How to Count number of instances in character variables and then filter?

[dellerrrr]

Hello,

I am looking to create a new variable that will count the occurrence of a set of columns if they have characters inside of them. The dataset I am working with contains ER visits within a year of people who eventually died from X. So there are roughly 3 thousand visits and they are already sorted by a variable that ensures each patient has all there er visits in sequence. So for example if patient A went to the ER 5 times in the past year, there are 5 rows with different information but all the same patient id. I want to form a new variable that counts the number of procedures per patient. Each visit has a procedure1-procedure20 with some missing and a lot of blank. Is there any way to form a count array or something that can count each procedure and then filter it by patient, then to add up the total?

Thank you.

[Jagadishkatam]

please provide sample data for better response.

[dellerrrr]

I.E. the data is something like this

Paitent ID   Proc1   Proc2   Proc3   Proc4   Proc5...   Proc20

12               2344    1233      894      

12               1902

12               111      2344       9001    2020   

12               234

13               232      3934       3434

13               1444

A patient can appear from anywhere from 1 time to up to 150 times in a row

[Jagadishkatam · Accepted solution]

please try the below code, please replace the number of variables , at the moment since in the sample data we have 3 variables i mentioned 3, but depending on the number of variables you can replace this 3 in array as well as cnt variable derivation. The CNT2 variable will have the actual count you are expecting. Hope this helps.

data have;
input Paitent_ID:   Proc1:   Proc2:   Proc3:   ;
cards;
12 2344 1233 894   
12 1902 . .
12 111 2344 9001    
12 234 . .
13 232 3934 3434
13 1444 . .
;

data want;
set have;
by Paitent_ID;
array vars(3) proc1-proc3;
cnt=3-nmiss(of proc1-proc3);
retain cnt2;
if first.Paitent_ID then cnt2=cnt;
else cnt2+cnt;
if last.Paitent_ID;
run;

[dellerrrr]

Thank you!

[PGStats]

Change your data structure to long (patient_ID visit_ID proc_ID) and everything will get simpler to do in SAS.

[novinosrin]

data have;
input Paitent_ID:   Proc1:   Proc2:   Proc3:   ;
cards;
12 2344 1233 894   
12 1902 . .
12 111 2344 9001    
12 234 . .
13 232 3934 3434
13 1444 . .
;

data want;
do until(last.Paitent_ID);
set have;
by  Paitent_ID;;
count=sum(count,n(of pro:));
end;
run;