How do I resolve a conflict between my protocol and my final synthesis method?

Compare the prespecified synthesis method with the one you used, document the reason and its effect on findings, then report the deviation in your review. PRISMA 2020 asks reviews to report the methods used and what studies were found and included, so a changed synthesis method belongs in that reporting.

1. Reconstruct both methods side by side. Write the protocol's synthesis plan in one column: the effect measure, the pooling model, and any planned subgroup or sensitivity analysis. In the second column, record what you actually did. The result is a specific list of differences, such as a switch from a fixed-effect to a random-effects model or a dropped subgroup analysis.

2. Trace each difference to its reason. For every row, record the trigger: unexpected heterogeneity, sparse events, missing data, or a correction to the original plan. Cochrane guidance on meta-analysis covers effect measures, fixed and random effects, heterogeneity, missing data, and sensitivity analyses, so name which of these conditions changed and when you noticed it.

3. Test the impact on the findings. Re-run the prespecified analysis where the data allow and compare the pooled estimate, its confidence interval, and the P value with the final result. Cochrane guidance on interpreting results treats confidence intervals, exact P values, and absolute and relative effects as the quantities that carry the conclusion. If the conclusion holds under both methods, say so; if it changes, the deviation is material and the final conclusion should rest on the method you can justify.

4. Decide between reporting and reverting. Revert to the protocol when the prespecified method remains feasible and the deviation was procedural. Report the deviation when the original method cannot be applied to the retrieved data or when it produces a misleading pooled result. Either way, state the change, the reason, and its effect on the findings in the methods and results.

5. Record the deviation where readers will find it. Place the comparison in the methods section, note the reason beside the affected analysis, and reflect the final method in the abstract and flow of included studies. Use the PRISMA 2020 checklist to confirm that the reported methods match the analyses you present.