Systematic Reviews: Data Extraction

The data you extract depends on the question you're trying to answer. Different research questions have different data needs. Typical data elements often include: 

• Study characteristics (author, year, title, location)
• Participant characteristics (age, sex, ethnicity, diseases) 
• Methodology (study type, patient recruitment/selection/allocation, level of evidence, study quality)
• Intervention (control, quantity, dose, route of administration, duration, setting)
• Outcomes (unit and method of measurement, length of follow-up, number of patients (included, withdrawn)

• A Practical Guide: Data Extraction for Intervention Systematic Reviews: a guide written by Covidence but not specific to Covidence software.
• Data Extraction (AHRQ): PPT slides with guidance, examples, and advice on how to handle common problems
• PRISMA 2020 Expanded Checklist (Sect. 9-11) describes how to report data collection and data items for outcomes and other variables. For even more detail, read their Elaboration and Explanation
• Cochrane Handbook Ch.5.3: Describes what data to collect
• DECiMAL: Data extraction for complex meta-analysis

The data you extract goes into your evidence and summary tables. These can include: 

• Quality assessment of individual studies (Risk of Bias) 
• Applicability assessment 
• Summary of findings table
• Characteristics of included studies 
• Quantitative data synthesis
• Sensitivity analysis
• Grading the strength of evidence
• Forest plot 

To evaluate the quality of evidence for each outcome, use GRADE. It classifies the evidence in four levels (very low to high) and offers 2 grades of recommendations: "strong" and "weak." 

There are many tools to extract data. You can create your own form in Word or Excel. Some software helps you do it, but you still have to input which data you want to extract: 

• Cochrane data collection form for interventions (RCTs and non-RCTs)
• Covidence: (Available to MD Anderson staff) allows you to create a data extraction form with text fields, single-choice items, section headings and section subheadings.
• RevMan is free software to manage Cochrane reviews. It creates collection forms for descriptive information on population, interventions, and outcomes, and quality assessments, as well as for data for analysis and forest plots. The form elements may not be changed, and data must be entered manually.