Build a Summary of Findings table outcome by outcome: record the pooled estimate, add the absolute effect, assign a GRADE certainty rating, and trace each value to its analysis record. GRADE guidance treats certainty as an outcome-level judgement, so each row needs its own rating rather than one rating for the whole review.
- List the review's outcomes and order them by importance to decision-making. For each outcome, confirm which analysis record supplies the pooled result, including the effect measure and the model used (fixed or random effects), as described in Cochrane Handbook Chapter 10. Expected result: one named analysis record per outcome row.
- Enter the pooled estimate from that record, then add the absolute effect for the same outcome. Keep the effect measure consistent between the two columns so the row reads as one coherent result. Expected result: relative and absolute values that match the same analysis.
- Assign a certainty rating for each outcome using the GRADE criteria, judging risk of bias, inconsistency, indirectness, imprecision, and publication bias at the outcome level. Expected result: a rating per row with the reasons for any downgrade noted.
- Write footnotes that explain each rating decision and any deviation from the pooled analysis, and link each footnote to the row it supports. Expected result: footnotes that a reader can match to specific rows.
- Build a row-level source map that ties every pooled estimate, absolute effect, and certainty rating to its analysis record and footnote. Expected result: a completed table in which each value can be traced back to its source.
GRADEpro workflows support assembling Summary of Findings tables, and the GRADE Working Group describes the criteria used for certainty judgements. Use those resources to keep rating language consistent across rows.