R help - Jun 2011 - UnoC function in survAUC for censoring-adjusted C-index

Hello,

I am having some trouble with the 'censoring-adjusted C-index' by Uno et
al, in the package survAUC. The relevant function is UnoC.
The question has to do with what happens when I specify a time point t for the
upper limit of the time range under consideration (we want to avoid using the
right-end tail of the KM curve).

Copying from the example in the help file:

TR <- ovarian[1:16,]
TE <- ovarian[17:26,]
train.fit  <- coxph(Surv(futime, fustat) ~ age,x=TRUE, y=TRUE,
method="breslow", data=TR)
lpnew <- predict(train.fit, newdata=TE)
Surv.rsp <- Surv(TR$futime, TR$fustat)
Surv.rsp.new <- Surv(TE$futime, TE$fustat)              

If time is left NULL, the maximum futime in the TE data is assumed:
UnoC(Surv.rsp, Surv.rsp.new, lpnew, time=NULL)
#same as 
UnoC(Surv.rsp, Surv.rsp.new, lpnew, time=max(Surv.rsp.new[,1]))

#which incidentally is the same as what one gets with the usual C-index function
that also extends to the max event time.
survConcordance(Surv.rsp.new ~ lpnew)$conc

Now, if I specify a range times 'tau' (as in Table III of the relevant
publication) I still get the same C for each timepoint:

tau <- Surv.rsp.new[, 1]
> tau 
 [1] 1040 1106 1129 1206 1227  268  329  353  365  377

UnoC(Surv.rsp, Surv.rsp.new, lpnew, time=tau)

 [1] 0.7333333 0.7333333 0.7333333 0.7333333 0.7333333 0.7333333 0.7333333
0.7333333 0.7333333 0.7333333

#even worse, if I sort tau in ascending times I get 0's
UnoC(Surv.rsp, Surv.rsp.new, lpnew, time=sort(tau))
 [1] 0 0 0 0 0 0 0 0 0 0 

This is a potentially very useful function so it would be great if someone could
explain the behaviour here.

Many thanks

Eleni Rapsomaniki

Research Associate/Statistician
Strangeways Research Laboratory
Department of Public Health and Primary Care

University of Cambridge
?

Note that the original C-index is implemented in various validation options
in the rms package related to the cph function (which calls coxph in
survival).
Frank
Eleni Rapsomaniki-2 wrote:
> 
> Hello,
> 
> I am having some trouble with the 'censoring-adjusted C-index' by
Uno et
> al, in the package survAUC. The relevant function is UnoC.
> The question has to do with what happens when I specify a time point t for
> the upper limit of the time range under consideration (we want to avoid
> using the right-end tail of the KM curve).
> 
> Copying from the example in the help file:
> 
> TR <- ovarian[1:16,]
> TE <- ovarian[17:26,]
> train.fit  <- coxph(Surv(futime, fustat) ~ age,x=TRUE, y=TRUE,
> method="breslow", data=TR)
> lpnew <- predict(train.fit, newdata=TE)
> Surv.rsp <- Surv(TR$futime, TR$fustat)
> Surv.rsp.new <- Surv(TE$futime, TE$fustat)              
> 
> If time is left NULL, the maximum futime in the TE data is assumed:
> UnoC(Surv.rsp, Surv.rsp.new, lpnew, time=NULL)
> #same as 
> UnoC(Surv.rsp, Surv.rsp.new, lpnew, time=max(Surv.rsp.new[,1]))
> 
> #which incidentally is the same as what one gets with the usual C-index
> function that also extends to the max event time.
> survConcordance(Surv.rsp.new ~ lpnew)$conc
> 
> Now, if I specify a range times 'tau' (as in Table III of the
relevant
> publication) I still get the same C for each timepoint:
> 
> tau <- Surv.rsp.new[, 1]
>> tau 
>  [1] 1040 1106 1129 1206 1227  268  329  353  365  377
> 
> UnoC(Surv.rsp, Surv.rsp.new, lpnew, time=tau)
> 
>  [1] 0.7333333 0.7333333 0.7333333 0.7333333 0.7333333 0.7333333 0.7333333
> 0.7333333 0.7333333 0.7333333
> 
> #even worse, if I sort tau in ascending times I get 0's
> UnoC(Surv.rsp, Surv.rsp.new, lpnew, time=sort(tau))
>  [1] 0 0 0 0 0 0 0 0 0 0 
> 
> This is a potentially very useful function so it would be great if someone
> could explain the behaviour here.
> 
> Many thanks
> 
> Eleni Rapsomaniki
> 
> Research Associate/Statistician
> Strangeways Research Laboratory
> Department of Public Health and Primary Care
> 
> University of Cambridge
> ?
> 
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
> http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
> 

-----
Frank Harrell
Department of Biostatistics, Vanderbilt University
--
View this message in context:
http://r.789695.n4.nabble.com/UnoC-function-in-survAUC-for-censoring-adjusted-C-index-tp3622181p3622289.html
Sent from the R help mailing list archive at Nabble.com.