FREE TEMPLATE
Free Data Extraction Table Template for Systematic Reviews
Download a free, bespoke systematic-review data extraction template for Excel or CSV — add, remove, or rename columns to match your protocol. Use it to record study characteristics, participants, interventions, comparators, outcomes, numerical results, risk of bias, and source locations before synthesis.
| Study ID | First author | Year | Country | Design | N | Population | Intervention | Comparator | Outcome | Events (I) | Total (I) | Events (C) | Total (C) | Mean (I) | SD (I) | N (I) | Mean (C) | SD (C) | N (C) | Effect size | SE | CI low | CI high | RoB overall | Notes |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
The Excel download includes the fuller field dictionary below (identifiers, citation, risk of bias, funding, and reconciliation columns) in addition to the quick-start columns shown in this table.
Data extraction form, table or spreadsheet: which do you need?
The three terms overlap, and reviewers use them loosely. In practice they describe different layers of the same job, and the download on this page covers all three.
Data extraction form
The list of fields you record for every included study: identifiers, design, population, intervention, outcomes, results, risk of bias. It can live on paper, in a web form, or as one row in a sheet. The field dictionary below is the content of the form.
Data extraction table
A grid with one row per study and one column per field. The template above is this layout. A shorter version of it, usually design, population, intervention, comparator and main outcomes, appears in a published review as the characteristics-of-included-studies table.
Data extraction sheet
The working copy reviewers actually type into — one sheet per extractor in Excel or Google Sheets, following the table layout above, before the two are reconciled into a final version.
Data extraction spreadsheet
The file that holds the table: Excel, Google Sheets or CSV. It is the most common working format for small reviews because it is free and familiar, but it has no built-in dual-extraction tracking or version history.
Using the template in Google Sheets or Excel
- Excel: open the .xlsx file directly. Google Sheets: upload it to Google Drive and open it with Sheets, or use File → Import with the CSV.
- Freeze the header row and the Study ID column so they stay visible while you scroll across the fields.
- Add drop-down lists (data validation) for fields with fixed answers, such as study design and risk-of-bias judgement, so both extractors enter the same wording.
- Keep one copy per extractor and reconcile in a third sheet, rather than editing the same file at the same time.
A spreadsheet works well for a small review with one or two extractors. It does not track who changed what or flag disagreements between reviewers; if you need that, extract in a free EvidenceFlow project instead.
What should a systematic review data extraction form include?
A complete data extraction form records more than the numbers that go into a forest plot. The groups below — also reflected in the downloadable Excel workbook — cover what most Cochrane-style clinical reviews extract for each included study.
Study identifiers and report identifiers
- Study ID (one per underlying study)
- Report ID(s) — a study can have several reports: registration entry, primary paper, follow-up paper, conference abstract
Citation, DOI/registry ID, and source location
- Full citation (author, year, journal)
- DOI, where available
- Registry ID (e.g. trial registration number, PROSPERO ID)
- Source location for the record (page, table, or figure)
Study design, setting, country, and recruitment period
- Study design (e.g. RCT, cohort, case-control, diagnostic-accuracy)
- Country and setting
- Recruitment period
Population and eligibility criteria
- Population description
- How the study's population maps to your protocol's eligibility criteria
Intervention and comparator details
- Intervention (or exposure), dose/duration where relevant
- Comparator (placebo, usual care, active comparator)
Outcome definitions, domain, instrument, and timepoint
- Outcome domain
- Precise outcome definition as reported
- Instrument or measure used
- Timepoint(s) of measurement
Numerical results for continuous and binary outcomes
- Binary: events and totals per arm
- Continuous: mean, SD, and N per arm
- Record raw values before any transformation
Analysis population and effect measures
- Analysis population (e.g. intention-to-treat, per-protocol)
- Effect measure, effect estimate, SE, and 95% CI as reported
Risk of bias
- Tool used (e.g. RoB 2, ROBINS-I, QUADAS-2)
- Domain-level judgements
- Overall judgement
Funding and conflicts of interest
- Funding source as declared by the study
- Conflicts of interest as declared by the study authors
Extraction source location, notes, and reviewer reconciliation
- Source location for each critical value (page/table/figure)
- Extractor notes, including quotes where useful
- Reconciled value once discrepancies are resolved, with a note on how they were resolved
Adapting the template by study type
This is one adaptable starting template, not a single instrument validated for every review design. Adjust the columns below to match the study designs in your protocol.
Randomized controlled trials
Add columns for the randomization method, allocation concealment, and blinding, and use RoB 2 domain judgements. Intention-to-treat is the usual default analysis population unless your protocol specifies otherwise.
Observational studies
Add a column for confounder adjustment: whether the reported effect estimate is adjusted or unadjusted, and which variables were adjusted for. Use ROBINS-I rather than RoB 2, since ROBINS-I includes a confounding domain that RoB 2 does not.
Diagnostic-accuracy studies
Replace the binary/continuous outcome block with a 2×2 test-result table against the reference standard (true positive, false positive, false negative, true negative), and record QUADAS-2 domain judgements instead of RoB 2 or ROBINS-I.
Method note: adapt and pilot this template against your own review protocol before full extraction begins — it is not a validated universal instrument.
How to write a data extraction plan
A data extraction plan is the part of your review protocol that fixes what you will extract and how, before anyone opens a full text. Writing it down first limits the temptation to choose outcomes or timepoints after seeing the results. A workable plan states:
- Outcomes and timepoints. List the outcomes from your protocol and say which measure or timepoint you will take when a study reports several.
- Fields. Choose the fields you need from the dictionary below and drop the rest, so every column has a reason to exist.
- Multiple reports. Say how you will link several papers to one study (one Study ID, several Report IDs) and which report wins if numbers differ.
- Missing or unclear data. Decide in advance when you will contact authors, how you will mark data as not reported, and which conversions you will allow (for example, from a standard error to a standard deviation).
- Who extracts. Name the extractors, state that at least two extract independently, and say how disagreements are resolved.
- Pilot and freeze. Test the form on two or three included studies, fix ambiguous fields, then lock a version and record any later change.
- Source locations. Record the page, table or figure for each value so anyone can check it against the paper.
Chapter 5 of the Cochrane Handbook (cited below) covers these decisions in detail. For the wider workflow, see the data extraction and risk-of-bias guide, or the full field-by-field extraction checklist this template is built from.
Worked example
| Field | Example value |
|---|---|
| Study ID | S1 |
| Citation | Fictional Example et al. (teaching example only) |
| Design | Randomized controlled trial (example) |
| Population | Adults with a fictional example condition |
| Intervention vs. comparator | Example intervention A vs. placebo |
| Outcome, timepoint | Illustrative primary outcome at 12 weeks (example) |
| Events / total, intervention | 18 / 92 (example) |
| Events / total, comparator | 27 / 90 (example) |
| Effect estimate (95% CI) | Risk ratio 0.65 (95% CI 0.39–1.10) — illustrative |
| RoB overall | Low (example) |
Dual extraction and quality control
- Define outcomes, timepoints, comparisons, and effect measures in the protocol before extraction starts
- Pilot the form on a small number of included studies and adjust columns before full extraction
- Keep study IDs distinct from report IDs — one study can have several reports
- Record a source location (page, table, or figure) for each critical value
- Preserve raw values before transformations (e.g. keep events/totals, not just a derived proportion)
- Label missing values explicitly as "Not reported", "Not applicable", or "Cannot determine" — never leave a blank without knowing why
- Use independent extraction for critical data where feasible, with two extractors working separately
- Resolve discrepancies with discussion or a third reviewer, and record how each was resolved
- Version the final extraction file so later changes are traceable
Methodology references
Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372:n71. doi.org/10.1136/bmj.n71. Li T, Higgins JPT, Deeks JJ (editors). Chapter 5: Collecting data. In: Cochrane Handbook for Systematic Reviews of Interventions (current version).
Continue your review workflow
Extraction usually sits between protocol and synthesis. Related free tools and guides:
- Draft a systematic review protocol before extraction begins, so extraction fields match pre-specified outcomes. If the protocol covers trial design, the sample size calculator helps justify your planned N.
- Run title and abstract screening to reach your included-studies list before extraction starts.
- Generate a PRISMA 2020 flow diagram alongside extraction, or read the step-by-step PRISMA 2020 flow-diagram guide.
- Convert extracted means and SDs into a standardized effect size with the effect size calculator, then pool them with the I² heterogeneity calculator.
- Start a free EvidenceFlow review to extract against imported PDFs with dual review and a path into meta-analysis.
Methods write-up: Data extraction and RoB methods guide
Why these columns
The template covers the fields Cochrane-style clinical reviews typically extract: bibliographic identity, PICO elements, dichotomous 2×2 counts, continuous mean/SD/N per arm, a reported effect size with SE or CI, and an overall risk-of-bias judgement.
You can leave unused columns blank. EvidenceFlow's project extraction table uses this same schema, so starting here does not lock you into a spreadsheet if you later import studies into a workspace.
FAQ
Frequently asked questions
Is this template free?+
Yes. Fill it in the browser and download CSV or Excel with no account. Payment is not required for manual extraction later in EvidenceFlow; AI auto-fill is the paid upgrade.
Should I extract OR/RR myself or raw counts?+
Prefer raw events and totals (or mean/SD/N). Pooled analysis can derive OR, RR, RD, MD, or SMD from those counts instead of copying a published effect that used a different model.
How many reviewers should extract?+
Two independent extractors with a third to resolve disagreements is the usual standard. Keep a Notes column for quotes and page numbers from the PDF.
What is the difference between a data extraction form and a data extraction table?+
The form is the set of fields you record for each study; the table is the grid that holds them, with one row per study. This template is a table you can use as your form. A shorter version of the table, the characteristics of included studies, is what usually appears in the published review.
Can I use this data extraction template in Google Sheets?+
Yes. Upload the .xlsx file to Google Drive and open it with Google Sheets, or import the CSV. Freeze the header row, add drop-down lists for fixed-answer fields, and keep separate copies for each extractor so you can reconcile them afterwards.
What should a data extraction plan include?+
The outcomes and timepoints you will take, the fields you will collect, how multiple reports of one study are linked, what you will do with missing data, who extracts and how disagreements are resolved, and a pilot on a few studies before you freeze the form.
Can I add custom fields?+
In the CSV or Excel file, add columns on the right. In EvidenceFlow, project extraction supports extra columns after you start a free project.
Is this a single validated instrument I can cite as-is?+
No. It is an adaptable starting template covering fields common to Cochrane-style reviews. Adapt and pilot it against your own protocol before full extraction — see the study-type notes and quality-control checklist below.
Need collaborative extraction with reviewer-level history and a path to meta-analysis? Start a free EvidenceFlow review.
Start free project