Date: Thu, 8 Jan 2009 14:59:27 -0800
Reply-To: Haris <Karovaldas@GMAIL.COM>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: Haris <Karovaldas@GMAIL.COM>
Subject: Re: PHREG: Left Truncate vs. Covariate
Content-Type: text/plain; charset=ISO-8859-1
The definition for LEFT CENSORING I am using is from Paul Allison's
text on survival: you left censor events that "occurred EARLIER than
some time t."
The definition of LEFT TRUNCATION from the same source: you left
truncate "periods of time when it is known that a particular
individual is not at risk for some reason."
In my case, according to these definitions, TRUNCATION is clearly the
case, not CENSORING.
Trouble with using days after discharge is in the fact that patient
who died 30 days after MI and 2 days after MI could have the same 1-
day survival times if the first one was in the hospital for 29 days
and the second one for 1 day.