[R] Problem pyears and left truncated data

Silke Zachariae silke.zachariae at imise.uni-leipzig.de
Mon Sep 19 13:53:41 CEST 2016

Dear all,

for a project I want to calculate cancer incidence rates by decades (0-20,20-30,30-40,..,70-80).
I have left truncated data, my Surv object has the form Surv(time=age,time2=age2, event, type="counting").

I tried different ways to get results for n, events, and person-years, but the results are not as expected.

The best results are for 
pyears(Surv(age,age2,event)~cut(age,c(0, 20,30,40,50,60,70,80))
             +group, data,scale=1)
However, the person-years is too high, since observation time of patients that start before 20 years of age are added completely to the observation decade of 0-20.

Person 1: 18 -31 years of age observation time
Person 2: 18 - 19 years of age observation time

Result gained: n = 2, pyears = 14  for 0-20 years
Result expected: n=2, pyears=3 for 0-20 years

Furthermore, the documentation says, that the cut function is not advised to use, but tcut does not work with my data.

What do I have to do to use the pyears-function for left truncated data? Are there any working examples for that case?

Many thanks and best wishes, Silke Zachariae

More information about the R-help mailing list