Forest plot

A data frame will be used for the basic layout of the forest plot. Graphical parameters can be set using the forest_style function.

forest draws the table and the confidence intervals. The other parts of the plot are added with functions that take the plot as their first argument, so they can be chained with the pipe |>:

set_xaxis Limits, tick marks and scale of the x-axis, and vertical lines
set_labs Title, x-axis labels, footnote, arrow labels and legend labels
scale_sizes Point sizes scaled by study weights
set_style Graphical parameters

These functions build the plot again, so they must be used before the plot is edited with edit_plot, add_text, insert_text, add_border or add_grob. The plot stays a gtable at every step, and can be combined with other plots, e.g. with patchwork::wrap_elements.

Usage

forest(
  data,
  est,
  lower,
  upper,
  sizes = 0.4,
  ref_line = NULL,
  ci_column,
  is_summary = NULL,
  nudge_y = 0,
  fn_ci = makeci,
  fn_summary = make_summary,
  index_args = NULL,
  style = NULL,
  ...
)

Arguments

data: Data to be displayed in the forest plot
est: Point estimation. Can be a list for multiple columns and/or multiple groups. If the length of the list is larger than then length of ci_column, then the values reused for each column and considered as different groups.
lower: Lower bound of the confidence interval, same as est.
upper: Upper bound of the confidence interval, same as est.
sizes: Size of the point estimation box, can be a vector or a list. The value is a multiple of one line of text, so 1 draws a point as tall as the base_size of the theme. The same scale applies to the summary diamond. Values are used as they are, unless scale_sizes is used to read them as study weights; useful values are roughly between 0.2 and 1.5, and a warning is given when the plot is drawn if they are outside 0.1 to 2.
ref_line: X-axis coordinates of the reference line, the value of no effect. If NULL (default), it is 1 if the x-axis is on a log scale (see set_xaxis) and 0 otherwise. Provide an atomic vector if different reference line for each ci_column is desired.
ci_column: Column number of the data the CI will be displayed.
is_summary: A logical vector indicating if the value is a summary value, which will have a diamond shape for the estimate. With multiple groups the diamonds are stacked in the same cell and the summary rows are made taller to fit them, so a larger nudge_y may be wanted.
nudge_y: Vertical adjustment to nudge groups by, must be within 0 to 1. Defaults to 0; for grouped forest plots a value of 0 is bumped to 0.1 automatically so that group CIs do not overplot. Set explicitly to override.
fn_ci: Name of the function to draw confidence interval, default is makeci. You can specify your own drawing function to draw the confidence interval, but the function needs to accept arguments "est", "lower", "upper", "sizes", "xlim", "pch", "gp", "t_height", "nudge_y". Please refer to the makeci function for the details of these parameters.
fn_summary: Name of the function to draw summary confidence interval, default is make_summary. You can specify your own drawing function to draw the summary confidence interval, but the function needs to accept arguments "est", "lower", "upper", "sizes", "xlim", "gp". Please refer to the make_summary function for the details of these parameters.
index_args: A character vector, name of the arguments used for indexing the row and column. This should be the name of the arguments that is working the same way as est, lower and upper. Check out the examples in the make_boxplot.
style: Style of the forest plot created with forest_style. A theme created with the superseded forest_theme is also accepted. The style can also be set or changed later with set_style.
...: Other arguments passed on to the fn_ci and fn_summary, or named in index_args. An argument none of them takes gives an error, as it would not be used. The arguments of earlier versions are also accepted here, with a message the first time each of them is used in a session: use set_xaxis instead of xlim, ticks_at, ticks_digits, ticks_minor, x_trans and vert_line, set_labs instead of arrow_lab, xlab, title and footnote, and style instead of theme.

Value

A forest plot object, a gtable of class forestplot.

Examples

library(grid)
# Read provided sample example data
dt <- read.csv(system.file("extdata", "example_data.csv", package = "forestploter"))

# Keep needed columns
dt <- dt[,1:6]

# indent the subgroup if there is a number in the placebo column
dt$Subgroup <- ifelse(is.na(dt$Placebo),
                      dt$Subgroup,
                      paste0("   ", dt$Subgroup))

# NA to blank or NA will be transformed to carachter.
dt$Treatment <- ifelse(is.na(dt$Treatment), "", dt$Treatment)
dt$Placebo <- ifelse(is.na(dt$Placebo), "", dt$Placebo)
dt$se <- (log(dt$hi) - log(dt$est))/1.96

# Add blank column for the forest plot to display CI.
# Adjust the column width with space.
dt$` ` <- paste(rep(" ", 20), collapse = " ")

# Create confidence interval column to display
dt$`HR (95% CI)` <- ifelse(is.na(dt$se), "",
                             sprintf("%.2f (%.2f to %.2f)",
                                     dt$est, dt$low, dt$hi))

# Define a style
st <- forest_style(base_size = 10,
                   ref_line = gpar(col = "red"),
                   footnote = gpar(col = "#636363", fontface = "italic"))

# Draw the plot and add the axis and labels with a pipe
p <- forest(dt[,c(1:3, 8:9)],
            est = dt$est,
            lower = dt$low,
            upper = dt$hi,
            sizes = dt$se,
            ci_column = 4,
            ref_line = 1,
            style = st) |>
  set_xaxis(xlim = c(0, 4), ticks_at = c(0.5, 1, 2, 3)) |>
  set_labs(arrow = c("Placebo Better", "Treatment Better"),
           footnote = "This is the demo data. Please feel free to change\nanything you want.")

# Print plot
plot(p)

Usage

Arguments

Value

See also

Examples