Screening
The screening stage in a literature review involves the selection of studies identified during the search stage of the review against inclusion criteria. This stage ensures that only studies which meet the inclusion criteria are included in the final review.
Review guidelines typically recommend a two-stage screening process:
- Title and abstract
The first stage involves an initial screening of the title and abstract against the selection criteria.
- Full text
The second stage assesses the eligibility of the full text of studies that pass the first stage screening.
Alert
The screening process should be planned and documented in your review protocol. Screening should involve at least two researchers to conduct the screening with a plan to resolve discrepancies, for example including a third researcher in the screening of split decisions.
Document your screening process
Documenting and reporting the screening process is essential for transparency and reproducibility of a literature review.
PRISMA flow diagram
Many review guidelines recommend using the PRISMA flow diagram as it illustrates the screening process clearly.
The PRISMA flow diagram enables researchers to report:
- the number of studies identified through the different searches
- the number of duplicate records
- the number of records screened
- the number of full text articles assessed for eligibility.
Note
Researchers should note the number of results from each individual search to record in the PRISMA flowchart. Furthermore, the PRISMA flow diagram also enables researchers to outline reasons for excluding studies from the review.
Screening software
Screening software should be used to manage the screening process. This software allows researchers to upload search results and enables multiple screeners to nominate studies for inclusion or exclusion. Search results should be dedupliated prior to screening, either in a reference management tool or through the screening software.
Below are a couple of screening software that you could apply in your review:
- LitQuest
Developed by Deakin, LitQuest assists with screening and data extraction. Machine learning functionality increases efficiency. For more information check out LitQuest knowledge base.
- Covidence
A software that supports screening, data extraction and critical appraisal. It also has machine learning functionality for efficiency. For further information check out the Covidence page on the Systematic and systematic-like review toolkit.
Caution
Most screening software now incorporates a machine learning element. As researchers select studies for inclusion or exclusion, the software learns the characteristics of studies likely to meet the inclusion criteria. To enhance efficiency in the screening stage, the software then orders the remaining studies to be assessed based on the likelihood prediction they’ll be included in the review.
Title and abstract screening stage
The process of title and abstract screening involves researchers reviewing the title and abstract of each identified study to determine if it meets the selection criteria for inclusion. The results from the searches should be collated and deduplicated in either a reference management tool (e.g. EndNote) or a screening tool (e.g. Covidence).
Depending on the type of the review, this task is usually conducted by at least two researchers, with a method to resolve any disagreements, for example, a third researcher. Screening software such as Covidence or LitQuest can assist in managing this process.
Deduplication
There is often overlap between different databases, leading to duplicate records for the same studies in the search results. Deduplicating these results removes duplicates, providing researchers with a clean set of results to begin screening.
Reference management software like EndNote, and screening software such as Covidence or LitQuest, offer deduplication functions to streamline this process.
Full text screening stage
The second phase of screening involves evaluating the full text of studies that have been identified for inclusion after the initial title and abstract screening.
Screening the full text is similar to the title and abstract screening. The difference is that the entire study is assessed for inclusion. Every inclusion criterion must be met for the study to be included. Once again, depending on the review guidelines, it is usual for at least two researchers to screen with a plan to resolve any disagreement.
The full text of these selected articles should be located and stored in either the screening software or reference management software.
Locating full text resources
Library and online searches
You can locate the full text by performing a title search in Deakin Library’s Discovery layer or via Google Scholar. To facilitate easy access to full texts, set up Deakin in your Library Links within the Google Scholar settings.
LibKey Nomad
LibKey Nomad is a browser extension that connects you to online resources that Deakin Library has access to. This means you can access our journal articles wherever you are on the web. An icon will appear when you come across an online resource that the Library has a subscription to.
Interlibrary loan service
If Deakin does not have access to the full text of an article, you can request it through the interlibrary loan service.
Reference management software
Some reference management software offers options to find the full text. Export the selected references from the screening software in RIS format and import them into the reference management software. Select the relevant references and use the ‘find full text’ function. You will be prompted to authenticate using your Deakin credentials. While this feature can be a shortcut to identifying full texts, it has limited success.
Locating additional resources after screening
Once screening is completed, use the included studies as seeding references for cited reference searching. This additional search uses the reference lists and articles that have cited the seeding references to identify whether there are any additional studies that meet the selection criteria that were not identified through any other search techniques.