Data Extraction Form for Systematic Reviews: Free Template, Fields and Examples

A data extraction form is the table you fill in for every included study. If it is designed well, your results section almost writes itself. If it is designed badly, you end up rereading 70 papers at the end of the review.

In short: a data extraction form records the same details from every included study: identification (author, year, DOI), aims, design, setting, participants, intervention or exposure, comparison, outcomes and how they were measured, results relevant to each review question, limitations, risk-of-bias items and where in the paper each value came from. Map every field to a review question, pilot the form on a few studies, and have a second person check key values. Use the template below, or extract into a table with the data extraction agent and verify each value.

Data extraction form template

Copy these fields into a spreadsheet, one row per study. Remove fields your review does not need and add review-specific ones.

Fields for different review types

• Intervention reviews: add intervention details (dose, duration, delivery, adherence), follow-up times and numbers analysed in each arm.
• Qualitative reviews: replace numerical results with themes, participant quotes and the authors' interpretations, and add the theoretical framework and data analysis method.
• Diagnostic accuracy reviews: add index test, reference standard, threshold and the 2×2 table (true and false positives and negatives).
• Prevalence reviews: add case definition, sampling method, numerator and denominator.

How to design and pilot the form

1. Start from your review questions. Each field should help answer a question, describe the evidence base or assess quality. Dr. Faheem Ullah's form maps fields directly to individual research questions, so the evidence table answers them without rereading papers.
2. Define every field. Write a one-line instruction and allowed values (for example, design: RCT / cohort / qualitative) so two people extract the same way.
3. Pilot it on 3–5 studies. Two reviewers extract independently, compare, and fix fields that were unclear or missing.
4. Separate reported results from interpretation. Use one field for what the study reports and another for your notes.
5. Record where each value came from. A page or table reference makes checking and updating much faster.
6. Lock the form and log changes. If you must change it mid-review, record the change and re-check earlier studies.

Single or double extraction?

The Cochrane Handbook recommends that at least two people extract data independently, at least for key outcomes and results, to reduce errors. Rapid reviews often use one extractor with a second person checking. Report which approach you used in your methods.

Tools for data extraction

• Spreadsheets (Excel, Google Sheets): flexible and free; fine for most reviews.
• Systematic review software such as Covidence, EPPI-Reviewer or SRDR+: build forms, assign extractors and resolve conflicts.
• AI-assisted extraction: pull predefined fields from many PDFs into a table at once. Treat it as a first pass. Check every value against the paper, and watch for results taken from the abstract or introduction instead of the results section.

In this SciSpace webinar, the presenter builds a data extraction form and then extracts data from 73 papers with AI:

Key moments: 12:27 snowballing · 14:40 building a data extraction form · 17:11 analysing extracted data with thematic analysis · 21:10 extracting data from 73 papers at once with AI. For designing an extraction framework around a research model, see this session, including how AI flags papers where only the abstract was analysed.

Watch: from data extraction to a research model

This SciSpace webinar picks up after extraction, showing how extracted data feeds synthesis and a research model.

Related guides

Frequently asked questions

注意转载授权 · 阅读原文