Date:         Tue, 29 Nov 2011 11:57:36 -0800
Reply-To:     Paul Miller <pjmiller_57@YAHOO.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Paul Miller <pjmiller_57@YAHOO.COM>
Subject:      Re: Merge/coding problem using index, find,
              or some other SAS function (with have and need data)
In-Reply-To:  <1322595470.95931.YahooMailClassic@web161602.mail.bf1.yahoo.com>
Content-Type: text/plain; charset=iso-8859-1

Hello Everyone,

Think I found the solution. It looks like I need to use the indexw function as in the code below.

Thanks,

Paul

proc sql;
        create table test as
        select  *
        from    drugs as d left join names as n
        on      indexw(lowcase(desc), lowcase(names), " /") > 0;
quit;


--- On Tue, 11/29/11, Paul Miller <pjmiller_57@yahoo.com> wrote:

> From: Paul Miller <pjmiller_57@yahoo.com>
> Subject: Merge/coding problem using index, find, or some other SAS function (with have and need data)
> To: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
> Received: Tuesday, November 29, 2011, 1:37 PM
> Hello Everyone,
>
> I have some data about chemotherapy and other medical
> procedures in a variable (called "desc") in one of my
> datasets. I want to code whether or not each entry in this
> variable is chemotherapy using information from a variable
> in another dataset (called "names").
>
> Often, the lowcase() version of the variables matches and
> so linking and coding are easy. In some cases though, the
> variable I'm trying to code is a longer string and I need to
> determine if the chemotherapy name in the second variable is
> somewhere inside it. An added complication is that "desc"
> can match more than one value in "names," but I don't want
> to wind up with multiple rows for "desc" whenever this is
> the case.
>
> I've provided some sample data below to illustrate what I'm
> trying to do. I've been trying to link the rows using proc
> sql and the index function but am unable to link properly
> for strings like "5FU/Fluorouracil" and "Leucovorin
> Calcium."
>
> Is there some easy way to get the result I need?
>
> Thanks,
>
> Paul
>
>
> data drugs;
>     input desc $ 1-25;
>     cards;
>     Venipuncture
>     Ondansetron
>     Bevacizumab
>     Carboplatin
>     5FU/Fluorouracil
>     Leucovorin Calcium
> ;
> run;
>
> data names;
>     input names $ 1-25;
>     cards;
>     Bevacizumab
>     Carboplatin
>     5FU
>     Leucovorin
> ;
> run;
>
> data need;
>     length desc $ 25;
>     input desc $& chemo $;
>     cards;
>     Venipuncture  No
>     Ondansetron  No
>     Bevacizumab  Yes
>     Carboplatin  Yes
>     5FU/Fluorouracil  Yes
>     Leucovorin Calcium  Yes
> ;
> run;
>