Skip to contents

ggplot2 support for prodlim objects

Source: R/ggprodlim.R
ggprodlim.Rd

ggplot2 support for Kaplan-Meier and Aalen-Johansen estimators

Usage

ggprodlim(
  x,
  type = x$type,
  cause,
  xlim,
  ylim,
  y_breaks,
  x_breaks,
  position_atrisk,
  atrisk_title = "Number at risk",
  atrisk_labels = NULL,
  atrisk_title_line = 0.14,
  atrisk_title_step = 0.06,
  atrisk_line_step = 0.04,
  atrisk_bottom_margin = NULL,
  facet_panel_spacing_y = NULL,
  conf_int = TRUE,
  conf_int_alpha = 0.2,
  percent = TRUE,
  timeconverter,
  color_var,
  facet_formula,
  palette = NULL,
  return_data = FALSE,
  ...
)

Arguments

x

An object obtained with prodlim.

type

Either "surv" or "risk". Defaults to x$type.

cause

For competing risk models: a cause (passed to as.data.table.prodlim), "all" for all causes, or "stacked" for a stacked cumulative incidence plot.

xlim, ylim

Limits for the axes.

x_breaks, y_breaks

Breaks for axes.#

position_atrisk

Numeric vector specifying x-positions where numbers at risk should be printed. If missing, a default grid is chosen.

atrisk_title

Title shown left of the at-risk table.

atrisk_labels

Optional replacement labels for strata/cause groups in the at-risk table (in plotting order).

atrisk_title_line

Numeric. Position of the at-risk title below the plotting region, as an offset from min(ylim) in units of diff(ylim).

atrisk_title_step

Numeric. Vertical distance from the title to the first at-risk row, in units of diff(ylim).

atrisk_line_step

Numeric. Vertical spacing between at-risk rows, in units of diff(ylim).

atrisk_bottom_margin

Numeric. Bottom plot margin in lines when the at-risk table is shown.

facet_panel_spacing_y

Optional numeric. Vertical spacing between facet rows in lines. If NULL and faceting is used together with an at-risk table, the default is 5 + n_colors. The same value is used as the default bottom plot margin below the lowest facet row.

conf_int

Logical. If TRUE and lower/upper are available, draw pointwise confidence intervals.

conf_int_alpha

Alpha for CI shading.

percent

Logical. If TRUE, y axis is shown in percent (0-100).

timeconverter

Optional character, see plot.prodlim for supported values (e.g., "years2months").

color_var

Covariate name must be included in colnames(x$X)) and in competing risk models the word "cause".

facet_formula

Formula for facet_wrap. May include covariate names colnames(x$X) and in competing risk models also the word "cause".

palette

Optional vector of colors. If provided, applied via scale_color_manual/scale_fill_manual. Default NULL (do not impose a palette).

return_data

Logical. If TRUE, return a list with plot and data.

...

Passed to as.data.table.prodlim; commonly newdata, times, and for competing risks cause.

Value

A ggplot object, or a list if return_data=TRUE.

Details

A ggplot2-based alternative to plot.prodlim that returns a ggplot object, enabling ggplot theming, facetting, and composition.

See also

plot.prodlim, as.data.table.prodlim

Examples

data(Melanoma,package = "riskRegression")
# marginal Kaplan-Maier
km <- prodlim(Hist(time,status != 0)~1,data = Melanoma)
ggprodlim(km)

# marginal Aalen-Johansen
aj <- prodlim(Hist(time,status)~1,data = Melanoma)
ggprodlim(aj)

ggprodlim(aj,facet_formula = ~cause)


ajsex <- prodlim(Hist(time,status)~sex,data = Melanoma)
ggprodlim(ajsex)
#> Warning: Removed 8 rows containing missing values or values outside the scale range
#> (`geom_step()`).

ggprodlim(ajsex,facet_formula = ~cause)
#> Warning: Removed 8 rows containing missing values or values outside the scale range
#> (`geom_step()`).

ggprodlim(ajsex,facet_formula = ~sex)
#> Warning: Removed 8 rows containing missing values or values outside the scale range
#> (`geom_step()`).

ggprodlim(ajsex,color_var = "sex")
#> Warning: Removed 8 rows containing missing values or values outside the scale range
#> (`geom_step()`).

ggprodlim(ajsex,color_var = "cause")
#> Warning: Removed 8 rows containing missing values or values outside the scale range
#> (`geom_step()`).


# stacking the causes
ggprodlim(ajsex,cause = "stacked")

# stacked with two covariates 
   ajepisex <- prodlim(Hist(time,status)~sex+epicel,data = Melanoma)
   ggprodlim(ajepisex,
             cause = "stacked",
             facet_formula = sex~epicel,
             atrisk_title_line = 0.2)