It is advisable to develop a data extraction form a priori including the variables to be collected and clear definitions for them. The form can be pilot tested by review team members to increase the reliability of the data extraction process. Having two or more people extract study data independently decreases the potential for error. Reviewers should consider contacting authors to verify assumptions made for missing or unclear information.
Previous Section: Assessing Risk of Bias of Included Studies