Date:         Wed, 24 Dec 2008 22:12:50 +0000
Reply-To:     naught@NIL.COM
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         naught@NIL.COM
Organization: Duke University, Durham, NC, USA
Subject:      Re: PHREG: counting process style and residuals
Comments: To: sas-l@uga.edu

Shawn Haskell <shawn.haskell@state.vt.us> wrote:
> well looking through survival papers and texts, I see "time-dependent"
> used to describe covariates that: 1) remain constant for an indiviudal
> throughout the duration of a study and violate the PH assumption, thus
> requiring extension of the Cox model, and 2) change periodically for
> an individual within the time of the study but may not necessarily
> violate the PH assumption - maybe someone gains weight and alters
> their expected hazard of heart-attack during a study, or something
> like that.  Maybe I'm all wet on this, but i would call the first
> "time-dependent" and the second "time-varying" so as not to be
> confused.  Maybe someone else can tell me if my understanding is
> bogus.
>

My experience is the same as yours, that the terms are often used
interchangeably.  I can't recall ever coming across an authoritative
source that definitely describes each term.  I agree with you in terms of
how I use them.

Mike Babyak