Systematic Literature Review: Screening

Systematic review screening software includes https://silvi.ai/ (link opens in a new window) and Rayyan (whose most features are free; the link opens in a new window). Carrie Price, a health professions librarian at Towson University, has produced an excellent guide on using Rayyan for screening and identifying duplicates (link opens in a new window), and a guide for Silvi AI is here: Silvi AI (link opens in a new window)

Reference management tools such as EndNote Web, Mendeley, and Zotero are also helpful in removing duplicate references.

EndnoteWeb (link opens in new window)
Mendeley (link opens in new window)
Zotero (link opens in new window)

Screening large numbers of articles can be complicated and time-consuming. Establishing a well-structured, transparent process for all reviewers to follow is a good idea. This journal article, Best Practice Guidelines for Abstract Screening Large-evidence Systematic Reviews and Meta-analyses (link opens in a new window), suggests some best practice processes to follow when screening for a systematic review.

Systematic review searches typically yield thousands, or even tens of thousands, of search results. Only a small handful of these will likely be relevant and ultimately included in the review. This journal article, Precision of Healthcare Systematic Review Searches in a Cross-sectional Sample (link opens in a new window), finds that, on average, most systematic reviews only include 3% of all results found in the search, with the majority being discarded at the screening stage. This is why it's so important to have robust screening criteria and a well-organized screening system.

After you have completed your database and grey literature searches and identified other papers through hand-searching, you will need to screen the titles and abstracts to determine if they meet the criteria for inclusion. These criteria should be predefined (i.e., stated in your protocol before you begin the searches). Inclusion criteria might relate to the following:

Date range of publication
Study design type
Whether a study focuses on your specific disease, condition, or patient population
Whether a study focuses mainly on your specific intervention
Whether a study took place in a particular country or a hospital context
This list is not exhaustive, and there are many other inclusion criteria you may wish to use. I think it's essential to be stated clearly in the Methods section of your protocol and systematic review, and that all co-authors understand them.

Generally, articles are screened against these criteria independently by at least two authors. Initially, you should screen the titles and abstracts, which will leave you with three groups of articles: include (those that meet all inclusion criteria), exclude (those that do not meet at least one of the inclusion criteria), and doubt (those for which it is unclear from their title and abstract if they meet all inclusion criteria). Once you have screened the titles and abstracts in this way, the full text for each article in the Doubt group should be obtained, and the same procedure should be applied. Therefore, at the end of the screening process, you will be left with two articles: include and exclude.

When preparing your systematic review, you should include a flow diagram illustrating the total number of studies identified and the number excluded at each screening stage. To generate PRISMA flow diagrams, you can use a standard PRISMA template for this diagram and the Key PRISMA 2020 documents (link opens in a new window).

When preparing your systematic review, you should include a flow diagram illustrating the total number of studies identified and the number excluded at each screening stage. You can use a standard ROSES template for this diagram and ROSES Flow Diagrams (link opens in a new window) to generate ROSES flow diagrams.