Testing for effect modification across race (4 strata) in Cox proportional model

sas_user4 — Tue, 03 May 2016 20:32:27 GMT

Hello,

I am running Cox regression and wanted to test effect modification across race (4 strata). This is the code I am using:

proc phreg data = ABC;

model asthmatt*asthma(0) = p_cvscore tp_cvscore/risklimit;
tp_cvscore=p_cvscore*log(asthmatt);
test_proportionality: test p_cvscore, tp_cvscore;

run;

Is testing for effect modification the same for proportionality assumption? I need explanation and if my code correct or wrong, please. Assuming the test is approriate in this case, what other options to use if the test of proportionality was sig? I mean what to do, what statistical solutions should I proceed with?

Thanks,

Re: Testing for effect modification across race (4 strata) in Cox proportional model

sas_user4 — Tue, 03 May 2016 20:37:06 GMT

To add, I got this error message (log says P_CVSCORE in TEST statement test_proportionality is not a parameter in the model) when I used this code:

proc phreg data = ABC;

class p_cvscore race1c education salary ;

model asthmatt*asthma(0) = p_cvscore tp_cvscore /risklimit;

tp_cvscore=p_cvscore*log(asthmatt);

test_proportionality: test p_cvscore, tp_cvscore;

run;

topic Re: Testing for effect modification across race (4 strata) in Cox proportional model in Statistical Procedures

Testing for effect modification across race (4 strata) in Cox proportional model

Re: Testing for effect modification across race (4 strata) in Cox proportional model