[R] cumulative incidence plot vs survival plot

[array chip]

Dear Terry,

Thanks for responding to my question. I am starting to understand CI now, but I 
still have a general question: in oncology medical literature, when the study 
endpoint is recurrence, it's so common to see publications with Kaplan-Meier 
survival curves (time to recurrence or recurrence-free survival) with death as 
being censored. Because death is a competing risk for recurrence, does this mean 
those KM curves are not appropriate, instead CI curves should be used?

Thanks very much!

John



----- Original Message ----
From: Terry Therneau <therneau at mayo.edu>
To: array chip <arrayprofile at yahoo.com>
Cc: r-help at r-project.org
Sent: Thu, June 30, 2011 6:48:13 AM
Subject: Re: cumulative incidence plot vs survival plot

> Hi, I am wondering if anyone can explain to me if cumulative incidence
> (CI) is ....

The cumulative incidence curve and the KM are not the same, when there
are multiple outcomes.  See the "etype" argument to survfit, which is
used to create CI curves (?survfit.formula).  For testing differences
between CI curves use the cmprsk library from Gray; it can also draw
curves by the survfit routine has a lot more flexibility.

Terry Therneau