Irin,

Then couldn't you just first run a "proc sort nodupkey" to get rid of the records that you don't want?

Art ------- On Tue, 3 Feb 2009 06:07:22 -0800, Irin later <irinfigvam@YAHOO.COM> wrote:

>Arthur, >The thing is that within the same id, desEAse and response� criteria I need to count just those with the different resp_date. >HOWEVER i also�do need�those patients� who have just one record (therefore I do need CK) > >I DO NOT NEED THOSE� with the same resp-date (even if they have the same response). >� >Thank you! I >Irin > > >--- On Mon, 2/2/09, Arthur Tabachneck <art297@NETSCAPE.NET> wrote: > >From: Arthur Tabachneck <art297@NETSCAPE.NET> >Subject: Re: SQL count based on the certain criteria >To: SAS-L@LISTSERV.UGA.EDU, "SUBSCRIBE SAS-L Irin Ros" <irinfigvam@YAHOO.COM> >Date: Monday, February 2, 2009, 6:58 PM > >Irin, > >I'm not sure if I correctly understand the problem. Does the following >come close?: > >data have; > input id $ disease $ response rep_date mmddyy10.; > cards; >10jan DM 1 08/10/2007 >10ian DM 1 12/02/2007 >10ian DM 2 12/02/2007 >10ian DM 3 12/02/2007 >10ian HF 1 12/02/2007 >12mar DM 2 12/03/2008 >12mar DM 2 07/03/2008 >34kat HF 2 11/01/2007 >34kat HF 3 10/02/2007 >45HAT CK 4 9/07/2008 >; > >proc sql; > create table want as > select id,disease,response,count(*) as count > from have > where disease ne 'CK' > group by id,disease,response >; >quit; > >Art >-------- >On Mon, 2 Feb 2009 15:40:04 -0800, Irin later <irinfigvam@YAHOO.COM> >wrote: > >>I have a dataset health with id, disease , response , rep_date fields. >>I need to know how many records within each id & certain disease have >the >same responses under different dates rep_dates: >>For example: >>id????? disease?? response??? rep-date >>10jan?? DM??????? 1??????????? 08/10/2007 >>10ian?? DM??????? 1??????????? 12/02/2007 >>10ian?? DM??????? 2??????????? 12/02/2007 >>10ian?? DM??????? 3??????????? 12/02/2007 >>10ian?? HF??????? 1??????????? 12/02/2007 >>12mar?? DM??????? 2??????????? 12/03/2008 >>12mar?? DM??????? 2??????????? 07/03/2008 >>34kat?? HF??????? 2??????????? 11/01/2007 >>34kat?? HF??????? 3??????????? 10/02/2007 >>45HAT?? CK??????? 4??????????? 9/07/2008 >>? >>My concern is the? combinations of the same ID, disease , RESPONSE'S >value and "only" records? (like id=45HAT10jan)? >>Based on the file I am going to do Proc Freq a Frequency with a count of >records of the same id, disease to create something like the? following >format report: >>? >>For DM: >>REC-CNT??????? RESPONSE VALUE???????????????? FREQ >>1??????????????????? 1????????????????????????????????????????????? 220 >>???????????????????? ?2????????????????????????????????????????????? 30 >>???????????????????? ?3???????????????????????????????????????????? ?48 >>????????????????????? .??????????????? . >>?????? >>2?????????????????? 1????????????????????????????????????????????? 300 >>???????????????????? 3??????????????????????????????????????????? ? 22 >>???????????????????? 5????????????????????????????????????????????? 2??? >> >> >>For HF:.........????????? . >>?How can I implement it in PROC SQL code? Could you please give me a hand? >>Thank you in advance, >>Irin >