Graphically evaluating adjusted HR as a function of time when PH are violated

[jennxxness]

Hi Everyone,

I am running a cause specific hazards regression model and my model currently violates the proportional hazards assumption for my independent variable "case" (checked by both looking at Schoenfeld residuals and the supremum test)

proc phreg data=work.MACE covs(aggregate);
class case (ref='0') income (ref='1') rurality (ref='0');
model time*event (0,2) = case income rurality/ RL ties=efron;
id patientgroup;
run;

My understanding for when the PH assumption is violated, is that I can choose to add a time interaction term for my variable case, to understand where the hazard ratio deviates/changes.

I have seen papers that graphically demonstrate how the adjusted hazard ratio changes as a function of time.

Does anyone have any advice on how I can demonstrate and program this?

Thank you!

[JOL]

Try re-posting under menu --> Analytics -> Statistical Procedures