interaction term with a continuous variable in PHREG

[Raymond-milan]

Hello, 

I'm trying to include an interaction term in my cox regression model; one is a categorical variable (severite2=Mild or severe) and the second is continous (AgeAtAgentDate). Althouh i can see the HR estimates for all other variables in the model (the table is attached), I cannot see the HR estimates for the interaction term (AgeAtAgentDate, severite, severite*AgeAtAgentDate). 

proc phreg data=RMDATA.CCDAnalysis NAMELEN=100 MULTIPASS;​

    class Severite2(ref="Mild")  / param=ref;​

    model TimetoHospitDeathcardioByYear*HospitDeathByCardio(0)= severite|AgeAtAgentDate Urbain male NSRG

        / SELECTION=B SLS=1 include=1 ties=discrete rl;​

    ods output ParameterEstimates=EstimCCD ModelBuildingSummary=MSCCD;​

OUTPUT OUT=OUT XBETA=x;​

run;

[Reeza]

---

@Raymond-milan wrote:

Hello, 

I'm trying to include an interaction term in my cox regression model; one is a categorical variable (severite2=Mild or severe) and the second is continous (AgeAtAgentDate). Althouh i can see the HR estimates for all other variables in the model (the table is attached), I cannot see the HR estimates for the interaction term (AgeAtAgentDate, severite, severite*AgeAtAgentDate). 

proc phreg data=RMDATA.CCDAnalysis NAMELEN=100 MULTIPASS;​

    class Severite2(ref="Mild")  / param=ref;​

    model TimetoHospitDeathcardioByYear*HospitDeathByCardio(0)= severite|AgeAtAgentDate Urbain male NSRG

        / SELECTION=B SLS=1 include=1 ties=discrete rl;​

    ods output ParameterEstimates=EstimCCD ModelBuildingSummary=MSCCD;​

OUTPUT OUT=OUT XBETA=x;​

run;

---

You need to add explicit hazardratio statements to get interaction estimates. See the HAZARDRATIO documentation for more details on the specification.