Once the decision has been made about which articles are to be included in the review, relevant data from each of those studies needs to be extracted. This should be done in a structured, systematic way and will usually be presented in tables in your final review report.

Tips

• Only extract the data that is relevant to your research question
• Create a table or data collection form so that you can summarise the data consistently for each study
• Convert all the data to the same units of measurement, where possible
• Ensure your table headings are easily interpreted by others
• Ensure any abbreviations or acronyms used in the table are explained in the footnotes to that table
• Pilot your data extraction method before you start – especially if more than one person is extracting the data

Useful Tools

Your data can be extracted into a simple table, but if you want a more detailed approach, there are a number of analysis tools that can help. See the following:

• NVivo (CSU supported and great for qualitative data analysis)
• PSPP or JASP (Open source software)
• Elamin, M. B., Flynn, D. N., Bassler, D., Briel, M., Alonso-Coello, P., Karanicolas, P. J., . . . Kunz, R. (2009). Choice of data extraction tools for systematic reviews depends on resources and review complexity. Journal of Clinical Epidemiology, 62(5), 506-510. doi:10.1016/j.jclinepi.2008.10.016

Study Design

Sometimes it's difficult to identify the study design of a particular paper. The Centre for Evidence Based Medicine in the UK provides a useful summary of study designs and their characteristics on their website.

More Information

• Cochrane Handbook – Chapter 5 Collecting Data

This is the section of your review where the findings responding to your initial research question(s), are determined. You will need to develop a synthesis, or critical overview, which integrates the key findings from all the studies you included while considering the methodological quality and other pertinent features of each of those studies (such as the sample size, population or context in which each one was conducted).     

Methods 

• Narrative, or descriptive: If you are not completing a full systematic review, or if you are comparing different study designs, this could be the best way for you to go.  There are no strict rules, but ensure you minimise bias and maximise credibility by carefully exploring the relationships between studies, assessing the robustness of the synthesis and theorising about the answer to your research question.
• Qualitative synthesis methods: Meta-synthesis, Meta-study or Meta-aggregation. These methods synthesise the findings from very different types of studies, clustering common characteristics into categories or using interpretative tools.
  • See Cochrane’s Core Library of Qualitative Synthesis Methodology for more information.
• Meta-analysis: To thoroughly synthesise the results of quantitative studies, this statistical technique results in an estimate of the “average” intervention effect. Data from different studies are weighted depending on the sample size and relevant criteria, and evaluated to work out a cumulative outcome. The results of a meta-analysis are often summarised in a forest plot. See:
  • What is a meta-analysis?
  • How to read a forest plot

Tools     

• JBI SUMARI - includes tools for data extraction and synthesis. See Access to JBI SUMARI for access instructions
• OpenMeta[Analyst] - open source, cross-platform software for meta-analysis
• OpenMEE - open source software for meta-analysis for ecologists and evolutionary biologists
• EPPI Reviewer 4: software for research synthesis - for the social sciences: you can sign up for a free one-month trial.
• The Risk of Bias 2 (RoB 2) tool - Cochrane tool used to reduce risk in Systematic Reviews.

Resources         

• Centre for Reviews and Dissemination, University of York. (2008). Systematic reviews: CRD's guidance for undertaking reviews in health care. (See Section 1.3.5 Data synthesis, pp. 45-76)