R help - Mar 2005 - Cox model qustion in R

Dear all,

I have a beginner's R question for coxph() and survfit().
I have a data set with survival time (no censoring) and treatment
indicator(0/1). I did something like

fit<-coxph( Surv(t)~treat,
method="breslow",data=data)
> fit
Call:
coxph(formula = Surv(t) ~ treat, data = data, method = "breslow")


       coef exp(coef) se(coef)     z       p
treat -1.96     0.140    0.243 -8.07 6.7e-16

Likelihood ratio test=80.3  on 1 df, p=0  n= 144 
> plot(survfit(fit))
But it only gives me the curve with combined treatment. How do I get a plot with
two curves for different treatments.
I don't want to do 
new<-data.frame(treat=c(0,1))
and plot(survfit(fit,newdata=new)) cause it gives you predicted curves for two
treatments.

Thanks you so much for your help.

Jimmy Chu
	[[alternative HTML version deleted]]

On Wed, 30 Mar 2005, Chao Zhu wrote:
>
> But it only gives me the curve with combined treatment. How do I get a 
> plot with two curves for different treatments. I don't want to do 
> new<-data.frame(treat=c(0,1)) and plot(survfit(fit,newdata=new))
> cause  it gives you predicted curves for two treatments.
>
I think you are going to have to be clearer about what you want.  If you 
want the Kaplan--Meier estimator for each treatment separately then
   plot(survfit(Surv(t)~treat))
will do it. Otherwise I don't know what you're trying to do.

By the way, do you really have no censoring in the time variable?

 	-thomas