You might want to run PROC ARBOR directly, which is the background procedure for the Decision Tree node in EM. Please refer to the documentation
http://support.sas.com/documentation/onlinedoc/miner/em43/proc_arbor.pdf. This is an olde version, but is still mostly valid.
For your case, the following generic code should work.
proc arbor data=;
input ...;
target disease;
save importance=;
score data= out=;
quit;
The SAVE IMPORTANCE= dataset contains variable importance results based on splits. The SCORE OUT= dataset contains the output results such as predictions for the score dataset DATA=.