Prev: Interesting finding: Proc SQL create quoted name dataset!
Next: Interesting finding: Proc SQL create quoted name dataset!
From: Trish Bous on 26 Jan 2010 19:17
The data are numeric:
Hi All,

I am having some problems selecting a substring of data, and I was hoping
someone might have some insight on the problem.

I am trying to select ICD-9 codes. My code works successfully in other
datasets, but I am not able to select my observations in the below dataset
and I am not sure why.

My code that works on other datasets is as follows:


data z.NMSC;
set z.namc94;
tag = 0;
array prx (3) diag1-diag3;
do i = 1 to 3;
if substr (prx {i}, 1, 3) in ('171', '172') then tag = 1;
end;
run;

In the dataset described below, the code does not select any observations.
However, if I change the code to the below, I select a few observations, but
not the entire set I want:


data z.NMSC;
set z.namc94;
tag = 0;
array prx (3) diag1-diag3;
do i = 1 to 3;
if prx{i} in ('172100', '172110') then tag = 1;
end;
run;

Any help on the problem? As a work around, I could list all possible codes
in the (), but there has to be a better way.


Dataset:

26 DIAG1 Num 8
27 DIAG2 Num 8

Example of data:

Obs DIAG1 DIAG2 DIAG3

1 172100 147790 900000
2 202509 900000 900000
3 184490 140190 144390
4 159900 207000 900000
5 172110 900000 900000
6 143590 900000 900000
7 141490 141390 900000
8 156400 145580 140190
9 147390 137230 149390
10 173530 147790 900000
11 149600 116200 900000
12 171900 100880 118500


Thank you!
 | 
Pages: 1
Prev: Interesting finding: Proc SQL create quoted name dataset!
Next: Interesting finding: Proc SQL create quoted name dataset!