[R] testing proportional hazard in a Cox model including a time-varying covariate

[Thomas Lumley]

On Mon, 3 Apr 2006, Boudreau Jean-Francois wrote:

> I am using a syntax like coxph(Surv(start, stop, event) ~ X, data) to 
> estimate the effect of X, which may change at each measurement (every 6 
> months). Is there anyone who knows a way to test the proportional hazard 
> assumption in that case?

You can use cox.zph(), or you can just put in interactions between X and 
start.

 	-thomas


>
>
>
> Thank you in advance
>
>
>
> Jean-François Boudreau
>
> Sherbrooke University
>
>
>
>
> 	[[alternative HTML version deleted]]
>
>

Thomas Lumley			Assoc. Professor, Biostatistics
tlumley at u.washington.edu	University of Washington, Seattle