Forest Plot

A forest plot is a graphical display used in meta analyses to present the results of individual studies and the overall pooled estimate in a single figure. Each study is represented by a horizontal line (the confidence interval) with a square at the point estimate, and the pooled result is shown as a diamond at the bottom.^[2] The name derives from the visual impression of a "forest" of lines.^[2]

The forest plot is the standard visual display for the results of a meta analysis. It allows readers to see at a glance the direction and magnitude of effect in each study, the precision of each estimate (reflected by the width of the confidence interval), the consistency of results across studies (whether the lines overlap), and the overall pooled result. A well constructed forest plot communicates more information in a single figure than pages of text. It makes the evidence transparent, allowing readers to assess the meta analysis critically rather than simply accepting the pooled number. Meta analyses are usually illustrated with forest plots, including in Cochrane reviews, and PRISMA 2020 asks reviews to present each study's effect estimate and its precision, ideally in structured tables or plots such as forest plots.^[3][4]

A forest plot is organized with study identifiers (typically first author and year) listed vertically on the left, and a horizontal scale showing the effect measure (risk ratio, odds ratio, mean difference, standardized mean difference, or hazard ratio) across the bottom or top.

For each study, a square is plotted at the point estimate, and a horizontal line extends from the lower to the upper bound of the confidence interval (usually 95%). The size of the square is proportional to the weight of the study in the meta analysis (typically the inverse of its variance), so larger, more precise studies have bigger squares.^[1][2] If the confidence interval line crosses the vertical line of no effect (1.0 for ratios, 0 for differences), the individual study result is not statistically significant at the chosen level.^[1]

At the bottom of the plot, the pooled estimate is displayed as a diamond. The center of the diamond represents the pooled point estimate, and the horizontal tips represent the confidence interval.^[2] If the diamond does not cross the line of no effect, the pooled result is statistically significant.^[1][2]

Forest plots may include additional information in columns alongside the figure, such as the exact point estimate and confidence interval for each study, the weight assigned to each study, the number of participants or events, and subgroup labels. Subgroup analyses are sometimes displayed in the same forest plot with subtotals for each subgroup.

Forest plots should be used whenever presenting the results of a meta analysis that combines effect estimates from multiple studies. They are appropriate for meta analyses of randomized controlled trials, observational studies, diagnostic test accuracy studies, and any other type of quantitative synthesis. Even when a formal pooled estimate is not calculated (for example, in a narrative synthesis with substantial heterogeneity), a forest plot can still be useful for visually displaying the individual study results. They are not appropriate for purely qualitative or non quantitative syntheses.

A hypothetical meta analysis of 12 randomized controlled trials evaluates the effect of probiotics on antibiotic associated diarrhea. The forest plot shows that 10 of 12 studies favor probiotics (risk ratio squares fall to the left of 1.0), though only 6 have confidence intervals that do not cross 1.0. Study weights range from 2.1% to 18.4%, with the two largest trials contributing the most to the pooled estimate. The pooled risk ratio diamond sits at 0.58 (95% CI: 0.50 to 0.68), indicating a 42% relative reduction in diarrhea risk. The diamond does not cross 1.0, and the I squared value printed below the plot is 38%, which falls where the Cochrane Handbook's overlapping ranges for "might not be important" (0% to 40%) and "may represent moderate heterogeneity" (30% to 60%) meet, so its importance depends on the context.