GRADE and 'Summary of Findings' tables

GRADE (Grading of Recommendations, Assessment, Development and Evaluation) is a common, sensible and transparent approach to grading quality of evidence and strength of recommendations in healthcare. 

GRADE allows you to evaluate the quality of evidence you are analysing. You can present this evaluation in a clear, transparent format using GRADEPro to create Summary of Findings tables.

 The newly revamped Cochrane Training website hosts a series of presentations to guide you through this process, beginning with an introduction to the approach and moving through a step-by-step guide on how to GRADE the evidence.

'Summary of Findings' tables (SOFT) 

SOFT should be planned at protocol stage and created after pooled analysis but before the results section is written.

Findings should be consistent and embedded throughout the review in:

  •                Abstract
  •          PLS
  •          Results – effects of interventions
  •          Discussion – quality of the evidence
  •          Authors’ conclusions – implications for research
You should include one table per comparison (not per outcome, as all the outcomes should be in the same table).

You should include up to 7 clinically important outcomes consistent with the review objectives. Prespecified outcomes should be included even if there is no data.

All GRADE decisions should be clearly described and quality assessed by two unbiased authors. Each Grade decision should be clearly justified. 

Results should be presented even if there’s no metaanalysis (i.e narratively).

Even if GRADEPro is used SOFT still needs manual editing.

Should include:

  • Comparison
  • PICO and setting
  • Length of follow-up
  • Outcome measurements
  • GRADE criteria

Downgrading criteria could include:

  • Study limitations (risk of bias)
  • Inconsistency (heterogeneity)
  • Indirectness
  • Imprecision
  • Publication bias - based on searching trial registries, grey literature, ongoing studies

Only downgrade for the above on studies contributing to the outcome.