topic array processing in SAS Data Management

array processing

kfluegge — Wed, 13 Feb 2019 13:17:52 GMT

I have the code below to identify psychiatric diagnoses from a principal diagnosis variable (prindiag). Instead of identifying them as yes or no, I would like to uniquely identify them by sub-category. That is, each of the 15 psychiatric diagnoses would be a different number (e.g., 1, 2, 3, 4, 5 ...). Is this possible to do within the data step?

data ip14aNTA; 
set ip14;
array diagvars {1} prindiag;
psyc_flag=0;
do i=1 to 1;
if diagvars{i} in: ( '295' , '296' ,'297' ,'298' ,'300' ,'301', '302', '306' ,'307', '308' , '309' ,'311', '312', '313', '314') 
and diagvars{i} not in ('30252')
then psyc_flag=1;
end;
run;

Re: array processing

PaigeMiller — Wed, 13 Feb 2019 13:46:48 GMT

The WHICHC function does this.

https://documentation.sas.com/?cdcId=pgmmvacdc&cdcVersion=9.4&docsetId=lefunctionsref&docsetTarget=p0jfvenvsqk24vn1q2ypospoq9ij.htm&locale=en

What I don't see is how having numbers 1 through 15 is superior for further analyses, or for reporting, than having the character strings '295' , '296' , etc.

Re: array processing

Astounding — Sat, 16 Feb 2019 00:56:55 GMT

It's mildly complicated by the truncation of in: but it can definitely be done. Here's an approach:

data want;
array temp {999} _temporary_;
if _n_=1 then do subcat = '295', '296', '297', '298', '300', '301', '302', '306',
      '307', '308', '309', '311', '312', '313', '314';
   k + 1;
   temp{k} = k;
end;
set have;
if prindiag not in : ('000', '30252', '   ') then 
psyc_flag = temp{input(prindiag, 3.)};
drop k subcat;
run;

The top part of the DATA step places the numbers 1 through 15 into matching spots of the temporary array.

The final computations omit diagnoses 30252, blank, and anything that begins with 000. Then it looks up the value 1 through 15 or missing, depending on the value of PRINDIAG.

Re: array processing

Tom — Sat, 16 Feb 2019 01:44:45 GMT

Your code doesn't really make much sense. You are looping over the character strings codes but then ignoring them.

Were you trying to make a sparse array using the numeric value of the codes as the index?

array temp {0:999} _temporary_;
if _n_=1 then do x = 295 to 298, 300 to 302, 306 to 309, 311 to 314;
   k + 1;
   temp{x} = k;
end;

Re: array processing

Astounding — Sat, 16 Feb 2019 01:47:48 GMT

Long day ... you're 100% right. Thanks!

Re: array processing

Patrick — Sat, 16 Feb 2019 07:14:30 GMT

@kfluegge

If you're just after re-coding the values in prindiag then code like below should do (untested as no sample data provided).

proc format;
  value $recodePrindiag (default=2)
    '295' = '01'
    '296' = '02'
    '297' = '03'
    '298' = '04'
    '300' = '05'
    '301' = '06'
    '302' = '07'
    '306' = '08'
    '307' = '09'
    '308' = '10'
    '309' = '11'
    '311' = '12'
    '312' = '13'
    '313' = '14'
    '314' = '15'
    other = '-1'
    ;
run;

data ip14aNTA;
  set ip14;
  prindiag_recoded=put(prindiag,$recodePrindiag.);
  psyc_flag= (prindiag_recoded ne '-1');
run;