For each subject, you apparently have zero or one observation with non-missing BUN_NHIGH_VA and BUN_NLOW_VA. For all observations with BUN_ND_LA='ticked" you want to apply those non-missing values.
This program is untested. If you were to provide your sample data in the form of a working data set, then I could vouch for this code:
data want;
set have ;
by subject notsorted;
if first.subject then ptr=.;
if BUN_NHIGH_VA^=. and BUN_NLOW_VA^=. then ptr=_n_;
if last.subject then do until (last.subject);
set have;
by subject notsorted;
if BUN_ND_LA='ticked' and ptr^=. then set have (keep=bun_nhigh_va bun_nlow_va) point=ptr;
output;
end;
run;
This reads each group twice. The first pass is to identify which observation has the non-missing values, setting ptr to the identified obs numbers. The second pass re-reads the group, and if BUN_ND_LA='ticked' it uses the "point=ptr" option to read the non-missing values prior to the output statement. Note if a group has non non-missing values, then ptr=. and is not used.
Click image to register for webinar
