Search code examples
rsurvival-analysissurvival

Get a 'survfit' object which will be the same size than the original data in case of ties? (survival, R)

I want the survival probability estimated by Kaplan-Meier estimator for each individual of my dataframe. The survfit(Surv(.)) function calculates the survival probability for each unique time ordered by decreasing order.

What would be an elegant way of getting the survival probability for each individual from a survfit object?

library(survival)
data = data.frame(cbind(id = 1:10, 
            time = c(2,3,4,5,2,3,8,9,10,11), 
            status = c(1,0,0,1,0,1,0,1,0,1)))
survfit(Surv(data$time, data$status)~1)$surv
 # 0.9000 0.7875 0.7875 0.6300 0.6300 0.4200 0.4200 0.0000
survfit(Surv(data$time, data$status)~1)$time
 #  2  3  4  5  8  9 10 11

Thanks in advance!

Solution

  • You can put the results in a data.frame and then just merge it back in

    merge(data, 
   with(survfit(Surv(data$time, data$status)~1), data.frame(time, surv))
)

    Here I just use with() to easily extract both columns from the survfit result.