RigorMed
← Research resources

Research methods guide

Meta-analysis data extraction table

Define extraction by study, group, outcome, and time point. For binary outcomes retain events and totals. For continuous outcomes retain sample size, mean, and standard deviation, or use protocol-defined estimates and precision.

Reviews and evidence synthesis

Three synthetic records illustrate binary and continuous outcomes and a shared comparator. Choose the effect measure before defining raw fields and analysis populations.

Three worked extraction rows

Also retain study and report ID, time, unit and source location.

Three worked extraction rows
ItemWorked exampleVerification
BinaryA01, 30-day readmission, 12/100 intervention and 18/98 control.Verify analyzed denominators and distinguish people from repeated events.
ContinuousB02, six-month SBP, 129.4 (SD 12.2, n=80) versus 133.0 (SD 11.8, n=78), in mmHg.Distinguish final from change scores and SD from SE; retain conversion rationale.
Multi-arm trialC03, two intervention arms share a 90-person comparator.Prespecify combining suitable arms or account for correlation without double counting.

Apply the method to your materials

  1. Separate raw and analysis values

    Save table references and raw values. Store conversions, direction changes and imputations in derived columns with methods.

  2. Verify independently

    Have a second extractor verify denominators, time points and units. Flag missing quantities and log author queries.

Questions that arise in practice

What if only a median and IQR are given?

Follow the protocol for author contact or suitable effect measures. Document conversion assumptions and assess sensitivity if converting.

Verified sources

The official sources below were checked on 2026-11-12. Access conditions and update dates are rechecked before publication.

  1. Cochrane Handbook, Chapter 5
  2. Cochrane Handbook, Chapter 10

RigorMed workspace

Start with your research materials

Bring your research question, available materials, and the steps you want to check.