Hi all, I'm trying to calculate European standardised incidence rates by year stratified by sex with my cohort data for survival data with a 15-year follow-up (2000–2015). However, the incidence rates I got after stsplit by the calendar year don't seem right. It seems like it's not using person-years but population N. Any suggestions on correcting this? The codes I used are below:
stset outcome, scale(365.25) id(id) failure(death==1) enter(enter_time) exit(end_date)
gen t=round(_t)
replace t=1 if t==0 & _d==1
stsplit period, after (time=mdy(1,1,2000)) at (1(1) 15)
dstdize _d t ageband if period==2000, by (sex) format (%9.6f) using (europend_stand)
The europend_stand file contains ageband population information of the European standard population. The incidence rates I got seem to be a 10-fold difference when compared to crude rates with other commands. Thanks for the help.
stset outcome, scale(365.25) id(id) failure(death==1) enter(enter_time) exit(end_date)
gen t=round(_t)
replace t=1 if t==0 & _d==1
stsplit period, after (time=mdy(1,1,2000)) at (1(1) 15)
dstdize _d t ageband if period==2000, by (sex) format (%9.6f) using (europend_stand)
The europend_stand file contains ageband population information of the European standard population. The incidence rates I got seem to be a 10-fold difference when compared to crude rates with other commands. Thanks for the help.
