Help for package ggsurvfit

Title:

Flexible Time-to-Event Figures

Version:

1.1.0

Description:

Ease the creation of time-to-event (i.e. survival) endpoint figures. The modular functions create figures ready for publication. Each of the functions that add to or modify the figure are written as proper 'ggplot2' geoms or stat methods, allowing the functions from this package to be combined with any function or customization from 'ggplot2' and other 'ggplot2' extension packages.

License:

MIT + file LICENSE

URL:

https://github.com/pharmaverse/ggsurvfit, https://www.danieldsjoberg.com/ggsurvfit/

BugReports:

https://github.com/pharmaverse/ggsurvfit/issues

Depends:

R (≥ 3.5)

Imports:

broom (≥ 1.0.0), cli (≥ 3.0.0), dplyr (≥ 1.0.3), ggplot2 (≥ 3.5.0), glue (≥ 1.6.0), gtable, patchwork (≥ 1.1.0), rlang (≥ 1.0.0), survival (≥ 3.6-4), tidyr (≥ 1.0.0)

Suggests:

covr, knitr, rmarkdown, scales (≥ 1.1.0), spelling, testthat (≥ 3.2.0), tidycmprsk (≥ 1.0.0), vdiffr (≥ 1.0.0), withr

VignetteBuilder:

knitr

Config/Needs/website:

cowplot, ggeasy, gghighlight

Config/testthat/edition:

Encoding:

UTF-8

Language:

en-US

LazyData:

true

RoxygenNote:

7.3.1

NeedsCompilation:

Packaged:

2024-05-08 20:28:46 UTC; sjobergd

Author:

Daniel D. Sjoberg

[aut, cre, cph], Mark Baillie [aut], Charlotta Fruechtenicht

[aut], Steven Haesendonckx

[aut], Tim Treis

[aut]

Maintainer:

Daniel D. Sjoberg <danield.sjoberg@gmail.com>

Repository:

CRAN

Date/Publication:

2024-05-08 20:50:02 UTC

ggsurvfit: Flexible Time-to-Event Figures

Description

Author(s)

Maintainer: Daniel D. Sjoberg danield.sjoberg@gmail.com (ORCID) [copyright holder]

Authors:

Mark Baillie bailliem@gmail.com
Charlotta Fruechtenicht charlotta.fruechtenicht@roche.com (ORCID)
Steven Haesendonckx shaesen2@its.jnj.com (ORCID)
Tim Treis tim.treis@outlook.de (ORCID)

Create a Survival Outcome from CDISC Data

Description

The aim of Surv_CNSR() is to map the inconsistency in data convention between the survival package and CDISC ADaM ADTTE data model.

The function creates a survival object (e.g. survival::Surv()) that uses CDISC ADaM ADTTE coding conventions and converts the arguments to the status/event variable convention used in the survival package.

The AVAL and CNSR arguments are passed to survival::Surv(time = AVAL, event = 1 - CNSR, type = "right", origin = 0).

Usage

Surv_CNSR(AVAL, CNSR)

Arguments

AVAL

The follow-up time. The follow-up time is assumed to originate from zero. When no argument is passed, the default value is a column/vector named AVAL.

CNSR

The censoring indicator where 1=censored and 0=death/event. When no argument is passed, the default value is a column/vector named CNSR.

Value

Object of class 'Surv'

Details

The Surv_CNSR() function creates a survival object utilizing the expected data structure in the CDISC ADaM ADTTE data model, mapping the CDISC ADaM ADTTE coding conventions with the expected status/event variable convention used in the survival package—specifically, the coding convention used for the status/event indicator. The survival package expects the status/event indicator in the following format: 0=alive, 1=dead. Other accepted choices are TRUE/FALSE (TRUE = death) or 1/2 (2=death). A final but risky option is to omit the indicator variable, in which case all subjects are assumed to have an event.

The CDISC ADaM ADTTE data model adopts a different coding convention for the event/status indicator. Using this convention, the event/status variable is named 'CNSR' and uses the following coding: censor = 1, status/event = 0.

Examples

# Use the `Surv_CNSR()` function with ggsurvfit functions
survfit2(formula = Surv_CNSR() ~ STR01, data = adtte) %>%
  ggsurvfit() +
  add_confidence_interval()

# Use the `Surv_CNSR()` function with functions from other packages as well
survival::survfit(Surv_CNSR() ~ STR01, data = adtte)
survival::survreg(Surv_CNSR() ~ STR01 + AGE, data = adtte) %>%
  broom::tidy()

Add Censor Marking

Description

Add a marking on the figure to represent the time an observations was censored.

Usage

add_censor_mark(...)

Arguments

...

arguments passed to passed to ggplot2::geom_point(...) with defaults shape = 3 and size = 2

Value

a ggplot2 figure

Examples

survfit2(Surv(time, status) ~ 1, data = df_lung) %>%
  ggsurvfit() +
  add_confidence_interval() +
  add_censor_mark() +
  scale_ggsurvfit()

Add Confidence Interval

Description

Add a confidence interval represented by either a ribbon or lines.

Usage

add_confidence_interval(type = c("ribbon", "lines"), ...)

Arguments

type

string indicating the type of confidence interval to draw. Must be one of c("ribbon", "lines")

...

arguments pass to geom.

type = 'ribbon': Defaults are ggplot2::geom_ribbon(alpha = 0.2, color = NA, ...)
type = 'lines': Defaults are ggplot2::geom_step(linetype = "dashed", na.rm = TRUE, ...)

Value

a ggplot2 figure

Examples

survfit2(Surv(time, status) ~ sex, data = df_lung) %>%
  ggsurvfit() +
  add_confidence_interval() +
  scale_ggsurvfit()

survfit2(Surv(time, status) ~ 1, data = df_lung) %>%
  ggsurvfit() +
  add_confidence_interval(type = "lines") +
  scale_ggsurvfit()

Add Legend Title

Description

Add a default or custom title to the figure legend.

Usage

add_legend_title(title = NULL)

Arguments

title

a string to override the default legend title. Default is NULL

Value

a ggplot2 figure

Examples

survfit2(Surv(time, status) ~ surg, data = df_colon) %>%
  ggsurvfit() +
  add_legend_title() +
  scale_ggsurvfit()

Add p-value

Description

add_pvalue("caption"): Add a p-value to the figure via ggplot2::labs(caption=)
add_pvalue("annotation"): Add a p-value text annotation via ggplot2::annotation("text")

P-values are calculated with survival::survdiff() or tidycmprsk::glance(). Examples of custom placement located in the help file for survfit_p().

When a competing risks figure includes multiple outcomes, only the p-value comparing stratum for the first outcome can be placed.

Usage

add_pvalue(
  location = c("caption", "annotation"),
  caption = "{p.value}",
  prepend_p = TRUE,
  pvalue_fun = format_p,
  rho = 0,
  ...
)

Arguments

location

string indicating where to place p-value. Must be one of c("caption", "annotation")

caption

string to be placed as the caption/annotation. String will be processed with glue::glue(), and the default is "{p.value}"

prepend_p

prepend "p=" to formatted p-value

pvalue_fun

function to round and style p-value with

rho

argument passed to survival::survdiff(rho=)

...

arguments passed to ggplot2::annotate(). Commonly used arguments are ⁠x=⁠ and ⁠y=⁠ to place the p-value at the specified coordinates on the plot.

Value

a ggplot2 figure

Examples

survfit2(Surv(time, status) ~ surg, df_colon) %>%
  ggsurvfit() +
  add_pvalue(caption = "Log-rank {p.value}") +
  scale_ggsurvfit()

survfit2(Surv(time, status) ~ surg, df_colon) %>%
  ggsurvfit() +
  add_pvalue("annotation", size = 5) +
  scale_ggsurvfit()

Add Quantile Annotation

Description

Add quantile information annotated on to the plot.

Usage

add_quantile(y_value = NULL, x_value = NULL, ...)

Arguments

y_value, x_value

Numeric value where the line segment will be drawn. Default is y_value=0.5 when both y_value and x_value are unassigned.

...

Named arguments passed to ggplot2::geom_segment() with default linetype = 2

Value

a ggplot2 figure

Examples

survfit2(Surv(time, status) ~ sex, data = df_lung) %>%
  ggsurvfit() +
  add_quantile(linetype = 2) +
  scale_ggsurvfit()

survfit2(Surv(time, status) ~ 1, data = df_lung) %>%
  ggsurvfit() +
  add_quantile(linetype = 2) +
  add_quantile(y_value = 0.9, linetype = 3) +
  scale_ggsurvfit()

survfit2(Surv(time, status) ~ sex, data = df_lung) %>%
  ggsurvfit() +
  add_quantile(linetype = 2, y_value = NULL, x_value = 10) +
  scale_ggsurvfit()

Add risk table

Description

Add risk tables below the plot showing the number at risk, events observed, and number of censored observations.

Usage

add_risktable(
  times = NULL,
  risktable_stats = c("n.risk", "cum.event"),
  risktable_group = c("auto", "strata", "risktable_stats"),
  risktable_height = NULL,
  stats_label = NULL,
  combine_groups = FALSE,
  theme = theme_risktable_default(),
  size = 3.5,
  ...
)

Arguments

times

numeric vector of times where risk table values will be placed. Default are the times shown on the x-axis. The times passed here will not modify the tick marks shown on the figure. To modify which tick marks are shown, use ggplot2::scale_x_continuous(breaks=).

risktable_stats

character vector of statistics to show in the risk table. Must be one or more of c("n.risk", "cum.event", "cum.censor", "n.event", "n.censor"). Default is c("n.risk", "cum.event").

"n.risk" Number of patients at risk
"cum.event" Cumulative number of observed events
"cum.censor" Cumulative number of censored observations
"n.event" Number of events in time interval
"n.censor" Number of censored observations in time interval

See additional details below.

risktable_group

String indicating the grouping variable for the risk tables. Default is "auto" and will select "strata" or "risktable_stats" based on context.

"strata" groups the risk tables per stratum when present.
"risktable_stats" groups the risk tables per risktable_stats.

risktable_height

A numeric value between 0 and 1 indicates the proportion of the final plot the risk table will occupy.

stats_label

named vector or list of custom labels. Names are the statistics from ⁠risktable_stats=⁠ and the value is the custom label.

combine_groups

logical indicating whether to combine the statistics in the risk table across groups. Default is FALSE

theme

A risk table theme. Default is theme_risktable_default()

size, ...

arguments passed to ggplot2::geom_text(...). Pass arguments like, size = 4 to increase the size of the statistics presented in the table.

Value

a ggplot2 figure

Customize Statistics

You can customize how the statistics in the risk table are displayed by utilizing glue-like syntax in the risktable_stats argument.

For example, if you prefer to have the number at risk and the number of events on the same row, you can use risktable_stats = "{n.risk} ({cum.event})".

You can further customize the table to include the risk estimates using elements c("estimate", "conf.low", "conf.high", "std.error"). When using these elements, you'll likely need to include a function to round the estimates and multiply them by 100.

add_risktable(
  risktable_stats =
    c("{n.risk} ({cum.event})",
      "{round(estimate*100)}% ({round(conf.low*100)}, {round(conf.high*100)})"),
  stats_label = c("At Risk (Cum. Events)", "Survival (95% CI)")
)

Formatting Numbers

You can also pass glue-like syntax to risktable_stats to format the numbers displayed in the risk table. This is particularly helpful when working with weighted survfit2 objects for which the risk table may display too many decimals by default e.g., for weighted patients at risk.

add_risktable(
  risktable_stats = c("{format(round(n.risk, 2), nsmall = 2)}",
                      "{format(round(n.event, 2), nsmall = 2)}"),
  stats_label = c("N effective patients at risk",
                  "N effective events")
)

Competing Risks

The ggcuminc() can plot multiple competing events. The "cum.event" and "n.event" statistics are the sum of all events across outcomes shown on the plot.

Examples

p <-
  survfit2(Surv(time, status) ~ sex, data = df_lung) %>%
  ggsurvfit() +
  add_censor_mark() +
  add_confidence_interval() +
  scale_ggsurvfit()

# using the function defaults
p + add_risktable()

# change the statistics shown and the label
p +
  add_risktable(
    risktable_stats = "n.risk",
    stats_label = list(n.risk = "Number at Risk"),
  )

p +
  add_risktable(
    risktable_stats = "{n.risk} ({cum.event})"
  )

p +
  add_risktable(
    risktable_stats = c("n.risk", "cum.event"),
    combine_groups = TRUE
  )

Use Symbol for Strata in Risk Table

Description

Replace the stratum level names with a color symbol in the risk tables. Use this function when stratum level names are long.

Usage

add_risktable_strata_symbol(
  symbol = NULL,
  size = 15,
  face = "bold",
  vjust = 0.3,
  ...
)

Arguments

symbol

UTF-8 code of shape to replace strata level with. Default is a rectangle ("\U25AC"). Other common options are circle ("\U25CF") and diamond ("\U25C6"). While a symbol is the most common string to pass here, any string is acceptable.

size, face, vjust, ...

arguments passed to a function similar to ggtext::element_markdown()

Value

a ggplot2 figure

Examples


p <-
  survfit2(Surv(time, status) ~ sex, data = df_lung) %>%
  ggsurvfit(linewidth = 1) +
  add_confidence_interval() +
  add_risktable(risktable_group = "risktable_stats") +
  scale_ggsurvfit()

 p + add_risktable_strata_symbol()
 p + add_risktable_strata_symbol(symbol = "\U25CF", size = 10)

Example phase III clinical trial data set

Description

Background The example simulated data set is based on large phase III clinical trials in breast cancer such as the ALTTO trial ⁠https://ascopubs.org/doi/abs/10.1200/JCO.2015.62.1797⁠. The example “trial” aims to determine if a combination of two therapies tablemab (T) plus vismab (V) improves outcomes for metastatic human epidermal growth factor 2–positive breast cancer and increases the pathologic complete response in the neoadjuvant setting (i.e. treatment given as a first step to shrink a tumor before the main treatment or surgery).

Usage

adtte

Format

The data set contains the following variables:

STUDYID: The study identifier. A code unique to the clinical trial
SUBJID: subject identifier. Numeric ID unique to each patient
USUBJID: unique subject identifier. Text ID combining study and patient IDs
AGE: age at randomisation (years)
STR01: Hormone receptor status at randomisation
STR01N: Hormone receptor positive (Numeric)
STR01L: Hormone receptor positive (Long format)
STR02: Prior Radiotherapy at randomisation
STR02N: Prior Radiotherapy at randomisation (Numeric)
STR02L: Prior Radiotherapy at randomisation (Long format)
TRT01P: Planned treatment assigned at randomisation
TRT01PN: Planned treatment assigned at randomisation (Numeric)
PARAM: Analysis parameter: Progression free survival
PARAMCD: Analysis parameter code
AVAL: Analysis value (time to event (years)
CNSR: Censoring (0 = Event, 1 = Censored)
EVNTDESC: Event description
CNSDTDSC: Censoring description
DCTREAS: Discontinuation from study reason

Details

The trial has four treatment arms, patients with centrally confirmed human epidermal growth factor 2-positive early breast cancer were randomly assigned to 1 year of adjuvant therapy with V, T, their sequence (T to V), or their combination (T+V) for 52 weeks.

The primary end point was progression-free survival (PFS) as defined by Cancer.gov: '“the length of time during and after the treatment of a disease, such as cancer, that a patient lives with the disease but it does not get worse. In a clinical trial, measuring the progression-free survival is one way to see how well a new treatment works”'.

A number of baseline measurements (taken at randomization) are also included such as age, hormone receptor status and prior radiotherapy treatment.

Additional details on reasons for study discontinuation and censoring event description are also included.

The data set adopts an abridged version of the CDISC ADaM ADTTE time to event data model. See here for more info on CDISC ADaM data standards https://www.cdisc.org/standards/foundational/adam and specifically the ADTTE time to event data model here https://www.cdisc.org/standards/foundational/adam/adam-basic-data-structure-bds-time-event-tte-analyses-v1-0.

Source

https://github.com/VIS-SIG/Wonderful-Wednesdays/tree/master/data/2020/2020-04-08

Formatted Copy of `survival::colon`

Description

This is a copy of the colon data set exported by the survival package. This data set, however, has column labels, numeric categorical variables are now factors with assigned levels, and we only include the recurrence outcome.

Usage

df_colon

Format

An object of class tbl_df (inherits from tbl, data.frame) with 929 rows and 14 columns.

Formatted Copy of `survival::lung`

Description

This is a copy of the lung data set exported by the survival package. This data set, however, has column labels and numeric categorical variables are now factors with assigned levels.

Usage

df_lung

Format

An object of class tbl_df (inherits from tbl, data.frame) with 228 rows and 10 columns.

Format p-value

Description

Round and format p-values

Usage

format_p(x, digits = 1)

Arguments

x

numeric vector of p-values

digits

number of digits large p-values will be rounded to. Default is 2, and must be one of 1, 2, or 3.

Value

a string

Examples

p_vec <- c(0.00001, 0.01111, 0.0500000, 0.15, 0.99999)
format_p(p_vec)
format_p(p_vec, 2)
format_p(p_vec, 3)

Plot Cumulative Incidence

Description

Plot a cumulative incidence object created with tidycmprsk::cuminc() or a multi-state object created with survfit2(). Read more on multi-state models here.

Usage

ggcuminc(
  x,
  outcome = NULL,
  linetype_aes = FALSE,
  theme = theme_ggsurvfit_default(),
  ...
)

Arguments

x

a 'survfit' object created with survfit2()

outcome

string indicating which outcome(s) to include in plot. Default is to include the first competing event.

linetype_aes

logical indicating whether to add ggplot2::aes(linetype = strata) to the ggplot2::geom_step() call. When strata are present, the resulting figure will be a mix a various line types for each stratum.

theme

a survfit theme. Default is theme_ggsurvfit_default()

...

arguments passed to ggplot2::geom_step(...), e.g. size = 2

Value

a ggplot2 figure

Details

Why not use cmprsk::cuminc()?

The implementation of cmprsk::cuminc() does not provide the data required to construct the risk table. Moreover, the tidycmprsk::cuminc() has a user-friendly interface making it easy to learn and use.

Examples


library(tidycmprsk)

cuminc(Surv(ttdeath, death_cr) ~ trt, trial) %>%
  ggcuminc(outcome = "death from cancer") +
  add_confidence_interval() +
  add_risktable() +
  scale_ggsurvfit()

cuminc(Surv(ttdeath, death_cr) ~ trt, trial) %>%
  ggcuminc(outcome = c("death from cancer", "death other causes")) +
  add_risktable() +
  scale_ggsurvfit()

# using the survival multi-state model
survfit2(Surv(ttdeath, death_cr) ~ trt, trial) %>%
  ggcuminc(outcome = "death from cancer") +
  add_confidence_interval() +
  add_risktable() +
  scale_ggsurvfit()

Plot Survival Probability

Description

Plot survival probabilities (and other transformations) using the results from survfit2() or survival::survfit(); although, the former is recommend to have the best experience with the ggsurvfit package.

Usage

ggsurvfit(
  x,
  type = "survival",
  linetype_aes = FALSE,
  theme = theme_ggsurvfit_default(),
  ...
)

Arguments

x

a 'survfit' object created with survfit2()

type

type of statistic to report. Available for Kaplan-Meier estimates only. Default is "survival". Must be one of the following or a function:

type	transformation
`"survival"`	`x`
`"risk"`	`1 - x`
`"cumhaz"`	`-log(x)`
`"cloglog"`	`log(-log(x))`

linetype_aes

theme

a survfit theme. Default is theme_ggsurvfit_default()

...

arguments passed to ggplot2::geom_step(...), e.g. size = 2

Value

a ggplot2 figure

Details

This function creates a ggplot figure from the 'survfit' object. To better understand how to modify the figure, review the simplified code used internally:

survfit2(Surv(time, status) ~ sex, data = df_lung) %>%
  tidy_survfit() %>%
  ggplot(aes(x = time, y = estimate,
             min = conf.low, ymax = conf.low,
             color = strata, fill = strata)) +
  geom_step()

Examples

# Default publication ready plot
survfit2(Surv(time, status) ~ sex, data = df_lung) %>%
  ggsurvfit() +
  scale_ggsurvfit(x_scales = list(breaks = seq(0, 30, by = 6)))

# Changing statistic type
survfit2(Surv(time, status) ~ sex, data = df_lung) %>%
  ggsurvfit(type = "cumhaz")

# Configuring KM line type to vary by strata
survfit2(Surv(time, status) ~ sex, data = df_lung) %>%
  ggsurvfit(linetype_aes = TRUE) +
  scale_ggsurvfit()

# Customizing the plot to your needs
survfit2(Surv(time, status) ~ 1, data = df_lung) %>%
  ggsurvfit() +
  add_censor_mark() +
  add_confidence_interval() +
  add_quantile() +
  add_risktable() +
  scale_ggsurvfit()

Align Plots

Description

Function accepts a list of ggplot objects, and aligns each plot to the same widths as the first passed plot. This utility function is used to align the risktable plots with the risk curve plots.

Usage

ggsurvfit_align_plots(pltlist)

Arguments

pltlist

list of ggplots

Value

a list of ggplot grobs

Examples

# construct a base plot
gg <-
  survfit2(Surv(time, status) ~ 1, data = df_lung) %>%
  ggsurvfit() +
  add_confidence_interval() +
  scale_ggsurvfit()

# create an area plot representing the number of subjects who experienced
df_risktable <-
  survfit2(Surv(time, status) ~ 1, data = df_lung) %>%
  tidy_survfit()
# the event and those that have been censored.
gg_risktable_figure <-
  df_risktable %>%
  ggplot() +
  geom_ribbon(aes(x = time, ymin = 0, ymax = cum.event), fill = "black") +
  geom_ribbon(aes(x = time, ymin = n.risk[1], ymax = n.risk[1] - cum.censor), fill = "grey") +
  theme_void() +
  theme(axis.text.y = element_text(size=8)) +
  scale_y_continuous(
    breaks = c(0, max(df_risktable$n.risk)),
    labels = c("Cum. Events",  "Cum.Censored")
  )

# align plots
lst_aligned_plots <- ggsurvfit_align_plots(list(gg, gg_risktable_figure))

# combine plots with patchwork
patchwork::wrap_plots(
  lst_aligned_plots,
  ncol = 1,
  heights = c(0.9, 0.1)
)

Build 'ggsurvfit' Object

Description

Function takes an object created with ggsurvfit() or ggcuminc() and prepares the plot for printing. If a plot also has a risk table, this function will build the risk table plots and return them either as list of plots or combined using patchwork::wrap_plots().

This can be particularly useful when you would like to place figures with risk tables side-by-side.

Usage

ggsurvfit_build(x, combine_plots = TRUE)

Arguments

x

an object of class 'ggsurvfit' or 'ggcuminc'

combine_plots

logical indicating whether to combine the primary plot and the risk tables. When TRUE, plot and risk table(s) are combined with patchwork::wrap_plots(). When FALSE and the plot has risk tables, they are returned in a list of gtable grobs. Default is TRUE.

Value

a list of ggplot2 objects or a single ggplot2 object

Examples

# construct plot
p <-
  survfit2(Surv(time, status) ~ surg, df_colon) %>%
  ggsurvfit() +
  add_risktable() +
  scale_y_continuous(limits = c(0, 1))

# build plots
built_p <- ggsurvfit_build(p, combine_plots = FALSE)

# reconstruct original figure print with risktables
patchwork::wrap_plots(
  built_p[[1]],
  built_p[[2]],
  built_p[[3]],
  ncol = 1,
  heights = c(0.70, 0.15, 0.15)
)

# place plots side-by-side (plots must be built before placement with patchwork)
patchwork::wrap_plots(
  ggsurvfit_build(p),
  ggsurvfit_build(p),
  ncol = 2
)

Global Options

Description

By default, ggsurvfit() and ggcuminc() uses the color aesthetic to draw curves stratified by treatment group. Moreover, in ggcuminc() when multiple outcomes are plotted on the same figure the linetype aesthetic is used to distinguish the curves among the various outcomes.

It is, however, sometimes desirable to use the linetype to stratify by treatment group and color by outcome. To obtain these figures, set the options("ggsurvfit.switch-color-linetype" = TRUE) option.

Examples

options("ggsurvfit.switch-color-linetype" = TRUE)
library(tidycmprsk)

cuminc(Surv(ttdeath, death_cr) ~ trt, trial) %>%
  ggcuminc(outcome = "death from cancer") +
  add_risktable() +
  scale_ggsurvfit()

cuminc(Surv(ttdeath, death_cr) ~ 1, trial) %>%
  ggcuminc(outcome = c("death from cancer", "death other causes")) +
  add_risktable() +
  scale_ggsurvfit()

# reset option
options("ggsurvfit.switch-color-linetype" = NULL)

Draw ggsurvfit object

Description

grid::grid.draw() methods for objects of classes 'ggsurvfit' and 'ggcuminc'. These are implemented to allow users to directly call ggplot2::ggsave() on 'ggsurvfit' figures.

Usage

## S3 method for class 'ggsurvfit'
grid.draw(x, recording = TRUE)

## S3 method for class 'ggcuminc'
grid.draw(x, recording = TRUE)

Arguments

x

an object of class 'ggsurvfit' or 'ggcuminc'

recording

A logical value to indicate whether the drawing operation should be recorded on the Grid display list.

Value

None

Examples

survfit2(Surv(time, status) ~ surg, data = df_colon) %>%
  ggsurvfit() %>%
  grid.draw()

library(tidycmprsk)
cuminc(Surv(ttdeath, death_cr) ~ trt, trial) %>%
  ggcuminc() %>%
  grid.draw()

Print ggsurvfit object

Description

Print ggsurvfit object

Usage

## S3 method for class 'ggsurvfit'
print(x, ...)

## S3 method for class 'ggcuminc'
print(x, ...)

Arguments

x

an object of class 'ggsurvfit' or 'ggcuminc'

...

These dots are for future extensions and must be empty.

Value

a printed ggplot2 figure

Examples

print(survfit2(Surv(time, status) ~ surg, data = df_colon))

Objects exported from other packages

Description

These objects are imported from other packages. Follow the links below to see their documentation.

dplyr: %>%
ggplot2: %+%, ggplot_add
grid: grid.draw
survival: Surv, survfit

Apply Scales

Description

The most common figure created with this package is a survival curve. This scale applies modifications often seen in these figures.

scale_y_continuous(expand = c(0.025, 0), limits = c(0, 1), label = scales::label_percent()).
scale_x_continuous(expand = c(0.015, 0), n.breaks = 8)

If you use this function, you must include all scale specifications that would appear in scale_x_continuous() or scale_y_continuous(). For example, it's common you'll need to specify the x-axis break points. scale_ggsurvfit(x_scales=list(breaks=0:9)).

To reset any of the above settings to their ggplot2 default, set the value to NULL, e.g. y_scales = list(limits = NULL).

Usage

scale_ggsurvfit(x_scales = list(), y_scales = list())

Arguments

x_scales

a named list of arguments that will be passed to ggplot2::scale_x_continuous().

y_scales

a named list of arguments that will be passed to ggplot2::scale_y_continuous().

Value

a ggplot2 figure

Examples

ggsurvfit <-
  survfit2(Surv(time, status) ~ surg, data = df_colon) %>%
  ggsurvfit(linewidth = 1) +
  add_confidence_interval()

# use the function defaults
ggsurvfit + scale_ggsurvfit()

# specify additional scales
ggsurvfit +
  scale_ggsurvfit(x_scales = list(breaks = seq(0, 8, by = 2)))

Step ribbon statistic

Description

Provides stairstep values for ribbon plots

Usage

stat_stepribbon(
  mapping = NULL,
  data = NULL,
  geom = "ribbon",
  position = "identity",
  na.rm = FALSE,
  show.legend = NA,
  inherit.aes = TRUE,
  direction = "hv",
  ...
)

StatStepribbon

Arguments

mapping

Set of aesthetic mappings created by aes(). If specified and inherit.aes = TRUE (the default), it is combined with the default mapping at the top level of the plot. You must supply mapping if there is no plot mapping.

data

The data to be displayed in this layer. There are three options:

If NULL, the default, the data is inherited from the plot data as specified in the call to ggplot().

A data.frame, or other object, will override the plot data. All objects will be fortified to produce a data frame. See fortify() for which variables will be created.

A function will be called with a single argument, the plot data. The return value must be a data.frame, and will be used as the layer data. A function can be created from a formula (e.g. ~ head(.x, 10)).

geom

which geom to use; defaults to "ribbon"

position

Position adjustment, either as a string naming the adjustment (e.g. "jitter" to use position_jitter), or the result of a call to a position adjustment function. Use the latter if you need to change the settings of the adjustment.

na.rm

If FALSE, the default, missing values are removed with a warning. If TRUE, missing values are silently removed.

show.legend

logical. Should this layer be included in the legends? NA, the default, includes if any aesthetics are mapped. FALSE never includes, and TRUE always includes. It can also be a named logical vector to finely select the aesthetics to display.

inherit.aes

If FALSE, overrides the default aesthetics, rather than combining with them. This is most useful for helper functions that define both data and aesthetics and shouldn't inherit behaviour from the default plot specification, e.g. borders().

direction

hv for horizontal-vertical steps, vh for vertical-horizontal steps

...

Other arguments passed on to layer(). These are often aesthetics, used to set an aesthetic to a fixed value, like colour = "red" or size = 3. They may also be parameters to the paired geom/stat.

Format

An object of class StatStepRibbon (inherits from Stat, ggproto, gg) of length 3.

Value

a ggplot2 figure

References

https://groups.google.com/forum/?fromgroups=#!topic/ggplot2/9cFWHaH1CPs

Examples

survfit(Surv(time, status) ~ 1, data = df_lung) %>%
  survival::survfit0() %>%
  broom::tidy() %>%
  ggplot(aes(x = time, y = estimate, ymin = conf.low, ymax = conf.high)) +
  geom_step() +
  geom_ribbon(stat = "stepribbon", alpha = 0.2)

Create survival curves

Description

Simple wrapper for survival::survfit() except the environment is also included in the returned object.

Use this function with all other functions in this package to ensure all elements are calculable.

Usage

survfit2(formula, ...)

Arguments

formula

a formula object, which must have a Surv object as the response on the left of the ~ operator and, if desired, terms separated by + operators on the right. One of the terms may be a strata object. For a single survival curve the right hand side should be ~ 1.

...

Arguments passed on to survival::survfit.formula

data: a data frame in which to interpret the variables named in the formula, subset and weights arguments.
weights: The weights must be nonnegative and it is strongly recommended that they be strictly positive, since zero weights are ambiguous, compared to use of the subset argument.
subset: expression saying that only a subset of the rows of the data should be used in the fit.
na.action: a missing-data filter function, applied to the model frame, after any subset argument has been used. Default is options()$na.action.
stype: the method to be used estimation of the survival curve: 1 = direct, 2 = exp(cumulative hazard).
ctype: the method to be used for estimation of the cumulative hazard: 1 = Nelson-Aalen formula, 2 = Fleming-Harrington correction for tied events.
id: identifies individual subjects, when a given person can have multiple lines of data.
cluster: used to group observations for the infinitesimal jackknife variance estimate, defaults to the value of id.
robust: logical, should the function compute a robust variance. For multi-state survival curves or interval censored data this is true by default. For single state data see details, below.
istate: for multi-state models, identifies the initial state of each subject or observation. This also forces time0 =TRUE.
timefix: process times through the aeqSurv function to eliminate potential roundoff issues.
etype: a variable giving the type of event. This has been superseded by multi-state Surv objects and is deprecated; see example below.
model: include a copy of the model frame in the output
error: this argument is no longer used
entry: if TRUE, the output will contain n.enter which is the number of observations entering the risk set at any time; extra rows of output are created, if needed, for each unique entry time. Only applicable if there is an id statement.
time0: if TRUE, the output will include estimates at the starting point of the curve or ‘time 0’. See discussion below.

Value

survfit2 object

`survfit2()` vs `survfit()`

Both functions have identical inputs, so why do we need survfit2()?

The only difference between survfit2() and survival::survfit() is that the former tracks the environment from which the call to the function was made.

The definition of survfit2() is unremarkably simple:

survfit2 <- function(formula, ...) {
  # construct survfit object
  survfit <- survival::survfit(formula, ...)

  # add the environment
  survfit$.Environment = <calling environment>

  # add class and return
  class(survfit) <- c("survfit2", "survfit")
  survfit
}

The environment is needed to ensure the survfit call can be accurately reconstructed or parsed at any point post estimation. The call is parsed when p-values are reported and when labels are created. For example, the raw variable names appear in the output of a stratified survfit() result, e.g. "sex=Female". When using survfit2(), the originating data frame and formula may be parsed and the raw variable names removed.

Most functions in the package work with both survfit2() and survfit(); however, the output will be styled in a preferable format with survfit2().

Examples

# With `survfit()`
fit <- survfit(Surv(time, status) ~ sex, data = df_lung)
fit

# With `survfit2()`
fit2 <- survfit2(Surv(time, status) ~ sex, data = df_lung)
fit2

# Consistent behavior with other functions
summary(fit, times = c(10, 20))

summary(fit2, times = c(10, 20))

Calculate p-value

Description

The function survfit2_p() wraps survival::survdiff() and returns a formatted p-value.

Usage

survfit2_p(x, pvalue_fun = format_p, prepend_p = TRUE, rho = 0)

Arguments

x

a 'survfit2' object

pvalue_fun

function to round and style p-value with

prepend_p

prepend "p=" to formatted p-value

rho

argument passed to survival::survdiff(rho=)

Value

a string

Examples

sf <- survfit2(Surv(time, status) ~ sex, data = df_lung)

sf %>%
  ggsurvfit() +
  add_confidence_interval() +
  add_risktable() +
  scale_ggsurvfit() +
  labs(caption = glue::glue("Log-rank {survfit2_p(sf)}"))

sf %>%
  ggsurvfit() +
  add_confidence_interval() +
  add_risktable() +
  scale_ggsurvfit() +
  annotate("text", x = 2, y = 0.05, label = glue::glue("{survfit2_p(sf)}"))

Survfit Plot Themes

Description

Returns ggplot list of calls defining a theme.

theme_ggsurvfit_default(): Builds on theme_bw() with increased text sizes.
theme_ggsurvfit_KMunicate(): Theme to create KMunicate-styled figures. doi:10.1136/bmjopen-2019-030215

Usage

theme_ggsurvfit_default()

theme_ggsurvfit_KMunicate()

Value

a ggplot2 theme

Examples

survfit2(Surv(time, status) ~ sex, data = df_lung) %>%
  ggsurvfit(theme = theme_ggsurvfit_default()) +
  scale_ggsurvfit()

Risk Table Themes

Description

Returns ggplot list of calls defining a theme meant to be applied to a risk table.

Usage

theme_risktable_default(axis.text.y.size = 10, plot.title.size = 10.75)

theme_risktable_boxed(axis.text.y.size = 10, plot.title.size = 10.75)

Arguments

axis.text.y.size

text size of the labels on the left of the risk table

plot.title.size

text size of the risk table title

Value

a ggplot2 figure

Examples

p <- survfit2(Surv(time, status) ~ 1, data = df_lung) %>%
  ggsurvfit() +
  scale_ggsurvfit()

# default ------------------------------------
p + add_risktable(theme = theme_risktable_default())

# larger text --------------------------------
p +
  add_risktable(
    size = 4,
    theme = theme_risktable_default(axis.text.y.size = 12,
                                    plot.title.size = 14)
  )

# boxed --------------------------------------
p + add_risktable(theme = theme_risktable_boxed())

# none ---------------------------------------
p + add_risktable(theme = NULL, risktable_height = 0.20)

Tidy a cuminc object

Description

The tidycmprsk package exports a tidier for "cuminc" objects. This function adds on top of that and returns more information.

Usage

tidy_cuminc(x, times = NULL)

Arguments

x

a 'cuminc' object created with tidycmprsk::cuminc()

times

numeric vector of times. Default is NULL, which returns all observed times.

Value

a tibble

Examples

library(tidycmprsk)

cuminc(Surv(ttdeath, death_cr) ~ trt, trial) %>%
  tidy_cuminc()

Tidy a survfit object

Description

The broom package exports a tidier for "survfit" objects. This function adds on top of that and returns more information. The function also utilizes additional information stored when the survfit object is created with survfit2(). It's recommended to always use this function with survfit2().

Usage

tidy_survfit(
  x,
  times = NULL,
  type = c("survival", "risk", "cumhaz", "cloglog")
)

Arguments

x

a 'survfit' object created with survfit2()

times

numeric vector of times. Default is NULL, which returns all observed times.

type

type of statistic to report. Available for Kaplan-Meier estimates only. Default is "survival". Must be one of the following or a function:

type	transformation
`"survival"`	`x`
`"risk"`	`1 - x`
`"cumhaz"`	`-log(x)`
`"cloglog"`	`log(-log(x))`

Value

a tibble

Examples

survfit2(Surv(time, status) ~ factor(ph.ecog), data = df_lung) %>%
  tidy_survfit()

ggsurvfit: Flexible Time-to-Event Figures

Description

Author(s)

See Also

Create a Survival Outcome from CDISC Data

Description

Usage

Arguments

Value

Details

See Also

Examples

Add Censor Marking

Description

Usage

Arguments

Value

See Also

Examples

Add Confidence Interval

Description

Usage

Arguments

Value

See Also

Examples

Add Legend Title

Description

Usage

Arguments

Value

See Also

Examples

Add p-value

Description

Usage

Arguments

Value

See Also

Examples

Add Quantile Annotation

Description

Usage

Arguments

Value

See Also

Examples

Add risk table

Description

Usage

Arguments

Value

Customize Statistics

Formatting Numbers

Competing Risks

See Also

Examples

Use Symbol for Strata in Risk Table

Description

Usage

Arguments

Value

See Also

Examples

Example phase III clinical trial data set

Description

Usage

Format

Details

Source

Formatted Copy of survival::colon

Description

Usage

Format

Formatted Copy of survival::lung

Description

Usage

Format

Format p-value

Description

Formatted Copy of `survival::colon`

Formatted Copy of `survival::lung`