How do I fix an absolute effect calculation that used the wrong baseline risk?

Replace the baseline risk with a value drawn from the review's target population, then recompute the absolute effect and its confidence interval and update the Summary of Findings table.

Correct the baseline first

The baseline risk is the assumed event rate in the population the review addresses. Cochrane Handbook Chapter 15 treats applicability and indirectness as judgments about how well the included evidence matches that population, so a baseline taken from a different setting or risk group changes the absolute effect even when the relative effect is unchanged. Confirm the source of the baseline against the population, setting, and risk level the review targets before recomputing.

Recompute in order

1. Identify the baseline risk that matches the review's target population and record where it came from. Expected result: a documented baseline with a stated source and population.
2. Recompute the absolute effect by applying the relative effect to the corrected baseline. Expected result: a revised absolute effect for each outcome.
3. Recompute the confidence interval around the absolute effect using the corrected baseline. Expected result: an interval that reflects the new baseline rather than the old one.
4. Update the Summary of Findings table with the corrected absolute effect and interval, following GRADE Summary of Findings guidance. Expected result: a table consistent with the recomputed values.
5. Document the correction: the old baseline, the new baseline, its source, and the recomputed values. Expected result: an audit trail a reader or reviewer can follow.

Check the downstream effects

Revisit any certainty rating or conclusion that referenced the old absolute effect, since GRADE certainty judgments and Summary of Findings tables depend on the presented effects. If the corrected baseline comes from a different population than the included studies, note the indirectness in the table so readers see the basis for the absolute effect.

When the baseline is uncertain

If no baseline clearly matches the target population, present the recomputed effect for the plausible baseline range and state which population each value represents. That keeps the absolute effect tied to a defined population instead of a single unexplained number.