Correct the axis and column labels to match the effect measure actually pooled, then verify the estimate scale and revise any interpretation text that used the wrong measure. The Cochrane Handbook's meta-analysis guidance treats effect measures and forest plots as linked choices, so a label mismatch is a reporting error to fix at the source, not a cosmetic edit.
- Compare the plot's axis label and column headers against the effect measure named in the analysis methods and the pooled estimate. If the axis says one measure while the columns or methods section name another, record which one the underlying data actually used. The expected result is a single measure confirmed from the analysis record.
- Correct the axis label and column headers to that confirmed measure, and check the estimate scale at the same time. A ratio measure and a difference measure sit on different scales, so confirm the null value and direction markers match the measure before republishing. The decision point: if the scale does not fit the measure, rebuild the plot rather than relabel it.
- Document the correction in the analysis log or version note, stating the original label, the corrected label, and the measure confirmed from the data. This record supports the audit trail for the decision or handoff.
- Re-read the interpretation text and replace any sentence that describes the result using the wrong measure. Cochrane's guidance on interpreting results emphasizes confidence intervals and careful wording of pooled results, so check that the revised text matches the corrected scale and the confidence interval shown.
If the underlying data and the methods section disagree about which measure was used, treat the label as unresolved and confirm the measure with whoever ran the analysis before editing the plot.