[R] Confidence bands with function survplot

David Winsemius dwinsemius at comcast.net
Mon Dec 3 23:45:14 CET 2012


On Dec 3, 2012, at 1:19 PM, Tian3507 wrote:

>
> Dear all,
>
> I am trying to plot KM curves with confidence bands with function  
> survplot under package rms.
>
> However, the following codes do not seem to work. The KM curves are  
> produced, but the confidence bands are not there.
>
> Any insights? Thanks in advance.
>
> library(rms)
> ########data generation############
> n <- 1000
> set.seed(731)
> age <- 50 + 12*rnorm(n)
> label(age) <- "Age"
> sex <- factor(sample(c('male','female'), n, TRUE))
> cens <- 15*runif(n)
> h <- .02*exp(.04*(age-50)+.8*(sex=='female'))
> dt <- -log(runif(n))/h
> label(dt) <- 'Follow-up Time'
> e <- ifelse(dt <= cens,1,0)
> dt <- pmin(dt, cens)
> units(dt) <- 'Year'
> dd <- datadist(age, sex)
> options(datadist='dd')
> S <- Surv(dt,e)
> tte<-data.frame(dt,e,age,sex)
> ############male group KM curve with bands#############
> f <- survfit(Surv(dt,e)~1, data=tte,subset=sex=='male')
> survplot(f,conf="bands", conf.int=0.95)

Reading the "manual" would seem to be the first step:

" conf.int
..... This argument is ignored for fits from survfit, which must have  
previously specified confidence interval specifications."

Although I then noted that your code (and the obvious shift of  
conf.int as an argument to survfit) _both_ created the same confidence  
bands on my machine, so we appear to have different versions (or OS or  
machines)  running. Reading the help page for survfit leaves me a bit  
puzzled since the default for conf.int is 0.95, so I am unable to  
explain why you did not get the confidence bands.

OS= Mac 10.6.8; R= 2.15.2;  rms_3.5-0  Hmisc_3.9-3

-- 

David.



> ########notice no confidence band created#############
> ############female group KM curve with bands#############
> g <- survfit(Surv(dt,e)~1, data=tte,subset=sex!='male')
> survplot(g,conf="bands", conf.int=0.95,col=3, add=T)

-- 

David Winsemius, MD
Alameda, CA, USA




More information about the R-help mailing list