TimelyScholarFree Tools › Mantel-Haenszel & Peto Odds Ratio
Pool 2×2 tables across studies or strata with the Mantel-Haenszel odds ratio or risk ratio, or the Peto odds ratio for rare events. Get the pooled estimate, 95% CI, heterogeneity and a forest plot.
FreeNo account neededYour data stays in your browser
| Study | Events (treatment) | No event (treatment) | Events (control) | No event (control) | |
|---|---|---|---|---|---|
| Mantel-Haenszel odds ratio | 0.62 (95% CI 0.57 to 0.68) |
|---|---|
| Test for overall effect | z = -11.54, p < 0.001 |
| Heterogeneity | Q = 163.94, df = 12, p < 0.001 |
| I² | 92.7%represents considerable heterogeneity (75-100%) |
Heterogeneity is substantial. A fixed-effect pooled estimate may be misleading; consider a random-effects model and explore the causes.
Mantel-Haenszel OR or RR, or Peto OR.
Events and non-events in each group, per study or stratum.
With the forest plot and heterogeneity statistics.
The Mantel-Haenszel method is a fixed-effect method for binary data. It gives more reliable results than inverse-variance weighting when studies are small or events are few, and it handles zero cells without continuity corrections.
Peto's method pools 'observed minus expected' event counts. It performs well when events are rare, group sizes are balanced and effects are not large, but can be biased otherwise.
The same Mantel-Haenszel calculation gives an odds ratio adjusted for a confounder when the 2×2 tables are strata of one study.
Fixed effect. If heterogeneity is substantial, a random-effects analysis may be more appropriate.
Studies with no events in either group give no information about the odds ratio and do not contribute to the pooled estimate.