Date:         Tue, 26 Jul 2011 13:49:29 -0500
Reply-To:     "Data _null_;" <iebupdte@GMAIL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         "Data _null_;" <iebupdte@GMAIL.COM>
Subject:      Re: identifying only the ICD 9 E codes in set of 15
Comments: To: BJ Mattson <bj.mattson@odh.ohio.gov>
In-Reply-To:  <201107261824.p6QFgVuW019235@waikiki.cc.uga.edu>
Content-Type: text/plain; charset=ISO-8859-1

You just need to keep a separate index for the E array.

data diag;
   infile cards missover;
   input (Diagnosis_Primary Diagnosis_2-Diagnosis_15) (:$8.);
   cards;
9945  E927  E8490 27651 7802  4019  2729
87343 9190  E8809 E8496 71690
92411 E9179 E8490 V4365 33818 71946
92320 E8889 E8490 25000 4019  412   V5861
8730  95901 V037  E9174 E8490
84500 8449  E927  E8490 25000
920   7802  E8859 E8498 4019  2729
92311 92401 V037  E8859 E8495 25000 4019  V5867
30000 88102 311   E956  E8490
9100  E8889 E8490 4019  2948  V5861 V1254
8820  E9208 E8490 3332
8261  V065  E916  E8490
8822  V065  E9203 E8490 4019  2729 4279
84500 E9289 E8490 4019  2469
82525 E9170 E8494 4019  53081 2729 20280
94405 94800 E9241 E8490
82535 E9174 E916  E8493 V065
72402 7244  71945 4439  3051  60000 2859  78321 42789 490   78900 56400
49120 73730 E8844
92401 5990  78830 4019  2948  2662  E8859 E8497 V145  V4364 V4365 V1005
V1581
72888 4019  2724  71536 2948  30000 3899  92411 E8859 E8490 V5866
V4364
;;;;
   run;
proc print;
   run;

*** Let's assume E means ICD 9 E;
data diag;
   set diag;
   array d[*] Diagnosis:;
   array e[*] $8 ecode1-ecode8;
   j = 0;
   do i = 1 to dim(d);
      if d[i] eq: 'E' then do;
         if not whichC(d[i],of e[*]) then do; *no dups;
            j + 1;
            e[j] = d[i];
            end;
         end;
      end;
   drop i j;
   run;
proc print;
   run;

On Tue, Jul 26, 2011 at 1:24 PM, BJ Mattson <bj.mattson@odh.ohio.gov> wrote:
> Toby
>
> I had't ever used the IFC function, so thank you for demonstrating it. Much
> more elegant than the kludge I worked out.
>
> Your code works except I need to have any existing E codes begining in
> ecode1 and filling in consecutively (ecode2, then ecode3, etc) so there are
> no skips.
>