In a follow-up, I have run below cases. The two cases give the same estimate:
data test1;
input strat trt $ responder $ count;
datalines;
1 Test Responder 4
1 Test NR 8
1 Pbo Responder 2
1 Pbo NR 11
2 Test Responder 0
2 Test NR 12
2 Pbo Responder 0
2 Pbo NR 13
;
run;
data test2;
input strat trt $ responder $ count;
datalines;
1 Test Responder 4
1 Test NR 8
1 Pbo Responder 2
1 Pbo NR 11
2 Test Responder 12
2 Test NR 0
2 Pbo Responder 13
2 Pbo NR 0
;
run;
proc freq data=test1;
weight count;
tables strat*trt * responder/commonriskdiff(column=2 cl=MR correct=NO);
run;
proc freq data=test2;
weight count;
tables strat*trt * responder/commonriskdiff(column=2 cl=MR correct=NO);
run;