R help - Feb 2012 - Packages/functions for competing risk analysis

Hi Rs,

I am analyzing a time to event dataset with several competing risks.

0 = Active by end of study
1 = Stopped treatment to start another treatment
2 = Lost
3 = Dead

My event of interest in Lost to Followup but starting a different treatment
and dying are competing risks.  All 1,2,3 events are events of exiting the
study, but it's only 2-LTFU that we are concerned with  (I know I am
repeating myself, sorry).

I can't seem to find a package or function that can handle 2 competing
risks for the main event.  I realize I may be using the well-known ones,
timereg and cmprisk, wrong.  And if I am, then I'd appreciate a correction.

Here's the model I tried (and it crashes)

comp.risk(Surv(futime, retention==2) ~ sex, data=dta, dta$retention,
causeS=c(1,3), resample.iid=1, n.sim=100)

I apologize, but I am cannot create the example dataset using code.  I am
not able.

The data, though is one to one.

ID   retention  sex  futime
------------------------------------
1       0           1     524.3


etc.

Thank you, in advance, if you can help me.

Best,

Jen

	[[alternative HTML version deleted]]

I have an update for this question.  Now that I have figured out how to do
this correctly I have a new question.

Is there a way to deal with late entries (left truncation) for competing
risk analysis in R?  For so-called period analyses?

I apologize if I haven't provided enough information.  I will be happy to
do so.

Best,

Jen



On Mon, Feb 27, 2012 at 11:09 PM, Jennifer Sabatier <
plessthanpointohfive@gmail.com> wrote:
> Hi Rs,
>
> I am analyzing a time to event dataset with several competing risks.
>
> 0 = Active by end of study
> 1 = Stopped treatment to start another treatment
> 2 = Lost
> 3 = Dead
>
> My event of interest in Lost to Followup but starting a different
> treatment and dying are competing risks.  All 1,2,3 events are events of
> exiting the study, but it's only 2-LTFU that we are concerned with  (I
know
> I am repeating myself, sorry).
>
> I can't seem to find a package or function that can handle 2 competing
> risks for the main event.  I realize I may be using the well-known ones,
> timereg and cmprisk, wrong.  And if I am, then I'd appreciate a
correction.
>
> Here's the model I tried (and it crashes)
>
> comp.risk(Surv(futime, retention==2) ~ sex, data=dta, dta$retention,
> causeS=c(1,3), resample.iid=1, n.sim=100)
>
> I apologize, but I am cannot create the example dataset using code.  I am
> not able.
>
> The data, though is one to one.
>
> ID   retention  sex  futime
> ------------------------------------
> 1       0           1     524.3
>
>
> etc.
>
> Thank you, in advance, if you can help me.
>
> Best,
>
> Jen
>
	[[alternative HTML version deleted]]