Search code examples
rggplot2survival-analysis

How can I draw median lines on a cumulative incidence plot in R and also include risk table / cumulative events table?

Here's a repeatable example of the code I would like to execute:

library(survival)
library(survminer)

km.fit <- survfit(Surv(rfstime, status) ~ 1, data = gbsg)

ggsurvplot(km.fit, fun = function(x) 1-x, data = gbsg, censor = FALSE, conf.int = TRUE,
           risk.table = TRUE, cumevents = TRUE, ylab = "Cumulative Incidence", 
           surv.median.line = "hv")

This code doesn't work and gives me the following error message:

Error in match(x, table, nomatch = 0L) : 
  'match' requires vector arguments

What I want is what the following code produces, but with median lines added to it:

ggsurvplot(km.fit, fun = function(x) 1-x, data = gbsg, censor = FALSE, conf.int = TRUE, 
           risk.table = TRUE, cumevents = TRUE, ylab = "Cumulative Incidence")

I figured out that I am getting the error mentioned previously because surv.median.line doesn't work when a function is called. However, I do need to be plotting this as cumulative incidence, not as survival, so the function is necessary.

A recommended work-around was to try and draw the median lines manually in ggplot, which I can do once I have saved this ggplot object to a variable and called it in the following way:

myplot <- ggsurvplot(km.fit, fun = function(x) 1-x, data = gbsg, censor = FALSE, conf.int = TRUE,
           risk.table = TRUE, cumevents = TRUE, ylab = "Cumulative Incidence")

myplot$plot + 
  geom_segment(aes(x = 0, xend = 1810, y = 0.5, yend = 0.5), linetype = "dashed") +
  geom_segment(aes(x = 1810, xend = 1810, y = 0, yend = 0.5), linetype = "dashed")

However, NOW the problem is that I don't see the risk table and cumulative incidence table anymore; I just have the plot. I do need the tables and it presents a lot of difficulty if I have to print them out and manage them separately somehow.

Any advice? How do I get everything I want (cumulative incidence plot, median lines, and tables) all in one plot? Is this even possible to do, or am I out of luck?

Solution

  • The issue is that you manipulated the plot but did not assign it back to ggsurvplot object, i.e. do myplot$plot <- myplot$plot + .... Then print myplot to get a plot of the survival curves inducing the risk table:

    library(survival)
library(survminer)

km.fit <- survfit(Surv(rfstime, status) ~ 1, data = gbsg)

myplot <- ggsurvplot(km.fit,
  fun = function(x) 1 - x, data = gbsg, censor = FALSE, conf.int = TRUE,
  risk.table = TRUE, cumevents = TRUE, ylab = "Cumulative Incidence"
)

myplot$plot <- myplot$plot +
  geom_segment(aes(x = 0, xend = 1810, y = 0.5, yend = 0.5), linetype = "dashed") +
  geom_segment(aes(x = 1810, xend = 1810, y = 0, yend = 0.5), linetype = "dashed")

myplot