Competing Risk Analysis with primary binary exposure not meeting PH assumption

cmccor — Mon, 31 Oct 2022 21:51:08 GMT

Hi,

I have a survival analysis that I am performing with competing risks (which I am very new at) comparing time to discontinuation of therapy as the outcome. I determined from several methods that my primary exposure, which is binary (1/0) violates the proportional hazards assumption. Since it is my primary exposure of interest, I can't stratify on it, so I figured I would go ahead and perform a piecewise survival analysis from day 0 - 60 and day 60+ as that is where the PH assumption is violated. However, since it is a competing risk analysis, when I try to input the BAYES statement to divide the time axis, I get an error statement that the BAYES statement is not available for the Fine and Gray competing-risk analysis. So I'm a little stuck on what to do since I want to make sure to include the competing risk analysis instead of removing it and treating all the censoring as the event?

The code I have is:

proc phreg data=data;
Class older_45 (ref='0') / Param=reference;
Model CENS_time*CENS_CMPRSK(0) = older_45 / RL EVENTCODE=1;
bayes piecewise=hazard (intervals=(60));
run;

Older_45 is my exposure (an individual being older at age 45 at the start of therapy, yes or no)

topic Competing Risk Analysis with primary binary exposure not meeting PH assumption in Statistical Procedures

Competing Risk Analysis with primary binary exposure not meeting PH assumption