GGsurvplot shows wrong percentages in risktable I used ggsurvplot to draw failure(with fun = "event" argument) probabilites at different time points. The % numbers in the Risktable are not correct. Do you know why? Is it possible to change that manually that is change the column  pct.risk
in the data.survtable dataframe ?
  f0 <- survfit(Surv(data$period1, event) ~ 1, 
             weights=frequency, conf.type="plain", 
             conf.int=0.9, data = data)
        
 t <- ggsurvplot(f0, data = data, ylab = " ", risk.table = 
                   "abs_pct" ,xlim = c(0,135),
                   break.x.by = 27, fun = "event" )   


 A: Both the frequencies and percentages in the table are being scaled up in the ggsurvplot() call, when the weights = frequency argument is used in the call to survfit(). This also happens when you plot a survival curve, rather than the event curve as you have plotted.
This is a bug and should be reported to the survminer authors/github. Here's a reproducible example, using the lung cancer dataset from the survival:: package.
library(survival)
library(survminer)
data(lung)

f0 <- survfit(Surv(time, status) ~ 1, data = lung)
ggsurvplot(f0, ylab = " ", risk.table = "abs_pct", fun = "event" ) 
# This gives the resulting plot, which looks fine.



# Now tweak the example code to make each line represent ten people:
lung2 <- lung
lung2$freq <- 10

f1 <- survfit(Surv(time, status) ~ 1, weights=freq, data = lung2)
ggsurvplot(f1, risk.table = "abs_pct", fun = "event" ) 

# Resulting plot... due to the first n isn't fully printed but is n=2280 (228 x 10) 
# but I didn't want to clutter the code by tweaking the font size.


This quite clearly indicates that not only are the frequencies being scaled in the table by the factor of ten (good!) but so are the percentages (not good!).
