R help - Jun 2006 - Selective Survival Statistics with R

Hello friends and fellow R users,

I have a problem to which I have been unable to find a solution: I am
gathering survival data on patients undergoing treatment with a new kind
of stent. I want to generate survival data and plot survival curves of
these patients based (among other things) on the treating physician. My
data set has been tabulated in the following manner:

Date	(the date the stent was implanted)   
Description (diameter of the stent)
Patient (name)
MRN (ID number)
Age (age in years of the patient)
Physician (last name of the implanting physician)
status (0 or 1)
days (days alive since the stenting procedure)
Cr 
INR
BR
MELD

The data set has over ten physicians. Following the examples in
"Introductory Statistics with R" I have been able to draw cumulative
survival curves and even the survival curves by physician but there are
so many curves on the graph when I do, it is uninterpretable.

I would just like to plot the survival curves of say the top two or
three. That would be much more useful. I have been able to create a
"Surv" object out of my own survival data and that of a colleague in
the
following way using indexing:
>Surv(days[viatorr[6]=="Pierce"|
viatorr[6]=="Ed"],status[viatorr[6]=="Pierce"|viatorr[6]=="Ed"]==1)
[1] 558+ 474+ 472+ 446+ 443+ 429+ 401  395+ 390  390+ 362+ 354  344+ 342
326+
[16] 300+ 284+ 280+ 246+ 237+ 233+ 233  230+ 230+ 225+ 215  199+ 191+
191  184+
[31] 161+ 153  150  129+  69+  52+  38+

I can get even further:

surv.ee.gp<-Surv(days[viatorr[6]=="Pierce"|
viatorr[6]=="Ed"],status[viatorr[6]=="Pierce"|viatorr[6]=="Ed"]==1)
> survfit(surv.ee.gp)
Call: survfit(formula = surv.ee.gp)

      n  events  median 0.95LCL 0.95UCL
     37       9     Inf     390     Inf

But now if I want to plot the data using the following command
>plot(survfit(surv.ee.gp)~Physician)
I receive an error:

Error in model.frame(formula, rownames, variables, varnames, extras,
extranames,  :
        invalid variable type

I have tried indexing Physician, but that fails as well:
> plot(survfit(surv.ee.gp)~Physician[viatorr[6]=="Pierce"|
viatorr[6]=="Ed"]==1)
Error in model.frame(formula, rownames, variables, varnames, extras,
extranames,  :
        invalid variable type

I apologize for this long-winded explanation but I am new to this
program and even newer to Statistics and I wanted to make sure the
context and problem were clear (hopefully).

I would appreciate any guidance in a way forward. 

Thank you,
Greg

On Mon, 2006-06-05 at 18:54 -0400, Barker, Chris [SCIUS]
wrote:
> Its probably easiest/fastest for you either to subset your dataset
> first, or else simply use the subset option in survfit()
> 
>  e.g. 
> 
> survfit( ) has a subset option,  
> survfit( Surv( , ) ~ physician , subset=='Jones")
> 
Chris,

Thank you very much for your kind reply. Using subset worked. I had to
modify the syntax a little from what you posted:

survfit(Surv(days,status==1)~Physician,subset(viatorr,viatorr[6]=="Pierce")

where viatorr is the name of my data set.

Applying plot to the function above generated a survival curve for my
patients. I was also able to plot survival curves for other physicians
as well. This is great!