Message-ID: <Pine.A41.4.44.0305200808140.171844-100000@homer07.u.washington.edu>
Date: 2003-05-20T15:15:12Z
From: Thomas Lumley
Subject: survit function and cox model with frailty
In-Reply-To: <Pine.SOL.4.44.0305200914220.28571-100000@mspacman.gpcc.itd.umich.edu>
On Tue, 20 May 2003, Thomas W Blackwell wrote:
> I reassure you that you CAN get to the bottom of this.
> I think you're making progress. One step going forward
> might be to look at help("model.frame"), since that's
> what causes the initial error message, and think about
> how the arguments get matched in survfit().
>
Actually I think one CAN'T get to the bottom of this. I don't think
survfit works on models with frailties.
I don't actually understand what the intent is (and why the multiple time
periods with identical covariates), but it isn't going to be at all
straightforward to do a proper prediction for a single new case: the
survival curve should be the survival distribution with the frailty
integrated out, which is hard.
It should be possible to do a prediction setting the frailty to 1, but it
isn't. Given that coxph will fit user-defined penalised likelihoods
survfit would have to be fairly clever to guess what it was supposed to do
in each circumstance.
-thomas