topic Re: Best Way to Do a Frequency Table for Patient Responses in SAS Programming

Best Way to Do a Frequency Table for Patient Responses

unwashedhelimix — Fri, 18 Oct 2024 04:45:25 GMT

I have a data set of hospital patient IDs, and the data shows which of 2 pizza brands (Brand A or B) each unique individual like (they can like one, both, or neither). It looks something like this:

ID    A    B

21   Yes  No
22  No   No
23  Yes  Yes
24  No   Yes
25  Yes  Yes
26  Yes  No
27  No   Yes
.
.
.
.

;

Now, based on that data, I want to know how many patients liked A, B, only A, only B, both, neither one, either one.

What's an effecient approach to this? I was thinking of doing a table where the columns are for A and B, then the rows are Yes or No. Then for each entry, it will have the count of yes/no for A and yes/no for B, but that doesn't seem to be very helpful in showing a report that gives me what I want.

Any ideas on what I can code to produce a report with those values?

Re: Best Way to Do a Frequency Table for Patient Responses

PaigeMiller — Fri, 18 Oct 2024 10:56:29 GMT

Combine the variables A and B into a single variable. Then run PROC FREQ.

data have1;
    set have;
    ab=cats(A,'|',B);
run;
proc freq data=have1;
    tables ab;
run;

Re: Best Way to Do a Frequency Table for Patient Responses

ballardw — Fri, 18 Oct 2024 17:46:02 GMT

My approach is similar to @PaigeMiller only uses an option on tables statement that doesn't require creating a new variable:

proc freq data=have;
   tables a*b / missing list ;
run;

I include Missing so if you have any missing values they appear in the body of the report table.

@unwashedhelimix wrote:

I have a data set of hospital patient IDs, and the data shows which of 2 pizza brands (Brand A or B) each unique individual like (they can like one, both, or neither). It looks something like this:
ID    A    B

21   Yes  No
22  No   No
23  Yes  Yes
24  No   Yes
25  Yes  Yes
26  Yes  No
27  No   Yes
.
.
.
.

;
Now, based on that data, I want to know how many patients liked A, B, only A, only B, both, neither one, either one.

What's an effecient approach to this? I was thinking of doing a table where the columns are for A and B, then the rows are Yes or No. Then for each entry, it will have the count of yes/no for A and yes/no for B, but that doesn't seem to be very helpful in showing a report that gives me what I want.

Any ideas on what I can code to produce a report with those values?

Re: Best Way to Do a Frequency Table for Patient Responses

JOL — Fri, 18 Oct 2024 20:13:37 GMT

data pizza_likes;
input ID $ A $ B $;
datalines;
21 Yes No
22 No No
23 Yes Yes
24 No Yes
25 Yes Yes
26 Yes No
27 No Yes
;
run;

proc freq data=pizza_likes;
tables A*B / nocol nopercent norow;
run;

proc sql;
select
sum(A='Yes') as Likes_A,
sum(B='Yes') as Likes_B,
sum(A='Yes' and B='No') as Likes_Only_A,
sum(A='No' and B='Yes') as Likes_Only_B,
sum(A='Yes' and B='Yes') as Likes_Both,
sum(A='No' and B='No') as Likes_Neither,
sum(A='Yes' or B='Yes') as Likes_Either
from pizza_likes;
quit;