Once you've completed the critical appraisal of your included articles, it is time to start reviewing all of the articles more in depth and pulling out usable information for your write up. Data extraction is a step in the review process that requires rigid workflows to ensure that all of the included studies are analyzed in the same way each time.
While Covidence has the option of completing data extraction, we will not be using it for this purpose.
Finally, because of the nature of this group assignment, you will need to analyze the qualitative and quantitative articles separately, as they use different methodologies.
When it comes to completing data extraction, there are a few things to remember about the steps:
The data that gets collected during the extraction phase usually serves a few purposes. You may collect all of the elements to craft a citation later, map each article to how it responds to the PICO(T) elements, additional data points your group is interested in analyzing, detailed statistical elements of articles, and more.
Data extraction can be very robust, so make sure to focus on collecting the elements that your group has enough time to then analyze and make conclusions about. The timeline for this assignment will dictate how detailed you're able to get with your data extraction, but that should not curb your creativity!
While templates (see below) can give you an idea of different data points to extract, the best resource for getting to know what should be analyzed in a systematic review is the Cochrane Handbook's Chapter 5.3 What Data to Collect and all of it's sub-chapters (5.3.1-5.3.7).
Qualitative vs. Quantitative Data Extraction
As a reminder, you'll want to collect and tabulate qualitative and quantitative data separately. Qualitative data is usually brought together in a narrative/thematic review style, while quantitative data is usually brought together with a meta-analysis.
This may lead to separate visualizations and conclusions in the review - as data being gathered using the different methodologies may not result in the same recommendations for practice.
Example Chart to Include In Your Final Paper
|Citation||Year of Publication||Sample Size||Recruitment Format||Start/End Date of Data Collection||Attrition Rate||Country of Study||Study Type||Intervention Type||Follow-Up Timeline||Age Range||etc.|
|Person, A. et al.||2015||150||January 14, 2014 - June 21, 2014||2 individuals dropped out||Canada||Cohort||CBT||6 month follow-up||18+|
|Human, B. et al.||2019||10||July 1, 2015 - December 20, 2018||None||United Kingdom (England)||Cohort||CBT/Exposure Therapy||6 and 12 month follow-up||17-24|
|Individual, C. et al.||2021||121||Social Media||November 1, 2018 - February 28, 2019||40 individuals dropped out||Iran||Cohort||CBT/Journalling||No follow-up||25-40|
|Author, D. et al.||2000||268||Email, Social Media,||November - December 1998||10 individuals didn't complete treatment||United States||RCT||CBT||1 week follow-up||16+|
For systematic reviews that are in the process of being published, you can see a lot of time and effort is put into creating a validated and reliable tool to conduct data extraction. The data points are often outlined at the beginning planning phase of the review, and they are tested multiple times for strong inter-rater reliability before doing the final extraction of included articles.
For the purposes of this assignment, you are short on time, so your group should work within the timeline you have and create something that will give you robust enough analyses to make conclusions, but may not exactly resemble something you would have used if your review was being published.
There are different ways you can put together a data extraction tool for your group to use. The easiest way is to create a spreadsheet in Excel or Google Sheets (so you can share within your group). You will be able to list each included article as a row, and extracted data points as columns.
Alternatively, you can use a Google or Qualtrics form for capturing the data from each included article. This may not be as time efficient to use - but there are many good examples out there already. Your group would need to create their own that fits with the data points you are looking to analyze.