Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gotzsche PC, Ioannidis JPA, Clarke M, Devereaux PF, Kleijnen J, Moher D. The PRISMA statement for reporting systematic reviews and meta-analysis of studies that evaluate health care interventions: explanation and elaboration. PLoS Med. 2009; 6(7): 1-28. doi:10.1371/journal.pmed.1000100.
Steps for writing a systematic review
1. Formulate a research question
Consider whether a systematic review is needed before starting your project. Has someone already written one on your topic? Librarians can help you find out. Also, decide if you have enough time and resources to conduct a systematic review. Keep in mind that it could take longer than a year to complete if you are working alone without any source of data. Identify a team of collaborators to work with you or probably your supervisor could help you with information. This reduces the risk of bias.
Begin your systematic review by putting your topic into the “Well-Built Clinical Question” framework. Your question should address 4 basic elements, known as PICO. Refer to our Evidence-Based Practice guide for a review.
2. Develop a research protocol
A research protocol is a detailed plan for how you will study a biomedical or health sciences problem. Systematic review protocols should include details such as:
- objectives of your project;
- specifics on the methods and processes that will be used;
- eligibility criteria for individual studies (such as study design);
- how data will be extracted from individual studies; and
- what analyses will be performed.
Consider registering your review protocol for free with PROSPERO. PROSPERO accepts reviews with health-related outcomes, including topics related to:
- health and social care,
- public health,
- crime and justice, and
- international development.
3. Conduct literature search
The Institute of Medicine recommends partnering with a librarian when developing your search strategy. Your goal is to find all the relevant studies on your topic, so your literature search should be thorough. A librarian will help you select databases relevant to your topic and create a detailed search strategy. You also want to search outside the normal academic publishing model. This is commonly known as “grey literature.” This might involve searching conference proceedings, browsing pharmaceutical company websites, and contacting experts in the field. The grey literature is important because negative results are less likely to be published. This will affect the reliability of your findings.
Keep detailed records as you search. Document details such as:
- databases you searched and what years those databases covered;
- dates the searches were first run and updated;
- details of strategies used, including search terms; and
- numbers of results obtained.
You may also want to use a citation manager. These tools can create bibliographies, remove duplicate records, and share citations with a team.
4. Select studies per protocol
The screening of studies should be performed by 2 reviewers. Use the criteria documented in your protocol. The first round of screening can be based on titles and abstracts. A second round of screening should be conducted by reviewing the full-text articles for selected studies. Reviewers should keep a log of excluded studies, including the reason they were omitted.
5. Appraise studies per protocol
A team of at least 2 reviewers should evaluate the methodological quality of a selection of full text articles. Use a checklist to determine whether studies meet the criteria of the protocol. Questions you might want to consider include:
- were the patients assigned to groups using a random allocation;
- was the allocation sequence concealed to patients and clinicians;
- could the conclusions of the study have been influenced by bias?**
6. Extract data
Create a data extraction form. Assign at least 2 reviewers to extract data from the included studies. Cochrane has several templates that researchers can adapt for their projects.
7. Analyze results
Create a table of study results and prepare a forest plot to demonstrate the relative strength of treatment effects, if relevant. Analyze data for issues such as variation across studies (heterogeneity) and sensitivity of findings. Consider running a meta-analysis of results. Make list of excluded studies available to interested readers.
8. Interpret results
Consider limitations (including biases), strength of evidence, applicability, economic effects, and implications for future practice or research. The Cochrane Handbook provides detailed guidance on interpreting results and drawing conclusions.
Stedman TL. Stedman’s Medical Dictionary. Philadelphia: Lippincott, Williams & Wilkins; 2006.