Can´t use survfit on some data.frames - r

Question

I have a dataset I´m going to use for survival analysis, and it seems to be working fine when I use the whole set. However, once I slice it into smaller dataframes using data[which(data$variable1=="somevalue")]the thing seems to break down.
Most of the resulting smaller dataframes work fine, but some are a problem. In the problematic ones, I can use summary(survfit(Surv(time, status)~variable2, data=smalldataframe))$surv without a problem, but when I try summary(survfit(Surv(time, status)~variable2, data=smalldataframe), time=5)$surv, it throws Error in array(xx, dim = dd) : negative length vectors are not allowed.
I´ve tried looking at the data, to see if I have any weird values, like negative times, but there aren´t any. Besides, if there were a problem with that, the full dataframe should be throwing an error too, but it doesn´t. All the smaller dataframes are created using the same line of code, so I also don´t understand why they are acting differently. And mostly, I don´t understand why summary(survfit(...))$surv works fine, as does plot(survfit(...)), but when I want to calculate survival at a specific time, it suddenly doesn´t like the data anymore.
Here´s one of the offending dataframes
test <-
structure(list(time2 = c(0.15, 2.08, 2.06, 0.32, 39.45, 39.09,
2.57, 3.64, 13.57, 36.57, 36.26, 0.78, 0.1, 33.94, 3.1, NA, 1.77,
28.38, 1.24, NA, 1.87, 25.83, 2.62, 1.57, 1.6, 22.74, 21.03,
20.54, 20.03, 0.97, 19.35, 18.09, 2.61, 17.68, NA, 3.85, 3.52,
11.22, 11.52, 11.04, 10.51, 1.68, 10.4, 10.61, 9.01, 9.05, 7.8,
0.11, 4.83), status = c(1, 1, 1, 1, 1, 0, 1, 1, 0, 0, 0, 1, 1,
0, 1, NA, 1, 1, 1, NA, 1, 0, 1, 1, 1, 0, 0, 0, 0, 1, 0, 0, 1,
0, NA, 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 0), cas_dg = c(1,
2, 2, 2, 3, 3, 3, 3, 3, 3, 3, 3, 3, 4, 4, 4, 4, 4, 5, 5, 5, 5,
6, 6, 6, 6, 6, 6, 6, 7, 7, 7, 7, 7, 7, 8, 8, 8, 8, 8, 8, 8, 8,
8, 9, 9, 9, 9, 9)), .Names = c("time2", "status", "cas_dg"), row.names = c(NA, -49L), class = "data.frame")
The call that is giving me trouble is summary(survfit(Surv(time2, status)~cas_dg, data=test), time=5)$surv and that only with some of the smaller dataframes.

Answer 1

You need to use argument extend=TRUE in summary; according to ?summary.survfit:
extend: logical value: if TRUE, prints information for all specified
‘times’, even if there are no subjects left at the end of the
specified ‘times’. This is only valid if the ‘times’
argument is present.
So for your sample data, you can do:
fit <- survfit(Surv(time2, status) ~ cas_dg, data = test);
summary(fit, time = 5, extend = TRUE)$surv;
#[1] 0.0000000 0.0000000 0.5555556 0.5000000 0.3333333 0.5714286 0.6000000
#[8] 0.6666667 0.8000000