topic Re: How do I do wildcards in Proc Format in SAS Programming

How do I do wildcards in Proc Format

renalstats — Mon, 20 Nov 2017 22:49:35 GMT

What I am trying to do, is use a wildcard for character variables (in this case, ICD-9 codes) within proc format.

To keep things simple (because a lot of ICD coding has a lot of suffixes and we only care for the primary prefix), I am using coding for uncomplicated hypertension (ICD-9 code 401), which has 2 subcodes (4011, 4019). I do not want to add a format definition for 4011 and 4019 (this becomes much more arduous when dealing with many other ICD-9 codes). I simply want to create a format by the prefix, and everything after the prefix will be assigned the same format definition.

The prefix code I am interested in, in this case, is 401 (as a character).

I have tried using a colon

"401:" = "Hypertension Uncomplicated - ICD-9"

in proc format (in an attempt to include 4011 and 4019), to no avail.

Here is my code:

proc format; 
value $test
"401" = "Hypertension Uncomplicated - ICD-9"
; 
run; 
quit; 

data test_form; 
input PatID ICD9$;
datalines; 
1 401
2 4011
3 4019
; 

data test_form; 
set test_form; 
format ICD9 $test.; 
run; 

proc print data = test_form NOOBS;
title "Testing format wildcards"; 
run; 
title;

As for outputs:

Current Output:
PatID	ICD9
1	Hypertension Uncomplicated - ICD-9
2	4011
3	4019

Desired Output:
PatID	ICD9
1	Hypertension Uncomplicated - ICD-9
2	Hypertension Uncomplicated - ICD-9
3	Hypertension Uncomplicated - ICD-9

Re: How do I do wildcards in Proc Format

Tom — Mon, 20 Nov 2017 22:59:18 GMT

You can try ranges.

'401' - '4019' = '....'

So that will match any string that sorts between the two values. Like '4010','4012345678' etc.

Or you could try regular expressions, but I think you will need to switch to using an INFORMAT instead of a FORMAT. At least that is what it looks like on the documentation page.

http://documentation.sas.com/?docsetId=proc&docsetVersion=9.4&docsetTarget=n1jriq5xib5j45n1pwpwzk311v0p.htm&locale=en

Re: How do I do wildcards in Proc Format

renalstats — Mon, 20 Nov 2017 23:18:11 GMT

Using a range ("401" - "4019") did work, but in this particular case I know what the range of suffixes are.

I have not tried the regular expressions yet, will attempt.