Forest Plot Maker
Enter each study's estimate and its lower and upper confidence limits and get a forest plot: a square on a line for every study, an optional summary diamond, the line of no effect, and the numbers printed beside the plot. Ratio measures are drawn on a log scale with no effect at 1; differences on a linear scale with no effect at 0. The plot shows your numbers as given — it does not pool anything. Download as SVG or PNG. Free, no sign-up, and your data stays in your browser.
Write Adams 2015, 0.72, 0.48, 1.08, 14.2 — the study's name, its estimate, the lower and upper limits of its confidence interval, and optionally its weight in percent. A first line of column names is ignored. Use semicolons between cells if your numbers use decimal commas.
The plot draws exactly the numbers you enter. It does not pool studies or re-estimate anything — enter a pooled result yourself for the summary row.
6 studies
Show the plot as a table
| Study | Estimate | Lower limit | Upper limit | Weight (%) |
|---|---|---|---|---|
| Adams 2015 | 0.72 | 0.48 | 1.08 | 14.20 |
| Baker 2016 | 0.65 | 0.41 | 1.03 | 11.80 |
| Chen 2018 | 0.88 | 0.70 | 1.11 | 27.50 |
| Diaz 2019 | 0.59 | 0.37 | 0.94 | 10.40 |
| Evans 2021 | 0.81 | 0.66 | 0.99 | 36.10 |
| Overall (summary) | 0.78 | 0.68 | 0.89 |
Title and figure style
Takes the title out of the image and sets the figure number in bold and the title in italics above it — the APA 7 layout.
How to Use the Forest Plot Maker
- 1. Enter the studies. One per line:
Study, estimate, lower, upperand, if you have it,weightin percent. A first line of column names is ignored. - 2. Choose the scale. Pick Ratio for odds, risk and hazard ratios (log scale, no effect at 1) or Difference for mean differences, standardised mean differences and correlations (linear scale, no effect at 0).
- 3. Add a summary row if you have one. Type your pooled result as the last line and keep Draw the last row as a summary diamond ticked. Untick it to draw every row as a study.
- 4. Name things. Set the name of the measure and the interval, the number of decimals, and the labels for each side of the line of no effect, such as Favours treatment and Favours control.
- 5. Download. Take the SVG or PNG. The table under the plot lists every number for readers who cannot see the image.
The Input Syntax
Rows and what they draw
Study, OR, Lower, Upper, Weight % Adams 2015, 0.72, 0.48, 1.08, 14.2 Chen 2018, 0.88, 0.70, 1.11, 27.5 Evans 2021, 0.81, 0.66, 0.99, 36.1 Overall, 0.78, 0.68, 0.89
Each study is a square at its odds ratio on a line from its lower to its upper limit; squares are larger for studies with more weight. Overall has no weight and is drawn as a diamond, as it is the last row. The values are invented for illustration.
A row is left out, with a reason, if the estimate is not between its limits, if a ratio is zero or below, or if a number is missing. Use semicolons between cells if your numbers use decimal commas.
Why Ratios Use a Log Scale
A ratio of 2 means the same size of effect as a ratio of 0.5, in opposite directions. On a linear axis, 2 would be a whole unit from the line of no effect and 0.5 only half a unit, so protective effects would look smaller than harmful ones. On a log axis the two sit the same distance either side of 1, and the confidence intervals, which are symmetric on the log scale, look symmetric too. Differences have no such asymmetry, so they use a linear axis with no effect at 0.
What the Plot Does Not Do
This tool draws what you give it. It does not pool the studies, weigh them, test for heterogeneity or calculate a confidence interval; the summary diamond and the weights are whatever you typed. To combine studies, use the Meta-Analysis Calculator and enter its result here. In a systematic review, PRISMA 2020 asks authors to present, for each study, an effect estimate with its precision, ideally in a table or plot — a forest plot does that.
- Say what the measure is. The column header and axis title should name it: Odds ratio (log scale), Standardised mean difference.
- Label the direction. Favours treatment and Favours control tell the reader which side is good.
- Report the interval level. State it in the header (95% CI) and the figure note.
Frequently Asked Questions
- What is a forest plot?
- A forest plot shows the results of several studies on one axis. Each study is a square at its effect estimate with a horizontal line for its confidence interval, a vertical line marks no effect, and a diamond at the bottom often shows the pooled result. It makes agreement and disagreement between studies visible at a glance.
- How do I read a forest plot?
- Look at where each square sits relative to the line of no effect, and at whether its line crosses that line. An interval that crosses the line is compatible with no effect. A diamond that does not cross it shows a pooled effect whose interval excludes no effect. Larger squares are studies with more weight.
- Why is the line of no effect at 1 for some plots and 0 for others?
- Ratios such as odds, risk and hazard ratios equal 1 when there is no difference between groups. Differences — a mean difference, a standardised mean difference, a correlation — equal 0 when there is none. Choose the scale that matches your measure and the line moves to the right place.
- Does this tool do the meta-analysis?
- No. It draws the estimates, limits and weights you enter and nothing else. Pool the studies elsewhere, for example with the Meta-Analysis Calculator, and type the pooled estimate and its interval as the last row to have it drawn as a diamond.
- What do the sizes of the squares mean?
- When you enter weights, a larger square means the study carries more weight in the pooled result. Without weights every square is the same size. The weights are drawn as you give them; they are not calculated here.
- Why was one of my rows left out?
- The tool skips a row it cannot draw and says why: a number is missing, the estimate lies outside its limits, or a ratio is zero or negative on the log scale. Fix the row, or switch to the difference scale if zero or negative values are possible.