Build the row by extracting adverse-event counts and absolute risks from each study, then rate certainty with GRADE and flag sparse-event limits. The row should let a reader see the event counts behind each risk and the appraisal record behind each certainty rating.
- Define the safety outcome and its counting rule before extraction. Fix the event definition, the time window, and whether you count participants with at least one event or total events. Record this rule so every study is extracted the same way.
- Extract counts per arm from each study and enter them into your extraction record. Cochrane Handbook Chapter 10 covers effect measures and careful interpretation of pooled results; keep the raw counts visible so the row's risk cells trace back to source data.
- Calculate absolute risks per arm and the corresponding effect estimate. Report the risk in each group alongside the pooled estimate so readers can judge clinical meaning, not only relative change.
- Rate certainty for the safety outcome using GRADE. Cochrane GRADE resources describe the domains used to rate certainty in evidence syntheses; record the rating and the reason for any downgrade in the appraisal record.
- Note sparse-event limitations in the row. When events are few, state the counts and describe the result as imprecise; the certainty rating and the note should point to the same appraisal record.
- Cross-check study-level fields against the registration record. ClinicalTrials.gov data element definitions describe structured fields for outcomes and reported results; use them to confirm outcome definitions and reported counts where a registration record exists.
A completed row has cells for adverse-event counts, absolute risks, and a certainty rating, each traceable to an extraction or appraisal record.