If critera is met in any of several variables, create new variable and populate with 'yes'

[anle2079]

Hello, it has been a while since I've used SAS (using University Edition) and I'm not sure how to go about a problem I'm having.

I have a cohort of patients, and each patient has had data recorded at 5 visits. At each visit, it's recorded if a referral was needed (1) or not (0). So each patient might have data that looks like this:

ref1= 1

ref2=0

ref3=0

ref4=0

ref5=1

I need to create a new variable (referral) that is populated with 'yes' or 'no' if any one of the ref#=1; so if the patient ever needed a referral, during any visit, the new variable will reflect that.

Right now I am using the following code, but something isn't working. I feel like I need to use an array or DO loops, but like I say, it's been a while. Thank you for your help!

DATA want;

SET have;

IF (ref1-ref5) = .         THEN referred = ' ';

ELSE IF (ref1-ref5) = 1        THEN referred = ' Yes';

ELSE IF (ref1-ref5) = 0        THEN referred = ' No';

RUN;

[maguiremq · Accepted solution]

You can use an array or DO loop, but this is probably easier to understand and gets the job done.

data have;
input pt ref1-ref5;
datalines;
1 1 0 1 1 0
2 1 0 0 0 0
3 0 0 0 0 0
4 0 0 1 0 0
5 1 1 0 0 0
;
run;

data want;
	set have;
	if sum(of ref:) > 0 then referred = "Yes";
		else referred = "No";
run;

Obs 	pt 	ref1 	ref2 	ref3 	ref4 	ref5 	referred
1 	1 	1 	0 	1 	1 	0 	Yes
2 	2 	1 	0 	0 	0 	0 	Yes
3 	3 	0 	0 	0 	0 	0 	No
4 	4 	0 	0 	1 	0 	0 	Yes
5 	5 	1 	1 	0 	0 	0 	Yes

[ballardw]

Strongly suggest that you use 1/0 coding instead of "Yes" "No".

You can get counts of "yes" by summing, percent yes by means, if all values are the same with a Range of 0 or any different with a range=1, any yes if Max is 1, a 1/0 coded variable. And a few more tricks.

The code for a 1/0 coded variable would be

Referred = max(of ref1-ref5);

If you want to Display "yes" or "no" then create a format that does so:

Proc format;
   value yn
   1='Yes'
   0='No'
   other=' '
  ;
run;

and assign the YN. format just created when you need to see "Yes".

[Tom]

Why would you make the new variable using a different coding scheme than the original?

referred = max(of ref1-ref5);

The MAX() over a series of binary variables test if ANY are true.  The MIN() tests if ALL are true.