Use the below code.   data want; set have; by Subjid; retain ecref; if kitno ne . then ecref = kitno; run; ... View more

Then run this code, based on my earlier "have" dataset: proc sql; create table flag as select paramcd, case when count(distinct aval) = 1 then 'Y' else ' ' end as flag from have group by paramcd having flag = 'Y'; quit; data want; merge have flag ; by paramcd; if not first.paramcd then flag = ' '; run; ... View more

data have; input (usubjid parameter visit) (:$12.); cards; 1001 chemistry screening 1001 chemistry screening 1001 chemistry screening 1001 chemistry screening 1001 Hematology screening 1001 Hematology screening 1001 Hematology screening 1001 urinalysis screening 1001 urinalysis screening 1001 urinalysis screening 1001 chemistry Day1 1001 chemistry Day1 1001 chemistry Day1 1001 hematology Day1 1001 hematology Day1 1001 urinalysis Day1 ; data want; set have; by usubjid visit parameter notsorted; if first.parameter; run; ... View more

If you really want to change the shape of your dataset (a la Proc Transpose) then the following code works: data two ;   drop aeterm aedecod aesev ;   set one ;   array ae_vars (3) aeterm aedecod aesev ;   do i = 1 to 3 ;     ae = ae_vars(i) ;     if i > 1 then subjid = "";     output ;   end ; run ; ... View more

You can't fulfil the requirement, at least not in SAS. A numeric variable cannot contain mixed datetime and date values, this since formats can work only with either type. This is again feels like that someone who doesn't understand the implications for a requirement is set to specify one. And you with a quite distance from the end user just have to comply, no questions asked. As suggested by others, set a valid time value when it's missing. Check with your stakeholders for guidance. Or store the concatenated value in a string. ... View more

When you use the || operator unless you specifically trim or strip a variable then the variable is padded to its assigned length. Please look at the results of this example: data example; length x $ 25 long y z $ 50; x= 'short'; long= x||' word'; y = strip(x)||' word' ; z = catt(x,' word'); run; the CATT function (and related CATS) may be better choices than the || operator.   With out showing how you are getting that result it may be that the split line behavior is attempting to show 50 or 60 characters in a table or view that is only 20 or so wide and so the result is wrapping. ... View more

One variation:   newvariable = catx( ' (', dose, doseu) || ')' ; ... View more

A small example:   proc report data=sashelp.cars; where make="Acura"; columns model ("Miles per Gallon" MPG_City MPG_Highway); define model / format=$24.; define MPG_City / "City"; define MPG_Highway / "Highway"; run; Miles per Gallon Model City Highway MDX 17 23 RSX Type S 2dr 24 31 TSX 4dr 22 29 TL 4dr 20 28 3.5 RL 4dr 18 24 3.5 RL w/Navigation 4dr 18 24 NSX coupe 2dr manual S 17 24 The expression in parentheses in the columns statement does it.  ... View more

where is the subject column?   data want; set have; by subject; retain _date; if first.subject then call missing(_date); if not missing(date) then _date=date; else date=_date; run; ... View more