Systematic Review Service

This step is closely tied to Step 6: Assess the Quality. Parts of these steps are intertwined or simultaneous.

Data collection, also called data extraction, is performed independently by two reviewers for each included study.

First, members of the team, including the statistician if there is one, must work together to determine which data to collect from the studies that will be included in the review. This should be completed at the protocol stage and included in the protocol itself.

It is strongly recommended that once you have determined the data to extract, the team pilot the extraction template on several studies you know you want to include (different types of studies if applicable) to make sure it functions as anticipated and is understood by everyone performing extraction.

Then, collect the data you need to answer your research question. Frequently collected data includes but is not limited to:

• Basic study information: Authors, publication date, clinical trial number, funding sources.
• Study type: Prospective, retrospective, mixed methods, case series, controlled trial.
• Study characteristics: Setting, location, recruitment methods, eligibility criteria.
• Participants: Number included, number lost to follow-up, sex, age, race, ethnicity, diagnosis/es, comorbidities.
• Interventions/Exposures: Dosage, time frame, method of administration, staff performing intervention, follow-ups, exposure times.
• Comparisons: Placebo, standard-of-care, no comparison given.
• Outcomes: Measurements used, effects measured, time frames, adverse events.

Systematic reviews collect data from studies, not publications. If the same study is represented in multiple publications, determine which one is the most up-to-date or complete, and use that report to extract the data of interest.

If a publication does not include the necessary information, contact the authors to see if they can provide it. If they cannot, or if they do not reply, list this as a reason for exclusion.

Learn more about data collection from the Cochrane Handbook and the Institute of Medicine of the National Academies.

You need a strategy for data synthesis, ideally from the time of writing your protocol.

This is the criteria and methods for combining data for quantitative (e.g., meta-analysis) or qualitative (e.g., narrative summary) synthesis. You will also want to consider tables for evidence summaries based on inclusion criteria.

Note that the findings from your quality appraisal and certainty assessment will impact data synthesis as well.

Your MSK librarian can help create a separate Covidence project for pilot testing your data collection template.

Data collection or extraction is a step many researchers have difficulty with, so Covidence created a free eBook on this for researchers to download: A Practical Guide to Data Extraction for Intervention Systematic Reviews.

Covidence offers a fully customizable data extraction template. Once your team has finalized the template, Covidence has two reviewers perform the extraction, followed by conflict resolution either by consensus or by a third reviewer.

Learn more: