries)
• Canada and provinces (University of Alberta Libraries)
• Family Medicine
• Indigenous peoples (University of Alberta Libraries)
• Low and middle income countries (Cochrane Effective Practice and Organisation of Care) (last updated in 2012)
• Measurement properties
• Randomized controlled trials

In addition to using search filters or hedges, you may find it useful to consult other systematic reviews or knowledge syntheses related to at least one of the concepts of your search, to see how they developed the search strategy for the concept of interest. One trick, for example, is to search the Cochrane Library by restricting the term of interest to the title field in the record, and then looking at the search strategy in the full-text, as it is often well documented in Cochrane reviews. Ideally, you should check a few reviews given the quality of the searches can be quite heterogeneous.

RCT search filters/hedges

Use these pre-formulated search strategies (database-specific) to limit your results to randomized controlled trials or controlled clinical trials, depending on the strategy available. To do so, just copy and paste the appropriate strategy into the database search box and add it to your search with the Boolean operator AND.

CINAHL (EBSCOhost) filter/hedge for controlled clinical trials:

• See Glanville J, Dooley G, Wisniewski S, Foxlee R, Noel‐Storr A. Development of a search filter to identify reports of controlled clinical trials within CINAHL Plus. Health Information & Libraries Journal. 2019 Mar;36(1):73-90.

Embase (Ovid) filter/hedge for controlled trials:

Source: 6.3.2.2. What is in The Cochrane Central Register of Controlled Trials (CENTRAL) from EMBASE? http://handbook-5-1.cochrane.org/

(NB: does not remove animal studies as the MEDLINE version does)

MEDLINE (Ovid) highly sensitive filter/hedge for RCTs, modified with addition of randomised.ab.

Source: Box 6.4.c, http://handbook-5-1.cochrane.org/

PubMed highly sensitive filter/hedge for RCTs, modified with addition of randomised[tiab]:

Source: Box 6.4.a, http://handbook-5-1.cochrane.org/

McGill students and staff may download EndNote free of charge onto their personal computers at home or at the university. If you have a MAC, please see this notification FIRST.

If you are not sure which OS you have, please click here. Mac users with older operating systems (High Sierra, Sierra, El Capitan etc) download EndNote OS Mojave.

EndNote is installed on all library computers as well as other locations on campus.

FAQ FOR CITATION MANAGEMENT QUESTIONS

This FAQ is set up to accept natural language questions and to get you to the answer to your question as quickly as possible. For now, the FAQ is limited to questions related to EndNote, but this will be expanded in the future.

You can type your question below, or access the full FAQ at http://mcgill.libanswers.com/citation.  We welcome your feedback!

Exporting search results into EndNote

Note: This procedure helps you keep track of the information that will be needed to report the methods in the article and also to fill in the PRISMA flow diagram.

Have the right EndNote library open in the background before you begin exporting.

Note: We do not generally recommend using Safari as your browser during this process, although it can be used with some tweaking.

For more instructions on using EndNote or other software programs, please see the Citation Guide

The steps below cover Ovid MEDLINE, Embase Classic+Embase on Ovid, PubMed, and Web of Science, and CINAHL (EBSCOhost)

Exporting from Ovid Online

(MEDLINE, Embase, PsycINFO, Global Health)

We often start by exporting records from Ovid MEDLINE, if applicable (you may have decided to only use PubMed to search MEDLINE; those instructions are below):

1. In EndNote, create a new library, with a useful name like SearchConcepts-MEDLINE-yyyymmdd-xRecords.enl, for the records from the database in question (e.g., MEDLINE on Ovid) and have the library ready in the background
2. From Ovid, export complete reference in groups of <= 1000 (e.g., 1-1000, 1001-2000), do this until all of your records are exported (Export > Export To: EndNote; Select Fields to Display: Complete Reference; Export Citations)

3. Move batches of exported records to the library you created and check the final number of records in the library against the number of results listed in database
4. From Ovid MEDLINE: Download the search history for your records: Export (1, 1-10) > choose Microsoft Word format, check off Search History, and save the Word document to your documentation folder, with a useful filename in a format like SearchConcepts-yyyymmdd-Search-MEDLINE-xRecords, SearchConcepts-yyyymmdd-Search-Embase-xRecords...
   • Not including the records themselves, the search history can then be copied and pasted into an appendix of the article, to document the search strategy with exactness. Avoids transcription errors
5. Save the search history to your personal Ovid account (free) and identify the search by database name and date; this greatly facilitates rerunning the search later if an update is needed
6. Create a compressed version of the EndNote library for your records

Exporting from PubMed

1. In EndNote, create a new library, with a useful name like SearchConcepts-PubMed-yyyymmdd-xRecords.enl, for the records from the database in question (PubMed) and have the library ready in the background
2. From PubMed, Click on Send To > Choose Destination: File > Format: MEDLINE > Create File

3. In Firefox: Open with EndNote

4. If the file was saved to your computer instead of automatically being imported into EndNote, import the file into EndNote (In EndNote menu: File > Import > File and locate the saved PubMed file on your computer) > Select PubMed (NLM) filter (if not visible, click “Other Filters…” to find it)

5. Save the search history for your records: From PubMed: Click Create alert under the search box or from the Advanced screen, click Download history to maintain the line-by-line version (if applicable). The search in Create alert is saved in an agglomerated format but this is useful if you need to rerun the search later for an update and is also valid when including the search strategy in your manuscript
6. Create a compressed version of the EndNote library for your records

Exporting from Web of Science

1. In EndNote, create a new library, with a useful name like SearchConcepts-WoS-yyyymmdd-xRecords.enl, for the records from the database in question (Web of Science) and have the library ready in the background
2. In Web of Science, Click on the arrow for the Save to EndNote online dropdown menu and select Save to EndNote desktop
3. Export full records in groups of up to 500 (e.g., 1-500, 501-612)
4. Move batches of exported records to the library you created and check the final number of records in the library against the number of results listed in database
5. From Web of Science: Copy and paste the search history into a Word document, adjust the formatting
6. You can save the search history as a file on your computer and this file can later be uploaded to Web of Science to re-execute the search; this is useful later if an update is needed
7. Document which parts of Web of Science you are searching (Web of Science is composed of multiple databases and access to them is insititution-dependent)
8. Create a compressed version of the EndNote library for your records

Create a backup, compressed EndNote library for your records

• Before merging the libraries, create compressed libraries of each of the EndNote libraries from the individual database searches: Keep these compressed libraries for your records (you can also back these up as, e.g., RIS files, to save space)
• Using the uncompressed versions of those libraries, bring all the records from your separate searches in to one EndNote library if you haven’t done so already: From EndNote menu, File > Import > File > locate individual .enl files; Import: EndNote library. You will perform deduplication in this library.

Create a new EndNote library for deduplication and backup after deduplicating

Removing duplicates

See the instructions on deduplicating in EndNote.

Create a compressed library for backup after having removed all duplicates, with a filename like SearchTerms-yyyymmdd-Deduplicated—xRecords.enlx. This will be the library for screening.

Removing duplicates

Systematic review software may offer this functionality, or you can use one of the following methods to remove duplicates from a merged EndNote library.

Before deduplicating, you will need a merged EndNote library containing the records from all your separate EndNote libraries for the individual database searches, if you had previously exported records from each database into separate libraries:

• Create a new EndNote library that will contain the records from all the databases you searched (I like to put DEDUPING in the EndNote library name)
• Import the records from each EndNote library you created for the individual database searches:
  • Go to EndNote menu > File > Import > Import file
  • Next to "Import file", browse to find the .enl file (NOT the .enlx file) for each library and select "EndNote library" as the "Import option"
  • Once all the records have been added to this new library, check to make sure the final number of records, before removing duplicates, matches the sum of the records found in each database search

Using this merged library of records from your individual database searches, you are now ready to remove duplicates. Here are two methods you can use:

1) Earlier version of "Bramer method" for deduplicating, with steps provided in Word document format:

2) Paper describing more advanced configuration options for removing duplicates in EndNote: Bramer WM, Giustini D, de Jonge GB, Holland L, Bekhuis T. De-duplication of database search results for systematic reviews in EndNote. Journal of the Medical Library Association : JMLA. 2016;104(3):240-243. doi:10.3163/1536-5050.104.3.014

After deduplication

Create a compressed library for backup after having removed as many duplicates as possible, with a filename like SearchTerms-yyyymmdd-Deduplicated—xRecords.enlx. This will be the library for screening. You can export the deduplicated library to Rayyan (see Rayyan guide for information on proper format for the EndNote file), for example, or other synthesis software, to facilitate the screening process, which usually involves more than one reviewer.

Important information to document for the search

Record the information in a separate document:

Finding full text

Here are the steps I usually take when tracking down full-text articles. 

First, I like to perform these steps with VPN running in the background if off campus, for Google or Google Scholar in particular;

Given I use EndNote to collect my references (or other citation management software, if the option is available):

• I have usually exported my records to EndNote, so I use the Find Full Text feature in EndNote to automatically download as many full-text articles to my EndNote library as I can (usually in batches no greater than 20, but that's just a suggestion); they get added to the .Data folder but I don't usually access that folder as the article can be viewed and annotated directly from the EndNote reference
  • You can configure file naming conventions for your .Data PDF folder so that they're easier to identify, if you ever want to go in there and make a copy of a PDF, for example (otherwise, the PDFs are very hard to identify): Go to Preferences > PDF Handling and change the PDF Auto Renaming Options
• Instructions on setting up Find Full Text in EndNote

If that doesn't work or you don't use EndNote:

• I do a quick Google Scholar search http://scholar.google.ca for the article title and see if I can quickly get to full-text that way (I need to be using VPN for this to work well if off campus, or at least need to set up Google Scholar with “Find full text” links; 
• If that doesn’t work, I copy and paste the journal title (rather than the article title, although the article title can work too) into the McGill WorldCat Catalogue available at http://www.mcgill.ca/library
  • ​I often use quotations around the journal title for more precise results
  • I recommend using the full journal title over the abbreviated format
  • I often limit the results by format in the left-hand column to Journal/Magazine
  • When I find the right journal record, I click on “Access journal" and select a link that covers the date I am looking for
    • For e-access, I might have to click on a few links before getting to the right place: Pay particular attention to the information about the time coverage available for each electronic source to avoid unnecessary clicking
    • If it’s not available online, I check our print holdings in all relevant records for the journal; if it is only available in print at McGill, I click on “Place hold or request article/chapter scan” (be sure to fill in the citation information for the article or chapter in the item description field), or, if the article is in a journal that's on open shelves, I can go make a copy of it in person
• If it’s seemingly not available at McGill:
  • I do a quick Google search, putting the article title in quotations for precise results (using VPN if off campus)
  • If I can’t find it through Google, I submit an interlibrary loan request: http://www.mcgill.ca/library/services/otherloans/interlibrary