[datatable-help] Competing Risk Nomogram

dandrea dd.brigalla at gmail.com
Mon May 1 11:03:16 CEST 2017


Dear R users,
I have been using STATA for all my biostatistical analyses. For my new
project I needed a competing risk nomogram and switched to R. Sadly I am not
able to produce the nomogram.
I have a dataset of bladder cancer, and I want to build the nomogram for the
prediction of progression after 2 and 5 years. I first run a cox regression

>library(plyr)
>VH$SurvObj <- with(Surv(TimetoProg, Progression == 1))
>res.cox1 <- coxph(SurvObj ~ ConcomitantCIS + Tumorsize + Multifocal + LVI +
VH, data = VH)
>res.cox1
Call:
coxph(formula = SurvObj ~ ConcomitantCIS + Tumorsize + Multifocal + 
    LVI + VH, data = VH)

                  coef exp(coef) se(coef)     z      p
ConcomitantCIS1 -0.115     0.891    0.276 -0.42 0.6769
Tumorsize1       0.403     1.496    0.158  2.54 0.0110
Multifocal1      0.417     1.518    0.160  2.61 0.0091
LVI1             1.196     3.306    0.172  6.94  4e-12
VH1              1.921     6.826    0.160 12.00 <2e-16

First question: why do I not get a reasonable p value for LVI and VH?

>library(rms)
> mynom <- svycox.nomogram(.design = SurvObj, .model = Surv(TimetoProg,
> Progression==1) ~ ConcomitantCIS + Tumorsize + Multifocal + LVI + VH,
> .data = VHtrainset, pred.at = 24, fun.lab = "2yr Prob")
Error: $ operator is invalid for atomic vectors

and now I am really stuck!
I would really appreciate any help!

David



--
View this message in context: http://r.789695.n4.nabble.com/Competing-Risk-Nomogram-tp4733256.html
Sent from the datatable-help mailing list archive at Nabble.com.


More information about the datatable-help mailing list