|Date: ||Wed, 28 Jun 2006 13:10:34 -0400|
|Reply-To: ||Kevin Roland Viel <kviel@EMORY.EDU>|
|Sender: ||"SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>|
|From: ||Kevin Roland Viel <kviel@EMORY.EDU>|
|Subject: ||Re: UNIX datastep question|
|Content-Type: ||TEXT/PLAIN; charset=US-ASCII|
On Wed, 28 Jun 2006, plessthanpoinohfive wrote:
> Wow, this is LOTS of good info.
> One thing, I don't have away to tell if I have multiple records from a
> patient. Each row is uniquely id'd as a hospital stay. I would not assume
> that each record in independent. From what I remember of the documentation
> of the data there is nothing there that mentions any exclusion of non-unique
> patients. However, there will definitely be id's with multiple flags.
> Sorry to hear you're not here anymore. Where are you living?
There is a wealth of information available by also posting to SAS-L. I
highly recommend it. I have posted at least a hundred times in which I
was embarassingly WRONG. It is a great place to be corrected or to
expand. We are a gentle group :)
I am not familiar with your dataset. The lack of ability to
descriminate between unique patients is worrying. You can well imagine
that the more ill patients will be repeated multiples time-recurrent MI,
recurrent treatments, etc. You might wish to perform a sensitivity
analysis by attempting to identify candidate repeats via record linkage?
You likely have sex, age, BMI, and possibly other demographics that
might assist you. This is not my specialty, but there is at least one
expert on SAS-L and the good folks at GASUG might be a valuable place to
meet face to face, not to mention the SESUG meets in Atlanta in October :)
I now live in San Antonio. I have not wrestled any tornadoes yet, but I
have been warmly welcomed nonetheless :)
Department of Epidemiology
Rollins School of Public Health
Atlanta, GA 30322