Research methods guide
When a meta-analysis may not be appropriate
Pool studies when their questions and estimates can be meaningfully combined. Compare populations, interventions or exposures, outcome definitions, time points, designs, and available data.
A worked review includes clinic BP, mean home BP and self-reported control. Before pooling, check construct, estimand and sample independence.
How to report results without pooling
Define grouping, preserve estimates and intervals, and explain placement.
| Item | Worked example | Verification |
|---|---|---|
| Different constructs | Two studies report mmHg and another reports the proportion self-reporting improvement. | Tabulate objectively measured BP and self-report separately. |
| Different time points | One-month and twelve-month outcomes may capture different effects. | Group by protocol windows and display individual estimates and intervals. |
| Missing estimates or overlap | One study lacks variance and two reports share a cohort. | Query authors or report available quantities and link overlapping records. |
Apply the method to your materials
Use structured synthesis
Show study, contrast, direction, magnitude, precision and bias by group. Use a forest plot without a pooled diamond where appropriate.
Describe grouping and ordering
State outcome, time, population or design groupings and explain the actual comparability decisions.
Questions that arise in practice
Should we count significant P values instead?
Significance counts depend on sample size and precision. Show effect, uncertainty and applicability.
Verified sources
The official sources below were checked on 2026-11-12. Access conditions and update dates are rechecked before publication.
RigorMed workspace
Start with your research materials
Bring your research question, available materials, and the steps you want to check.
