Estimate OR on predictor levels

#1
Hi.

I work with a prediction model using logistic regression in the following model setup:

case_control = a b1 b2 b3 c

Each variable a,b*,c have five levels. I like to estimate the combined effect of b1,b2,b3 on different levels on b1,b2,b3.

How could I do this?
 
#3
I presume this should do it. The variables b1,b2,b3 have value 0 as reference for effect. If I treat them as continuous, I then try like this:

proc logistic data = _last_;
model case_control = a b1 b2 b3 c;
contrast 'combined_b1-b3' b1 1 b2 1 b3 1 / estimate = exp e;
run;

Can I read the result as that combined effect of b1 to b3 at level 1 for each predictor compared to 0 on each predictor?