Scientific integrity

Evidence review

Systematic review and critical appraisal with explicit, reproducible methods.

Method

How a review is conducted

  1. Register a protocol. Define the question, eligibility criteria and planned methods before searching, and register the protocol.
  2. Search comprehensively. Search multiple databases and trial registries with a documented, reproducible strategy.
  3. Assess risk of bias. Use RoB 2 for randomised trials and ROBINS-I for non-randomised studies.
  4. Synthesise appropriately. Pool results only where studies are similar enough, and explore heterogeneity rather than hide it.
  5. Rate certainty. Rate the certainty of each key outcome with GRADE and present it in a summary of findings table.
  6. Report fully. Report against PRISMA 2020, including the flow of studies and every deviation from the protocol.
  7. Appraise others’ reviews. Appraise existing systematic reviews with AMSTAR 2 before relying on them.
Tools

Appraisal and reporting tools

Handbooks

Reference handbooks

Textbook

Introduction to Meta-Analysis

Michael Borenstein, Larry V. Hedges, Julian P. T. Higgins et al.

2nd editionWiley2021
Examples

Systematic reviews and meta-analyses in the library

All evidence reviews

Need a review conducted or appraised? Contact the Evidence Review unit.