Skip to main content

Nursing Resources : Systematic Reviews and Meta-Analysis

Steps to Conduct a Systematic Review of Meta-Analysis

1. Identify a suitable supervisor and a clinically relevant topic of interest with PICO (Population, Intervention, Control, Outcome)

2. Check to see if a similar systematic review or meta-analysis has already been done (e.g. in PubMed or Google Scholar)

3. Register on PROSPERO

4. Create search terms for use in relevant databases (e.g. a combination of PubMed, PsycInfo, EMBASE (if available), Scopus, CINHAL, Cochrane, Google Scholar, etc.)

5. Consider a search of databases in other languages (e.g. http://oversea.cnki.net/)

6. Retrieve titles into a bibliographic software program (e.g. EndNote)

7. Remove duplicate references and note number of remaining studies

8. Consider using a systematic review software program (e.g. Rayyan or Covidence)

9. Two people independently review references at the title and abstract level as either ‘in’ or ‘out’ depending on whether they have met the criteria you listed in PROSPERO

10. All references that have been marked as ‘in’ by either reviewer should have full-text articles retrieved, with number of ‘in’ studies recorded

11. All full-text articles should be independently reviewed by two people as ‘included’, ‘excluded’ or ‘maybe’

12. For all articles where there is no consensus, a discussion between the two independent reviewers or a third reviewer should be undertaken

13. Create a table of excluded studies (with author/year and reason columns, e.g. wrong study design, different population from that of interest, did not contain usable data) with number of excluded studies recorded

14. For all included studies you will need to create a table of included studies (e.g. study author, date, setting, age of subjects, their gender, diagnosis, the comparison group where relevant, the intervention where applicable and outcomes)

15. Assess risk of bias in individual studies (e.g. Newcastle-Ottawa Scale for observational or Cochrane Risk of Bias Tool for randomized controlled trials)

16. Create PRISMA diagram with included and excluded studies

17. Extract data of interest from included studies. One person should extract data and a second person should validate it

18. If planning a meta-analysis use relevant software (e.g. RevMan)

19. Consider heterogeneity and publication bias

20. Write manuscript with reference to PRISMA checklist

21. At all stages meet regularly with a supervisor

 

Review Ebling Library's Systematic Review Subject Guide

Utilize software such as Covidence to organize items and inclusion/exclusion criteria