Seeking help with the ESTIMATE function in PROC GENMOD

[djohn051]

PROC GENMOD DATA=TEST;

CLASS
ID
THERAPY (REF="0" PARAM=ref)
AGE_GROUP (REF="0" PARAM=ref)
SEX (REF="F PARAM=ref);

MODEL COMPLETION=THERAPY AGE_GROUP SEX/ DIST=poisson LINK=log;

REPEATED SUBJECT=ID/TYPE=EXCH COVB CORRW;

ESTIMATE 'THERAPY 0 v. 1' THERAPY 1/EXP;

ESTIMATE 'AGE_GROUP 0 v. 1' AGE_GROUP 1/EXP;
ESTIMATE 'AGE_GROUP 0 v. 2' AGE_GROUP 0 1/EXP;
ESTIMATE 'AGE_GROUP 0 v. 3' AGE_GROUP 0 0 1/EXP;

ESTIMATE 'SEX female v. male' SEX 1/EXP;

RUN;

I am trying to write the code for a model that will give me the relative risk of completion of a program based on whether or not you received THERAPY (dichotomous), which AGE_GROUP (4 levels) you were in, and your SEX (dichotomous). I am new to the ESTIMATE function, but I need to use GENMOD because this is to calculate the relative risk. My concern lies in whether or not I correctly codes the estimates for each predictor variable?

[Rajesh3 · Accepted solution]

You must specify all levels of a variable. Please use variable level information from the output to make sure you are correctly understanding the reference category. Hope the below code helps!

PROC GENMOD DATA=TEST;

CLASS
ID
THERAPY (REF="0" PARAM=ref)
AGE_GROUP (REF="0" PARAM=ref)
SEX (REF="F PARAM=ref);

MODEL COMPLETION=THERAPY AGE_GROUP SEX/ DIST=poisson LINK=log;

REPEATED SUBJECT=ID/TYPE=EXCH COVB CORRW;

ESTIMATE 'THERAPY 0 v. 1' THERAPY 0 1/EXP;

ESTIMATE 'AGE_GROUP 0 v. 1' AGE_GROUP 0 1 0 0/EXP;
ESTIMATE 'AGE_GROUP 0 v. 2' AGE_GROUP 0 0 1 0/EXP;
ESTIMATE 'AGE_GROUP 0 v. 3' AGE_GROUP 0 0 0 1/EXP;

ESTIMATE 'SEX female v. male' SEX 0 1/EXP;

RUN;

 Thanks,

Rajesh.