j******1
发帖数: 62 | 1 PSA is a continuous variable and race has two levels, one is coded 1 and
the other is coded 2.
Question: What are the odds ratio estimates and the 95% CI for a 4 ng/ml
change in PSA within each race?
I tried to write the SAS code as below. But it does not work.
proc logistic data=lib1.pros descending;
model capsule =race psa race*psa/
units psa= 4 ;
contrast 'psa(race=1)' race 1 psa*race ?/estimate=exp;
contrast 'psa(race=2)' race 2 psa*race ?/estimate=exp;
run;
Thank you so much! | q********i
发帖数: 795 | 2 1. to calculate ors for each race, you dont need interaction, simply do
separate logit for each category using BY statement.
2. to get the or per 4 ng/nl, just multiple the point estimate by 4, and
then exponential it, same with the se | b*******y
发帖数: 30 | | s*****h
发帖数: 232 | 4 can you use contrast statement for continuous variables?
【在 j******1 的大作中提到】
: PSA is a continuous variable and race has two levels, one is coded 1 and
: the other is coded 2.
: Question: What are the odds ratio estimates and the 95% CI for a 4 ng/ml
: change in PSA within each race?
: I tried to write the SAS code as below. But it does not work.
: proc logistic data=lib1.pros descending;
: model capsule =race psa race*psa/
: units psa= 4 ;
: contrast 'psa(race=1)' race 1 psa*race ?/estimate=exp;
: contrast 'psa(race=2)' race 2 psa*race ?/estimate=exp;
|
|
