Forest Plot Generator

Figran

Paste your study results and get a forest plot and funnel plot for your paper.

  • Effect sizes, counts or means
  • Fixed and random effects
  • Free SVG and 300 DPI PNG
Data
Zoom
Color
StudyTreatmentControlEvents / NWeightRR (95% CI)Aronson 19484 / 12311 / 1395.1%0.41 (0.13–1.26)Ferguson & Simes 19496 / 30629 / 3036.4%0.20 (0.09–0.49)Rosenthal et al 19603 / 23111 / 2204.4%0.26 (0.07–0.92)Hart & Sutherland 197762 / 13598248 / 128679.7%0.24 (0.18–0.31)Frimodt-Moller et al 197333 / 506947 / 58088.9%0.80 (0.52–1.25)Stein & Aronson 1953180 / 1541372 / 145110.1%0.46 (0.39–0.54)Vandiviere et al 19738 / 254510 / 6296.0%0.20 (0.08–0.50)TPT Madras 1980505 / 88391499 / 8839110.2%1.01 (0.89–1.14)Coetzee & Berjak 196829 / 749945 / 72778.7%0.63 (0.39–1.00)Rosenthal et al 196117 / 171665 / 16658.4%0.25 (0.15–0.43)Comstock et al 1974186 / 50634141 / 273389.9%0.71 (0.57–0.89)Comstock & Webster 19695 / 24983 / 23413.8%1.56 (0.37–6.53)Comstock et al 197627 / 1691329 / 178548.4%0.98 (0.58–1.66)Random effects0.49 (0.34–0.70)0.050.10.20.512510Favors TreatmentFavors ControlHeterogeneity: τ² = 0.31; Q = 152.23, df = 12 (p < 0.001); I² = 92%Test for overall effect: Z = −3.97 (p < 0.001)
Options
0.10.20.51250.00.20.40.60.81.0RRStandard errorCenter and 95% pseudo-confidence limits: inverse variance
0.10.20.51250.00.20.40.60.81.0RRStandard errorCenter and 95% pseudo-confidence limits: inverse variance

Forest plot examples

StudyTreatmentControlEvents / NWeightRR (95% CI)Aronson 19484 / 12311 / 1395.1%0.41 (0.13–1.26)Ferguson & Simes 19496 / 30629 / 3036.4%0.20 (0.09–0.49)Rosenthal et al 19603 / 23111 / 2204.4%0.26 (0.07–0.92)Hart & Sutherland 197762 / 13598248 / 128679.7%0.24 (0.18–0.31)Frimodt-Moller et al 197333 / 506947 / 58088.9%0.80 (0.52–1.25)Stein & Aronson 1953180 / 1541372 / 145110.1%0.46 (0.39–0.54)Vandiviere et al 19738 / 254510 / 6296.0%0.20 (0.08–0.50)TPT Madras 1980505 / 88391499 / 8839110.2%1.01 (0.89–1.14)Coetzee & Berjak 196829 / 749945 / 72778.7%0.63 (0.39–1.00)Rosenthal et al 196117 / 171665 / 16658.4%0.25 (0.15–0.43)Comstock et al 1974186 / 50634141 / 273389.9%0.71 (0.57–0.89)Comstock & Webster 19695 / 24983 / 23413.8%1.56 (0.37–6.53)Comstock et al 197627 / 1691329 / 178548.4%0.98 (0.58–1.66)Random effects0.49 (0.34–0.70)0.050.10.20.512510Favors TreatmentFavors ControlHeterogeneity: τ² = 0.31; Q = 152.23, df = 12 (p < 0.001); I² = 92%Test for overall effect: Z = −3.97 (p < 0.001)

Risk ratio meta-analysis

StudyTreatmentControlEvents / NWeightRR (95% CI)randomAronson 19484 / 12311 / 1395.1%0.41 (0.13–1.26)Ferguson & Simes 19496 / 30629 / 3036.4%0.20 (0.09–0.49)Rosenthal et al 19603 / 23111 / 2204.4%0.26 (0.07–0.92)Hart & Sutherland 197762 / 13598248 / 128679.7%0.24 (0.18–0.31)Vandiviere et al 19738 / 254510 / 6296.0%0.20 (0.08–0.50)TPT Madras 1980505 / 88391499 / 8839110.2%1.01 (0.89–1.14)Coetzee & Berjak 196829 / 749945 / 72778.7%0.63 (0.39–1.00)random subtotal0.38 (0.22–0.65)alternateFrimodt-Moller et al 197333 / 506947 / 58088.9%0.80 (0.52–1.25)Stein & Aronson 1953180 / 1541372 / 145110.1%0.46 (0.39–0.54)alternate subtotal0.58 (0.34–1.01)systematicRosenthal et al 196117 / 171665 / 16658.4%0.25 (0.15–0.43)Comstock et al 1974186 / 50634141 / 273389.9%0.71 (0.57–0.89)Comstock & Webster 19695 / 24983 / 23413.8%1.56 (0.37–6.53)Comstock et al 197627 / 1691329 / 178548.4%0.98 (0.58–1.66)systematic subtotal0.65 (0.32–1.32)Random effects0.49 (0.34–0.70)0.050.10.20.512510Favors TreatmentFavors ControlHeterogeneity: τ² = 0.31; Q = 152.23, df = 12 (p < 0.001); I² = 92%Test for overall effect: Z = −3.97 (p < 0.001)Test for subgroup differences: Q = 1.86, df = 2 (p = 0.394)

Subgroup analysis

StudyTreatmentControlEvents / NWeight (fixed)Weight (random)OR (95% CI)Morton 19841 / 402 / 360.1%2.4%0.44 (0.04–5.02)Rasmussen 19869 / 13523 / 1351.0%9.7%0.35 (0.15–0.78)Smith 19862 / 2007 / 2000.3%4.7%0.28 (0.06–1.36)Abraham 19871 / 481 / 460.0%1.9%0.96 (0.06–15.77)Feldstedt 198810 / 1508 / 1480.3%8.4%1.25 (0.48–3.26)Shechter 19891 / 599 / 560.4%3.1%0.09 (0.01–0.74)Ceremuzynski 19891 / 253 / 230.1%2.6%0.28 (0.03–2.88)Bertschat 19890 / 221 / 210.0%1.5%0.30 (0.01–7.88)Singh 19906 / 7611 / 750.5%7.7%0.50 (0.17–1.43)Pereira 19901 / 277 / 270.3%2.9%0.11 (0.01–0.97)Shechter 19912 / 8912 / 800.6%5.0%0.13 (0.03–0.60)Golf 19915 / 2313 / 330.4%6.6%0.43 (0.13–1.44)Thogersen 19914 / 1308 / 1220.4%6.6%0.45 (0.13–1.54)LIMIT-2 199290 / 1159118 / 11575.0%14.3%0.74 (0.56–0.99)Shechter 19954 / 10717 / 1080.8%7.2%0.21 (0.07–0.64)ISIS-4 19952216 / 290112103 / 2903989.7%15.3%1.06 (1.00–1.13)Fixed effect1.01 (0.95–1.07)Random effects0.46 (0.31–0.70)0.010.050.21520Favors TreatmentFavors ControlFixed effect: Heterogeneity: τ² = 0.00; Q = 47.14, df = 15 (p < 0.001); I² = 68%Fixed effect: Test for overall effect: Z = 0.20 (p = 0.841)Random effects: Heterogeneity: τ² = 0.29; Q = 47.06, df = 15 (p < 0.001); I² = 74%Random effects: Test for overall effect: Z = −3.62 (p < 0.001)

Fixed and random effects

StudyTreatmentControlMean / SD / NWeightMD (95% CI)Edinburgh55/47/15575/64/15611.0%−20.00 (−32.47–−7.53)Orpington-Mild27/7/3129/4/3211.7%−2.00 (−4.83–0.83)Orpington-Moderate64/17/75119/29/7111.4%−55.00 (−62.77–−47.23)Orpington-Severe66/20/18137/48/189.6%−71.00 (−95.02–−46.98)Montreal-Home14/8/818/11/1311.4%−4.00 (−12.15–4.15)Montreal-Transfer19/7/5718/4/5211.7%1.00 (−1.12–3.12)Newcastle52/45/3441/34/3310.3%11.00 (−8.06–30.06)Umea21/16/11031/27/18311.6%−10.00 (−14.92–−5.08)Uppsala30/27/6023/20/5211.4%7.00 (−1.73–15.73)Random effects−15.11 (−32.64–2.43)−100−500Favors TreatmentFavors ControlHeterogeneity: τ² = 684.65; Q = 238.92, df = 8 (p < 0.001); I² = 99%Test for overall effect: Z = −1.69 (p = 0.091)

Mean difference

StudyOR (95% CI)AgeUnder 60 (Reference)60 or older1.45 (1.12–1.87)SexFemale (Reference)Male1.23 (0.94–1.61)12Favors TreatmentFavors Control

Odds ratios from a regression

Illustrative data

StudyEventsP for interactionHR (95% CI)Overall321/4020.79 (0.68–0.92)Age < 65145/1900.380.74 (0.58–0.95)Age ≥ 65176/2120.84 (0.67–1.05)Female118/1560.760.81 (0.63–1.04)Male203/2460.78 (0.64–0.95)1Favors TreatmentFavors Control

Trial subgroup hazard ratios

Illustrative data

0.0050.020.10.5210502000.00.51.01.52.0p < 0.010.01–0.050.05–0.10p > 0.10ORStandard errorCenter and 95% pseudo-confidence limits: inverse variance

Contour-enhanced funnel plot

0.010.050.215200.00.51.01.52.0Open circles: imputed studiesRRStandard errorCenter and 95% pseudo-confidence limits: inverse variance

Trim and fill

How to make a forest plot

  1. 1

    Choose a data type

    Select effect sizes, events or means.

  2. 2

    Import or edit

    Paste a table or select a study in the figure.

  3. 3

    Choose a model

    Select fixed, random or your own values.

  4. 4

    Download

    Save a forest plot or funnel plot.

How to read a forest plot

A forest plot, sometimes called a blobbogram, shows estimates from individual studies alongside a pooled result.

StudyTreatmentControlEvents / NWeightRR (95% CI)Aronson 19484 / 12311 / 1395.1%0.41 (0.13–1.26)Ferguson & Simes 19496 / 30629 / 3036.4%0.20 (0.09–0.49)Rosenthal et al 19603 / 23111 / 2204.4%0.26 (0.07–0.92)Hart & Sutherland 197762 / 13598248 / 128679.7%0.24 (0.18–0.31)Frimodt-Moller et al 197333 / 506947 / 58088.9%0.80 (0.52–1.25)Stein & Aronson 1953180 / 1541372 / 145110.1%0.46 (0.39–0.54)Vandiviere et al 19738 / 254510 / 6296.0%0.20 (0.08–0.50)TPT Madras 1980505 / 88391499 / 8839110.2%1.01 (0.89–1.14)Coetzee & Berjak 196829 / 749945 / 72778.7%0.63 (0.39–1.00)Rosenthal et al 196117 / 171665 / 16658.4%0.25 (0.15–0.43)Comstock et al 1974186 / 50634141 / 273389.9%0.71 (0.57–0.89)Comstock & Webster 19695 / 24983 / 23413.8%1.56 (0.37–6.53)Comstock et al 197627 / 1691329 / 178548.4%0.98 (0.58–1.66)Random effects0.49 (0.34–0.70)0.050.10.20.512510Favors TreatmentFavors ControlHeterogeneity: τ² = 0.31; Q = 152.23, df = 12 (p < 0.001); I² = 92%Test for overall effect: Z = −3.97 (p < 0.001)
  1. Square: the study estimate; its area reflects weight.
  2. Horizontal line: the confidence interval.
  3. Vertical line: the value indicating no effect.
  4. Diamond: the pooled estimate and confidence interval.
  5. I²: the proportion of variation associated with heterogeneity.

How to read a funnel plot

The vertical axis shows standard error, with more precise studies toward the top. The dashed triangle shows 95% pseudo-confidence limits around the inverse-variance fixed-effect estimate. Asymmetry can have several explanations.

Frequently asked questions

How do I make a forest plot?

Choose effect sizes with confidence intervals, event counts, or means and standard deviations. Edit the plot or import a table; an example is loaded to get you started.

What does I² mean?

I² estimates the share of observed variation attributable to heterogeneity. The Handbook describes 0–40% as potentially unimportant, 30–60% moderate, 50–90% substantial, and 75–100% considerable; interpretation depends on context. Its value can change with the τ² method.

Fixed effect or random effects?

Fixed effect assumes a common effect; random effects allow effects to vary across studies. Random effects with REML is the default; DerSimonian–Laird and HKSJ are available in Options.

Which method does this page use?

Random effects use inverse variance with REML or DerSimonian–Laird; fixed event counts use Mantel–Haenszel. Wald intervals are the default and HKSJ is optional. Checked against metafor 5.2.1. RevMan’s M-H random method uses different weights. SMD uses the exact Hedges correction and may differ slightly from RevMan.

Can I keep pooled results from RevMan, Stata or R?

Yes. Choose My own values and mark a row Summary to draw your supplied estimate and confidence interval.

How do I make a forest plot from Excel?

Copy the study table from Excel and paste it into Import. You can also paste results from R or Stata, or choose a CSV, TSV or XLSX file.

Does funnel plot asymmetry mean publication bias?

No. Asymmetry has several possible explanations. Sterne 2011 advises against asymmetry tests with fewer than 10 studies because power is usually low. Egger’s test can mislead with odds ratios; Harbord or Peters may be preferable.

Is my data uploaded?

No. Calculations and exports run in your browser. The draft remains when you refresh and is cleared when the browser tab is closed.

Example sources: Colditz 1994; Egger, Davey Smith and Altman 2001, Table 18.2; Normand 1999; Lee & Done 2004.

Make your forest plot

Turn your study results into a forest plot and funnel plot.