[R] Direct adjusted survival?

[Peter Jepsen]

Thank you for your reply, Thomas. I'm not quite sure whether survexp() does that. It seems that the idea of survexp() is to take the ratetable from a mortality table or Cox model based on one dataset and apply it to another dataset. I'm trying to adjust for confounding, so I want to take the ratetable from a Cox model based on one dataset and apply it to the SAME dataset. Here's an example of how I try to achieve this:

require(survival)
data(pbc)
## compare data to Cox model 
## fit to randomised patients in Mayo PBC data
coxph(Surv(time,status)~edtrt,data=pbc)
m<-coxph(Surv(time,status)~edtrt+log(bili),data=pbc)
m # log(bili) is a strong confounder
plot(survfit(Surv(time,status)~edtrt,data=pbc))
lines(survexp(~edtrt+ratetable(edtrt=edtrt,bili=bili),data=pbc,ratetable=m,cohort=TRUE),col="purple")

The lines that I hoped to be the survival probabilities for each edtrt-group adjusted for confounding by log(bili) are nearly identical to the KM-lines, and they certainly don't appear adjusted for the very strong confounding by log(bili). I'm not quite sure what they are, though.

Ghali et al. claim to have an S-plus implementation of the 'direct adjusted survival' method (Ghali WA, Quan H, Brant R, van Melle G, Norris CM, Faris PD, et al. Comparison of 2 methods for calculating adjusted survival curves from proportional hazards models. JAMA 2001;286:1494-1497). I have found the function here: http://stat.ubc.ca/~rollin/stats/S/surv.html. It is inserted below, but please note that I have made one modification.

I'm still very new to R, so I don't follow exactly what happens. It seems that avg.surv() wants edtrt as a factor that takes integer values?! I realize that this is changes the Cox model specification, but, anyway, this code produces a result that is much closer what I expected:

pbc$edtrt.fac<-as.factor(pbc$edtrt*2)
m2<-coxph(Surv(time,status)~edtrt.fac+log(bili),data=pbc)
fits<-avg.surv(m2, var.name="edtrt.fac", data=pbc, var.values=c("0","1","2"))
matlines(fits$time,fits$fits)

However, avg.surv() does not provide standard errors, hence my question regarding the Zhang paper. If anyone can help me sort out what is going on, I'd be very thankful.

Best regards,
Peter.

***
avg.surv <- function(cfit, var.name, var.values, data, weights)
{
	if(missing(data)) {
		if(!is.null(cfit$model))
			mframe <- cfit$model
		else mframe <- model.frame(cfit, sys.parent())
	}   else mframe <- model.frame(cfit, data)
	var.num <- match(var.name, names(mframe))
	data.patterns <- apply(data.matrix(mframe[,  - c(1, var.num)]), 1,
		paste, collapse = ",")
        data.patterns <- factor(data.patterns,levels=unique(data.patterns))
	if(missing(weights))
		weights <- table(data.patterns)
	else weights <- tapply(weights, data.patterns, sum)
        kp <- !duplicated(data.patterns)
	mframe <- mframe[kp,]
        obs.var <- mframe[,var.num]
        lps <- (cfit$linear.predictor)[kp]
        tframe <- mframe[rep(1,length(var.values)),]
        tframe[,var.num] <- var.values
        xmat <- model.matrix(cfit,data=tframe)[,-1]
        tlps <- as.vector(xmat%*%cfit$coef)
        names(tlps) <- var.values
        obs.off <- tlps[as.character(obs.var)]
        explp.off <- exp(outer(lps - obs.off ,tlps,"+"))
	bfit <- survfit(cfit, se.fit = F)	# Changed from "survfit.coxph" to "survfit"
        fits <- outer(bfit$surv,explp.off,function(x,y) x^y)
        avg.fits <-
           apply(sweep(fits,2,weights,"*"),c(1,3),sum)/sum(weights)
        dimnames(avg.fits) <- list(NULL,var.values)
        list(time=bfit$time,fits=avg.fits)
}
***
-----Oprindelig meddelelse-----
Fra: r-help-bounces at r-project.org [mailto:r-help-bounces at r-project.org] På vegne af Thomas Lumley
Sendt: 30. januar 2008 00:31
Til: Peter Jepsen
Cc: r-help
Emne: Re: [R] Direct adjusted survival?

On Wed, 30 Jan 2008, Peter Jepsen wrote:
>
> I am trying to find an R function to compute 'direct adjusted survival'
> with standard errors. A SAS-macro to do this is presented in Zhang X,
> Loberiza FR, Klein JP, Zhang MJ. A SAS macro for estimation of direct
> adjusted survival curves based on a stratified Cox regression model.
> Comput Methods Programs Biomed 2007;88:95-101. It appears that this
> method is not implemented in R. Can anyone prove me wrong?
>

This looks like what survexp() does. It's hard to be sure, since I can 
only find the abstract online.

 	-thomas

______________________________________________
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.