Systematic Reviews

The next step of the SR process is to read the full text of each included study and extract the pertinent data using a standardized data extraction/coding form. According to MECIR Box 5.4.a , data collection for systematic reviews should be performed using structured data collection forms. This is mandatory for Cochrane reviews, but advisable for other systematic reviews.

The aim of a meta-analysis is to collate quantitative data from the studies under review in order to provide a more precise effect of the results. This requires diving deep into the data of each of the publications. Scrutinizing the data can also help you uncover additional bias. For example, are the characteristics of both groups under investigation comparable? Data analysis starts with the extraction of the data from the primary articles. Then, the data needs to be standardized and analyzed. Finally, you need to present your findings in an appropriate manner. Tools exist that can assist in every step of the process. Which data you took, how it was extracted and what analysis you did on them all needs to be adequately described in the report.

Source: Heather Krause, Data Biographies: Getting to Know Your Data

The first step in the process is data extraction. This can be as simple as copying one value of interest from all the articles under review, to collecting large data sets in different formats. Key in this phase is to work meticulously. It can help if the data is extracted by two independent researchers who check each other’s work. Prepare a template that is suitable for capturing all the data.

Tools

• Templates
• DistillerSR
• Systematic review data repository
• Atlas TI
• Maxqda
• Plot digitzer

Once the data is properly retrieved it needs to be standardized so that the results of the different studies can be compared. Then, you can perform the desired analysis on the data. If you do any form of data-analysis (e.g., in a meta-analysis) use statistics to support the validity of your claims. This can be done in Excel in combination with an add-on designed specifically for performing meta-analysis. Alternatively you can do it in JASP, a free tool from the university of Amsterdam designed for statistics on datasets.

We strongly advise to consult a statistician during this phase of the systematic review!

When you performed the desired analyses on the data, you need to present them in a suitable way in the report. The classical way of representing the outcome of a meta-analysis is in a forest plot. Forest plots can be made with excel (Tutorial) or with JASP (Tutorial).

Always double check, to make sure that your tables and figures are legible!