Systematic Reviews

Partnering with a librarian

• Saves researchers time
• Ensure a comprehensive and rigorous search
• Gets maximum retrieval of relevant articles
• Minimizes article screening time by excluding non-relevant articles from the search strategy

Turning a research question into a search string is a multi-stepped process:

1. Break down your research question into searchable concepts
2. Write down a list of keywords and synonyms for each concept
3. Map each concept to relevant controlled vocabularies for each database (such as MeSH)

As you do test searches and read more relevant literature, you will be able to add additional keywords to your search concepts. It is recommended that you reach out to a librarian for assistance in generating keywords and mapping concepts to a controlled vocabulary.

Example Research question: Does obesity lead to heart disease and strokes?

It is important to finalize your inclusion and exclusion criteria after finalizing your research question but before developing your search strategy. Well-defined inclusion/exclusion criteria

• Reduce bias in the review
• Help craft a more targeted search strategy

Types of criteria could include

• Type of study participants
• Study design
• Intervention of interest
• Outcomes of interest
• Study setting
• Type of publication
• Date of publication
• Language of publication

When you are first refining your research topic and looking at the literature, you will most likely come across articles of interest that you know should be included in your review. These are referred to as exemplar articles. Exemplar articles can be used to test your search strategy and ensure that your search strategy is retrieving the kinds of literature you want to include in the review. In addition, citation chasing from exemplar articles is a great way to identify additional relevant literature. See below for some tools that can take advantage of your exemplar articles.

Databases

A systematic review aims to review as much of the literature as possible to minimize bias. As such, it is recommended that at least 3 databases are searched for a systematic review. 

As PubMed/MEDLINE excludes newer journals and health sciences disciplines, additional databases must be used to minimize publication bias. A librarian can suggest additional databases relevant to your research topic. See below for a video demonstrating how to use the library website to find databases relevant to your research topic.

Gray Literature

Gray literature is a term used to describe all sources of research or research-related literature that exists outside of peer-reviewed, academic journal articles.

Examples of grey literature include: conference abstracts, presentations, proceedings; regulatory data; unpublished trial data; government publications; reports (such as white papers, working papers, and internal documentation); dissertations/theses; patents;  and policies & procedures.

Why Search Gray Literature?

• To find information not available in traditional publication outlets
• To reduce publication bias in your results
• To get a complete picture of your research topic

During the planning stage, it is recommended that the team searches for existing systematic reviews on related topics. This helps to ensure that you are not duplicating an existing review (a review older than 10 years old can be updated), and can also be used to help inform the search strategy and inclusion/exclusion criteria you can use in your own review. Related systematic reviews can help you generate search terms and identify databases and journals of relevance to your topic. In addition, viewing the citations of those reviews might help you identify relevant research.

Many databases include search filters or limiters that will limit your search to just systematic reviews. If that is not available, searching for the phrase "systematic review" in the title of your search results will also work.

Once all terms have been identified, you need to put them together in a search string. You can export your search strategy in addition to the results, to use in your search documentation.

A search string will generally look like:

(Topic A term 1 OR Topic A term 2) AND (Topic B term 1 OR Topic B term 2) AND (Topic C term 1 OR Topic C term 2)

If searching PubMed with our example research question, the search string would look like:

(obesity OR overweight OR obese OR "morbidly obese" OR "Obesity"[Mesh] OR "Obesity, Morbid"[Mesh]) AND ("heart disease" OR "cardiac diseases" OR "heart disorders" OR cardiovascular OR "Heart Diseases"[Mesh]) AND (stroke OR "cerebrovascular accidents" OR "Stroke"[Mesh])

The search string above was developed for PubMed. When adapting the string for another database, you want to have the strings operate as similarly as possible. You would replace the MeSH terms with the controlled vocabulary of the other databases used.

The search string above is searching with both keywords and MeSH terms. The MeSH terms will be searched in the MeSH field. The keywords will be searched in all fields, like the title, abstract, journal name, etc.

In a systematic review, you will need to keep the search as similar as possible between different databases. In practice, you should only need to change a few things for each database:

• Applying consistent search limiters to each database
• Searching the same fields in each database
• Applying the controlled vocabulary of each database

As experts in database searching, librarians can help you translate your search string between different databases.

After the search strategy is developed, the search is translated and run in the identified databases and search engines. 

The results from each resource is transferred to a separate dated file in the citation manager.

The team members then perform a blind screen of the pilot search results.

The purpose of the pilot search and screen are to: 

• Ensure sensitivity and specificity of search strategy
• Provide feedback for necessary refinement of search strategy
• Ensure consistency among screeners for inclusion/exclusion criteria

Citation chasing is a technique you can use to identify relevant literature that may be missed in traditional systematic search methods. Citation chasing uses the citations of relevant articles to identify additional articles of relevance. 

Citation Chasing Process

1. Identify highly relevant articles for your review
   1. Consider using your exemplar articles
2. Review the references of your chosen articles
3. Identify articles relevant to your review and include them in the screening process
4. Review articles that have cited your chosen articles
5. Identify articles relevant to your review and include them in the screening process

There are some helpful tools that can streamline and automate this process.