Title: Publication-Ready Forest Plots
Version: 0.3.0
Description: Creates publication-ready forest plots from any tabular data containing point estimates and confidence intervals. Suitable for visualising results from regression models, meta-analyses, subgroup analyses, or any comparative study. Supports group and subgroup headings, summary estimates displayed as diamonds, grouped estimates with automatic colour and shape mapping, vertical dodging of multiple estimates within the same row, customisable text columns alongside the plot, and optional row striping. Provides a helper to export plots to PDF, PNG, SVG, or TIFF. Built on 'tinyplot' for clean, consistent visual styling with a minimal dependency footprint.
License: MIT + file LICENSE
Encoding: UTF-8
Language: en-GB
RoxygenNote: 7.3.3
Depends: R (≥ 4.1.0)
Imports: tinyplot (≥ 0.2.0)
Suggests: broom, covr, data.table, knitr, marginaleffects, MatchIt, quarto, rmarkdown, tibble, testthat (≥ 3.0.0)
Config/testthat/edition: 3
Config/Needs/coverage: covr
VignetteBuilder: quarto, knitr
URL: https://lorenzofabbri.github.io/forrest/, https://github.com/lorenzoFabbri/forrest
BugReports: https://github.com/lorenzoFabbri/forrest/issues
NeedsCompilation: no
Packaged: 2026-03-27 11:47:54 UTC; lorenzo
Author: Lorenzo Fabbri [aut, cre, cph]
Maintainer: Lorenzo Fabbri <lorenzo.fabbri92sm@gmail.com>
Repository: CRAN
Date/Publication: 2026-03-31 15:10:02 UTC

forrest: Publication-Ready Forest Plots

Description

Creates publication-ready forest plots from any tabular data containing point estimates and confidence intervals. Suitable for visualising results from regression models, meta-analyses, subgroup analyses, or any comparative study. Supports group and subgroup headings, summary estimates displayed as diamonds, grouped estimates with automatic colour and shape mapping, vertical dodging of multiple estimates within the same row, customisable text columns alongside the plot, and optional row striping. Provides a helper to export plots to PDF, PNG, SVG, or TIFF. Built on 'tinyplot' for clean, consistent visual styling with a minimal dependency footprint.

Author(s)

Maintainer: Lorenzo Fabbri lorenzo.fabbri92sm@gmail.com [copyright holder]

See Also

Useful links:

Create a forest plot

Description

Draws a publication-ready forest plot from a data frame. Each row represents one estimate — a study, a predictor, a model, a subgroup, or any other unit of analysis.

Usage

forrest(
  data,
  estimate,
  lower,
  upper,
  label = NULL,
  group = NULL,
  is_summary = NULL,
  weight = NULL,
  section = NULL,
  subsection = NULL,
  section_indent = TRUE,
  section_spacer = TRUE,
  section_cols = NULL,
  subsection_cols = NULL,
  ref_label = FALSE,
  ref_line = 0,
  log_scale = FALSE,
  xlim = NULL,
  xlab = "Estimate (95% CI)",
  title = NULL,
  header = NULL,
  cols = NULL,
  widths = NULL,
  stripe = FALSE,
  dodge = FALSE,
  pch = 15,
  shape = NULL,
  lwd = 2,
  cex = 1,
  col = NULL,
  cols_by_group = FALSE,
  legend_pos = "topright",
  legend_shape_pos = "bottomright",
  theme = "default",
  ...
)

Arguments

data

A data frame, tibble, or data.table.

estimate

Column name (string) for point estimates.

lower

Column name (string) for lower confidence interval bounds.

upper

Column name (string) for upper confidence interval bounds.

label

Column name (string) for row labels displayed on the y-axis or in the left text panel. If NULL, row numbers are used.

group

Column name (string) for a grouping variable. Rows that share a group value receive the same colour, and a legend is drawn automatically.

is_summary

Column name (string) of a logical vector. Rows where this is TRUE are drawn as filled diamonds (e.g. pooled or overall estimates) rather than squares with whiskers.

weight

Column name (string) of numeric row weights. When provided, point size scales as cex * sqrt(weight / max(weight)), so rows with larger weights appear with a bigger marker. Weights for summary rows are ignored (diamond size is fixed by cex).

section

Column name (string) for a grouping variable that determines section structure. Whenever the value of this column changes (run-length boundary), a bold section header row is automatically inserted before the group. Row order is preserved; no automatic sorting is applied. See also section_indent, section_spacer, and section_cols.

subsection

Column name (string) for a second-level grouping variable. Requires section. Inserts indented sub-headers beneath each section header. See also subsection_cols.

section_indent

Logical. If TRUE (default), label values of data rows within a section are automatically indented by two spaces (four spaces for rows within a subsection).

section_spacer

Logical. If TRUE (default), a blank spacer row is appended after the last row of each section.

section_cols

Named character vector. Names must be a subset of the names of cols. Values are column names in data whose first non-NA entry in each section is shown in that section's header row. Columns not listed here display "" in the header row. Use this to show section-level summaries (e.g. "k = 3 studies") next to the section header.

subsection_cols

Like section_cols but for subsection header rows.

ref_label

Logical. When TRUE and section is provided, rows with NA estimates that are present in the original data (i.e. reference category rows, not auto-generated headers) have " (Ref.)" appended to their label. Default is FALSE.

ref_line

Numeric. Position of the vertical reference line (e.g. 0 for differences, 1 for ratio measures on the natural scale). Set to NULL to suppress. Default is 0.

log_scale

Logical. If TRUE, apply a log transformation to the x-axis. Useful when plotting odds ratios, hazard ratios, or risk ratios on the natural scale. Default is FALSE.

xlim

Numeric vector of length 2 giving x-axis limits. Computed from the data when NULL (default). Confidence intervals that extend beyond xlim are clipped at the axis boundary and an arrow is drawn to indicate truncation.

xlab

Label for the x-axis. Default is "Estimate (95% CI)".

title

Plot title. Default is NULL (no title).

header

Optional header string placed above the label column. When cols is provided this appears above the left text panel; otherwise it is drawn above the topmost row on the y-axis.

cols

Named character vector specifying extra text columns to display to the right of the plot. Names become column headers; values are column names in data. Example: cols = c("OR (95\\% CI)" = "or_ci").

widths

Numeric vector of relative panel widths. When cols is NULL, ignored. Otherwise, length must equal 2 + length(cols): label panel, plot panel, then one entry per extra column. Sensible defaults are chosen automatically.

stripe

Logical. If TRUE, alternate rows are shaded with a light grey background to improve readability. Default is FALSE.

dodge

Logical or positive numeric. When TRUE (or a positive number), consecutive rows that share the same label value are grouped together and their confidence intervals are drawn with a small vertical offset so that they do not overlap. The shared label is displayed once at the centre of the group. Use together with group (for colour) and/or shape (for point characters) to distinguish the overlaid series. A numeric value sets the offset between rows in a group directly (in y-axis units); TRUE uses a default of 0.25. Default is FALSE.

pch

Point character for non-summary rows. Default is 15 (filled square). When shape is provided, pch is used only as a fallback for rows whose shape value is NA.

shape

Column name (string) for a shape variable. When provided, different values of the column are rendered with different point characters and a shape legend is drawn. Use together with group to distinguish two categorical dimensions simultaneously (e.g. colour = time period, shape = sex).

lwd

Line width for confidence interval whiskers. Default is 2.

cex

Point size multiplier. Default is 1.

col

Colour or character vector of colours. When NULL (default) and group is specified, the Okabe-Ito colorblind-safe palette is used. When NULL and no group, a single dark colour is used.

cols_by_group

Logical. Relevant only when dodge is active. When TRUE, each text column in cols is collapsed to one value per label group: the first non-empty entry within the group is displayed at the group centre y position. This produces a wide-format text table with one row per label and one column per condition. Populate each text column so that the value is non-empty only for the matching condition row and empty ("") for all others; forrest() picks up the right value automatically. When FALSE (default), text values are drawn at each individual row's dodged y position, keeping them aligned with their CI whiskers.

legend_pos

Position of the colour legend when group is supplied. Passed to legend(). Use NULL to suppress. Default is "topright".

legend_shape_pos

Position of the shape legend when shape is supplied. Passed to legend(). Use NULL to suppress. Default is "bottomright".

theme

Visual theme name ("default", "minimal", "classic") or a named list of style overrides. Default is "default".

...

Graphical parameters forwarded to the internal tinyplot call (e.g. cex.axis, cex.lab).

Details

Rows with NA estimates are treated as reference-category rows: they produce no point or confidence interval, and their label is rendered in regular (non-bold) font. To create section headers and spacers automatically, use the section (and optionally subsection) arguments instead of inserting NA rows by hand.

Value

Invisibly returns NULL. Called for its side effect of producing a plot.

Examples

# Basic forest plot: linear model coefficients
dat <- data.frame(
  predictor = c("Age (per 10 y)", "Female sex",
                "BMI (per 5 kg/m\u00b2)", "Current smoker"),
  estimate  = c(0.42, -0.38,  0.19, -0.31),
  lower     = c(0.22, -0.56, -0.02, -0.51),
  upper     = c(0.62, -0.20,  0.40, -0.11)
)
forrest(dat,
  estimate = "estimate",
  lower    = "lower",
  upper    = "upper",
  label    = "predictor",
  xlab     = "Regression coefficient (95% CI)"
)

# Section headers from a grouping column
dat2 <- data.frame(
  domain    = c("Lifestyle", "Lifestyle", "Clinical", "Clinical"),
  predictor = c("Physical activity", "Diet quality",
                "BMI (per 5 kg/m\u00b2)", "Systolic BP (per 10 mmHg)"),
  estimate  = c(-0.31, -0.18,  0.19,  0.25),
  lower     = c(-0.51, -0.36, -0.02,  0.08),
  upper     = c(-0.11, -0.00,  0.40,  0.42)
)
forrest(dat2,
  estimate  = "estimate",
  lower     = "lower",
  upper     = "upper",
  label     = "predictor",
  section   = "domain",
  xlab      = "Regression coefficient (95% CI)"
)

Save a forest plot to a file

Description

Writes the plot produced by forrest() (or any base-R plotting code) to a file. The graphics device is inferred from the file extension.

Usage

save_forrest(file, plot, width = 7, height = 5, dpi = 300, bg = "white")

Arguments

file

Output file path. Supported extensions: .pdf, .png, .svg, .tiff.

plot

A zero-argument function whose body calls forrest(). Evaluated inside the open graphics device.

width

Plot width in inches. Default 7.

height

Plot height in inches. Default 5.

dpi

Resolution in dots per inch for raster formats (.png, .tiff). Ignored for vector formats (.pdf, .svg). Default 300.

bg

Background colour. Default "white".

Value

Invisibly returns file.

Examples

dat <- data.frame(
  label    = c("Age (per 10 y)", "Female sex", "Current smoker"),
  estimate = c(0.42, -0.38, -0.31),
  lower    = c(0.22, -0.56, -0.51),
  upper    = c(0.62, -0.20, -0.11)
)
tmp <- tempfile(fileext = ".pdf")
save_forrest(tmp, function() {
  forrest(
    dat,
    estimate = "estimate",
    lower    = "lower",
    upper    = "upper",
    label    = "label",
    xlab     = "Regression coefficient (95% CI)"
  )
})