What Is a Forest Plot? A Practical Meta-Analysis Guide
Everything you need to read and construct publication-ready meta-analysis forest plots using Fixed and DerSimonian-Laird Random Effects.
Meta-analysis forest plots with Fixed/Random effects (DL), weights, I², τ², and Q.
Decoding squares, whiskers, and weights
Each row in a forest plot summarizes one independent study. The square sits at the point estimate (such as an Odds Ratio or Mean Difference) and its area scales with the study's inverse-variance weight—larger, more precise studies get bigger squares and narrower 95% confidence interval whiskers.
The line of no effect and the pooled diamond
A vertical dashed line marks no treatment difference: 1.0 for ratio measures (OR, RR, HR) on a log axis, or 0.0 for continuous Mean Differences. The diamond at the bottom shows the combined meta-analytic estimate, with its horizontal tips marking the pooled 95% CI. When the diamond stays entirely to one side of the null line, the overall effect reaches statistical significance.
Reporting I², τ², and Fixed vs. Random models
Cochran's Q tests whether study effects vary more than chance alone, while I² expresses the percentage of total variance attributable to true between-study heterogeneity (roughly 25% low, 50% moderate, 75% high). When clinical diversity is expected, report the DerSimonian-Laird Random-Effects model alongside τ².