Systematic reviews seek to answer a focused research question by collecting all relevant published and unpublished evidence that meets a predefined eligibility criteria. They are particularly useful for examining clinical interventions, diagnostic tests, biomarkers, and effectiveness of treatment.

Systematic reviews are commonly used to:

• Investigate conflicting results from primary research.
• Evaluate the evidence supporting current clinical practice, while identifying areas where practice may vary.
• Guide clinical decision-making.
• Summarize the entirety of available evidence on a focused research question.

Choosing to conduct a systematic review 

Before starting your project, determine whether a systematic review fits your research based on the guidelines below. 

A systematic review requires these elements: 

• A comprehensive, reproducible search and analysis of all the literature (including gray literature) relevant to the research topic.
• A team of researchers with expertise in the clinical topic area, systematic review methods, search skills, and quantitative methods.
• A willingness to realize undertaking a systematic review takes up a significant amount of time. Depending on team size, scope, and resources, this could be several months to a year.
• A team of at least two independent reviewers, including a third tiebreaker as needed who can use their judgement during the screening process. 

Locating registered and published systematic reviews

Before beginning the systematic review process, search for registered and in-progress reviews on your topic. This helps determine if similar reviews already exist, identify opportunities to update or expand prior research, and avoid unnecessary duplication.

In addition to databases such as PubMed and Web of Science, other recommended databases to search for both registered and published systematic reviews are the PROSPERO international systematic review registry and the Cochrane Database for Systematic Reviews (CDSR).

Guidance

Systematic reviews strive for objectivity and to minimize bias through rigorous, reproducible methodological standards.

Methodology can be found in the Cochrane Handbook for Systematic Reviews of Interventions or the Joanna Briggs Institute (JBI) Manual for Evidence Synthesis, which can be consulted for comprehensive guidance.

Reporting standards are guidelines that can help ensure transparency so that readers can understand and evaluate the methods and findings. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) are the most widely used reporting guideline for systematic reviews. This checklist supports transparent reporting, including methods, search strategy, results, and interpretation. 

Additional learning tools such as Cochrane Interactive Learning can be used to learn more in-depth about the systematic review process. 

Starting a systematic review

Selecting bibliographic databases

When conducting a search for a systematic review, choose databases that are appropriate for the research question. Multiple databases are usually searched, along with gray literature. These might include: 

• PubMed
• Embase
• Web of Science
• Scopus 
• Cochrane Library 
• TRIP

Depending on the topic, a database that is targeted to a discipline or region is also essential to achieving a comprehensive search. These might include:

• CINAHL (Nursing & Allied Health)
• PsycINFO (Psychology & Psychiatry)
• SocIndex (Sociology)
• Latin America and the Caribbean Literature on Health Sciences (LILACS) 
• African Index Medicus (AIM)
• Clinical Trials.gov
• WHO International Clinical Trials Registry Platform

Preliminary topic investigation  

Conducting a preliminary topic investigation before you undertake a systematic review is essential to understand the scope of the literature that has already been published on your research topic. 

This process typically involves:

• Searching relevant databases, such as PubMed,  Scopus, and Cochrane, to understand terminology and indexing terms, estimate the literature volume, and determine if recent systematic reviews already exist. 
• Reviewing previous systematic reviews to see whether there are any easily identifiable gaps in the research. 
• Applying the FINER (Feasible, Interesting, Novel, Ethical, Relevant) criteria to assess the practicability of the project’s topic. 

Research frameworks 

Using a research framework is one way to format your research question. A popular framework for systematic reviews is PICO (Population, Intervention, Comparison, Outcome), but it’s important not to force your research into a framework. Instead, you can break the question into separate concepts of your choice. The appropriate framework, if that’s the approach you take, depends on the review question and study design. For example, PEO may be used for exposure/qualitative questions, PCC is commonly used for scoping reviews, and SPIDER may be useful for qualitative or mixed-methods evidence synthesis.

Predefined inclusion/exclusion criteria

Your inclusion and exclusion criteria will be the foundation for why the research articles that are selected through screening are included in or excluded from the systematic review. Inclusion and exclusion are based on the core aspects of the developed research question. Well-defined inclusion and exclusion criteria strengthen the transparency of a screening process. 

Inclusion criteria establish the key parts of a systematic review research question to determine which results will be analyzed. These criteria can include characteristics of the population studied, the type of intervention, the setting, and type of study design. A restriction (date range, language, study design, etc.) should be scientifically justified because restrictions may introduce bias. Elements of the inclusion criteria, such as date range, should have an explained rationale because each element may introduce bias. Refer to Chapter 3 of the Cochrane Handbook for Systematic Reviews of Interventions or JBI Manual for Evidence Synthesis. 

Exclusion criteria establish key disqualifying characteristics that make a study ineligible for inclusion in the review.  Since systematic reviews aim to mitigate bias, any aspects that will limit the search will require a documented rationale for the decision. 

Search strategy 

Before registering a protocol for your systematic review, it’s important to develop a search strategy that outlines the databases to be searched, the search concepts, and the approach. These may be refined after the protocol registration, as the search strategy is tested and peer reviewed. All of the final executed search strategies should be reported in the completed review. A  systematic and reproducible search strategy should balance comprehensive retrieval with the relevance of results. Refining the search strategy in collaboration with team members or getting advice from an Informationist at the Health Sciences Library — or both — is essential. The search strategy should also be evaluated using PRESS (Peer Review of Electronic Search Strategies).

Search updates

Depending on the length of the systematic review, the search should be updated by rerunning the search strategy in the original databases and documenting the new search date. This update should occur before publication.

Registering a protocol

Creating a protocol to follow as you research and write your systematic review is considered best practice by both the Cochrane Handbook for Systematic Reviews of Intervention and the JBI Manual for Evidence Synthesis. Developing a protocol helps reduce potential bias by establishing methods in advance and publicly documenting the review plan topic by registering that protocol (e.g., in PROSPERO). Many journals require or strongly encourage protocol registration prior to manuscript submission. 

When writing a protocol, you’ll want to: 

• Follow the PRISMA-Protocols (PRISMA-P) template.
• Register the protocol in PROSPERO.

After registering, protocols can be amended, but any such changes should be documented and justified.

Covidence

Covidence is a systematic review management tool that supports key stages of the review process, including importing citations and search results, removing duplicate records, title and abstract screening, full-text review, risk-of bias assessment, and data extraction. While Covidence helps organize and streamline the review workflow, reviewers are responsible for applying methodological decisions based on the predefined protocol and eligibility criteria. You can sign up for Covidence through the Health Sciences Library.

Next steps of a systematic review

1. Execute finalized search strategy in each selected database, document the search date, database platform, and exact search strategy used.
2. Screen titles and abstracts, followed by full-text screening, based on predefined inclusion and exclusion criteria. 
3. Assess each study for bias and quality using critical appraisal tools that are appropriate for the included study designs.
4. Extract predefined data from each study that was chosen for inclusion.
5. If appropriate for the review, conduct a meta-analysis. 
6. Interpret results, while following the methodological guidelines (Cochrane Handbook for Systematic Reviews of Interventions or JBI Manual for Evidence Synthesis) 

For further guidance on next steps, refer to Cochrane Interactive Learning or Cochrane Handbook for Systematic Reviews of Interventions.