How to estimate overall hazard ratio in interaction PHREG

#1
Hi SAS friends,

I have a question about estimating main hazard ratios in Cox models with interactions.

Weberian asked a slighltly similar question (Hazardratio statement, interaction in Proc Phreg (cox-regression)) but it does not answer this.

I am looking at the interactive effects of X according to Y on death. Here is the code:

proc phreg data=Mortality_M3_72 covs(aggregate);
class X(ref=first) Y(ref=first);
model (monthfrombl, monthfrombl_stop)*death(0)= X|Y /RL TIES=EFRON;
id scrssn_n;

hazardratio "Effect of X in Y groups" X/diff=ref at(Y=all);
run;

I want both overall hazard ratio and interactive hazard ratio from the same model

I only get hazard ratios of X according to Y and vice versa. Here is the output





For instance, I want X=1vs0 in Y =all

How should I proceed?
Thanks for your help!
 

Attachments

hlsmith

Less is more. Stay pure. Stay poor.
#2
I am not sure what you want has utility. So you have two categorical independent variables, but want to know the base effect of X1 vs 0 across all Y's. I don't below you can get that when you put Y in the model, since the intercept will always be the reference group for both X and Y, and even if you could get it, the model is still conditional on the interactions being in the model, which have reference groups. I would be interested if you can find an instance in the literature where some presents such an estimate.

You could run a model without Y to get the X1 vs 0 estimate, but that would be a completely different model and if there really is an interaction the main effect of X would kind of be moot.

I might also throw out that if you are looking at all of these interactions, you should correct your alpha for false discovery or have some type of regularizing process to combat incidental findings that won't replicate.