It is really uneasy.
You need post real data to get pattern of starting a new line .
data have;
infile "c:\temp\temp.csv" lrecl=3000 length=len;
input have $varying3000. len;
if _n_=1 then delete;
if prxmatch('/^[a-z]{4,}/i',have) then group+1;
run;
data want;
do until(last.group);
set have;
by group;
length want $ 3000;
want=cats(want,have);
end;
length PERSON_ID DEPT_ID DATE_JOINED $ 100;
PERSON_ID=scan(want,1,'<|>');
DEPT_ID=scan(want,2,'<|>');
DATE_JOINED=scan(want,3,'<|>');
drop have;
run;