2
$\begingroup$

I am carrying out survival analysis for two populations of mammals. The data available is GPS collar data: date collared, date of death or date of end of study. I am using the non-parametric Kaplan Meier estimator in R to create survival curves. And I am using a log-rank test to compare between the survival curves of my two study sites. The data I used was number of days that each individual was collared (or alive) and whether or not they were alive at the end of the study.

I have two concerns with what I have done so far.

I haven't managed to find code to analaysis specifically for annual survival. To display annual survival I shortened the x-axis to 365 days. I worry that 1) this is not correct and 2) that my log-rank test only tests for differences between the 2 complete curves, and not between just one year.

Additionally, I found a paper (Pollock et al, 1989) which modified the Kaplan Meier estimator (originally for medical trials) to allow for a staggered entry design, useful for collaring studies. I have two study sites. At one study site all jackals were collared on the same day. For the other study site the jackals were collared over a range of dates. I followed a tutorial online and their data did share the same study start date. I am looking for annual survival, and not survival at a set point in time ie end of a specific season or at a specific event, and so I am unsure if the Pollock version would be more approporiate to use or not?

I did not use the Cox-proportional hazard because we have no covariates

$\endgroup$
1
$\begingroup$

I'm not sure if you are still looking, but generally you might try to use survival package and:

output<-summary(survfit(Surv(as.numeric(enter, exit), event)~(as.factor(year)), conf.type = "log-log", data=data)).

From your output you can extract year (output$strata) and survival values (output$surv) and SE and CI components (output$std.err, output$lower, output$upper)

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.