Posters · Glossary

What is a forest plot?

A forest plot is a meta-analysis chart that shows each study's effect as a square with a horizontal line for its confidence interval, set against a vertical line of no effect. A diamond at the bottom shows the pooled estimate, its width giving the pooled interval, so agreement between studies is visible at a glance.

A forest plot packs an entire systematic review's results into one figure, which makes it the chart readers look for first on a review poster. It also carries a lot of detail that becomes unreadable if the plot is simply shrunk to fit a board.

Nuwan Madhusanka · Co-founder

5 min read · Published 13 September 2026

The parts of a forest plot
Part	What it shows	How to read it
Study label	Each included study, often with its sample size or events	One row per study, sometimes grouped into subgroups
Square	The study's point estimate; its area reflects the study's weight	Bigger squares carry more weight in the pooled result
Horizontal line	The study's confidence interval	Longer lines mean less precise estimates
Vertical line of no effect	1 for ratios, 0 for differences	An interval crossing it is compatible with no effect
Diamond	The pooled estimate; its width is the pooled interval	Where the review's overall answer sits
Heterogeneity statistics	Measures such as I squared and tau squared	How much studies disagree beyond chance
Axis	Effect scale, usually logarithmic for ratios	Labels at each end say which side favours which group

Reading one from top to bottom

Start with the axis labels, which say which side of the line of no effect favours the intervention. Then scan the squares: are they mostly on one side, and do their lines overlap each other? The Cochrane Handbook explains that the area of each block indicates the weight assigned to that study, and the horizontal lines depict the range of intervention effects compatible with the study's result. Finally read the diamond, which the handbook describes as the summary result. If the diamond sits clear of the vertical line, the pooled estimate excludes no effect; if it touches or crosses, it does not.

Heterogeneity is part of the picture

Studies that point in different directions, or whose intervals barely overlap, suggest that the true effect varies. The I squared statistic describes the percentage of variation across studies due to heterogeneity rather than chance. The Cochrane Handbook gives a rough guide with deliberately overlapping bands: 0 to 40 percent might not be important, 30 to 60 percent may represent moderate heterogeneity, 50 to 90 percent substantial, and 75 to 100 percent considerable. It stresses that the importance of the value depends on the size and direction of effects and the strength of the evidence, not only the number.

Fixed and random effects

The diamond depends on the model. A fixed effect analysis assumes all studies estimate the same underlying effect. A random effects analysis, in the Cochrane Handbook's words, assumes the different studies are estimating different, yet related, intervention effects, and it produces a wider interval when studies disagree. The two can give different diamonds from the same squares. A poster should name the model and give the heterogeneity statistic, because a narrow fixed effect diamond over visibly scattered squares overstates how certain the pooled answer is.

Fitting a forest plot on a board

A plot with twenty studies and a column of numbers becomes unreadable at a metre when reduced to poster width. Options that work: show only the primary outcome, list studies in a meaningful order such as by weight or year, drop the numeric columns and keep the labels, and move full plots for secondary outcomes behind a QR code. Keep the axis labels large, because they carry the direction. The PRISMA statement expects results of syntheses to be presented, and a clear summary with the full plot linked meets that expectation better than a tiny one. A common compromise is a simplified plot of the six to ten largest studies with the pooled diamond, captioned with the total number of studies and participants.

Common mistakes

Mislabelling which side favours the intervention flips the meaning of the whole figure. Plotting ratios on a linear axis makes intervals look lopsided and exaggerates effects above one. Drawing all squares the same size hides which studies drive the result. Pooling studies with very different outcomes into one diamond produces a figure that looks authoritative but answers no clear question. And omitting heterogeneity statistics leaves the reader to guess whether the diamond represents agreement. A final error is presenting a subgroup plot without saying so, which leads readers to treat subgroups of one trial as independent studies.

Where it shows up in the poster builder

There is no forest plot among the ten chart kinds a poster figure offers, and chart figures hold categories and values with no interval whiskers. A forest plot drawn in statistics software could only sit in an image figure, and the seven image figures, found on four layouts, cannot be filled, since they have no upload control and image generation skips them. In practice a forest plot on a board built here is either described in a results section or replaced by a horizontal bar chart of each study's effect, with the pooled estimate, its interval and I squared stated in a key number block or the caption, and the full plot linked by QR code on the 37 layouts that carry one.

Questions people ask

Why is it called a forest plot?

The usual explanation is that the rows of lines and squares resemble a forest of trees, and the name echoes the saying about not seeing the wood for the trees, since the plot shows the individual studies and the overall result together. Whatever its origin, the name is now standard in systematic review methods and reporting guidance.

Can a forest plot be used outside meta-analysis?

Yes. The same layout is used to show subgroup results within a single trial, or the effects of several variables from one regression model. In those cases there may be no pooled diamond. The caption should say what each row represents, because a reader who assumes each row is a separate study will misread a subgroup plot.

What does it mean if one study's square is far from the others?

It may be an outlier because of a different population, intervention intensity, outcome measure or risk of bias. Reviewers check whether removing it changes the pooled result, a sensitivity analysis, and look for an explanation. A single outlying small study usually has little weight; an outlying large study deserves careful discussion on the poster.

Should the x axis be logarithmic?

For ratios such as risk ratios and odds ratios, yes. On a log scale a halving and a doubling sit the same distance from one, and intervals look symmetrical. For differences, such as mean differences, use a linear scale centred on zero. Label the axis with the actual values, not the logarithms, so readers can read effects directly.

What is a funnel plot and how is it related?

A funnel plot graphs each study's effect against its precision to look for small study effects, one sign of possible publication bias. In the absence of bias the points form a symmetrical inverted funnel. It uses the same study estimates as the forest plot but answers a different question, and it is unreliable with fewer than about ten studies.

How many studies are needed for a forest plot?

A pooled forest plot needs at least two studies, but with only a handful the pooled interval and heterogeneity statistics are imprecise. Many reviews still show a forest plot without pooling when studies are too few or too different, simply to display each study's estimate side by side. Say in the caption whether a pooled result was calculated.

Make one with posters

The button opens the generator with this use case already described. Change the wording to match your own.

Create a poster with OneCraft

Sources

Written and checked by the OneCraft team. Last checked 13 September 2026.