Reputation: 11
enter image description hereI did a cox regression model. The data was classfied according to a variable. The three group was 7,143,7 according the ECOG score. The 95%CI was wide. In this dataset, is it proper to fit a cox model?How to explain the wide CI?
sas:
data efffreq2;
set efffreq1;
if treat=1 then treat1=1; else treat1=0;
if treat=2 then treat2=1; else treat2=0;
run;
proc phreg data= efffreq2;
where fasfl="Y" and mcode="1";
model os*status(0)=treat1 treat2 tt1 tt2 /firth risklimits=pl itprint;
tt1=treat1*os;
tt2=treat2*os;
EFFECT1:TEST tt1,tt2;
run;
I tried the dummy variable ECOG and added the ECOG*TIME in the procedure. And i also tried the FIRTH estimate. It didnnot work.
Upvotes: 1
Views: 21