Message-ID: <Pine.A41.4.61b.0411091621170.129478@homer06.u.washington.edu>
Date: 2004-11-10T00:22:11Z
From: Thomas Lumley
Subject: frailty model for gap times
In-Reply-To: <6.1.2.0.1.20041109112147.01b36498@l.mail.virginia.edu>
On Tue, 9 Nov 2004, Lei Liu wrote:
> Hi there,
>
> I have a question on the frailty model for gap times with distinct baseline
> hazard for each gap. Can I use R coxph function to do it? For example, the
> following code assumes an identical baseline hazard for every gap time, but
> what if I want a distinct baseline hazard for each gap?
>
> coxph(Surv(gap,status)~trt+frailty(id))
>
You need a stratified model, eg if gapnumber is 1,2,3,4,.. within each
person
coxph(Surv(gap,status)~trt+frailty(id)+strata(gapnumber))
-thomas