restricted.residual.mean {timereg}

Estimates restricted residual mean for Cox or Aalen model

Description

The restricted means are the

\int_0^\tau S(t) dt

the standard errors are computed using the i.i.d. decompositions from the cox.aalen (that must be called with the argument "max.timpoint.sim=NULL") or aalen function.

Usage

restricted.residual.mean(out, x = 0, tau = 10, iid = 0)

Arguments

out

an "cox.aalen" with a Cox model or an "aalen" model.

x

matrix with covariates for Cox model or additive hazards model (aalen).

tau

restricted residual mean.

iid

if iid=1 then uses iid decomposition for estimation of standard errors.

Details

must have computed iid decomposition of survival models for standard errors to be computed. Note that competing risks models can be fitted but then the interpretation is not clear.

Value

Returns an object. With the following arguments:

mean

restricted mean for different covariates.

var.mean

variance matrix.

se

standard errors.

S0tau

estimated survival functions on time-range [0,tau].

timetau

vector of time arguments for S0tau.

Author(s)

Thomas Scheike

References

D. M. Zucker, Restricted mean life with covariates: Modification and extension of a useful survival analysis method, J. Amer. Statist. Assoc. vol. 93 pp. 702-709, 1998.

Martinussen and Scheike, Dynamic Regression Models for Survival Data, Springer (2006).

Examples

### this example runs slowly and is therefore donttest
data(sTRACE)
sTRACE$cage <- scale(sTRACE$age)
# Fits Cox model  and aalen model 
out<-cox.aalen(Surv(time,status>=1)~prop(sex)+prop(diabetes)+prop(chf)+
	       prop(vf),data=sTRACE,max.timepoint.sim=NULL,resample.iid=1)
outa<-aalen(Surv(time,status>=1)~sex+diabetes+chf+vf,
data=sTRACE,resample.iid=1)

coxrm <- restricted.residual.mean(out,tau=7,
   x=rbind(c(0,0,0,0),c(0,0,1,0),c(0,0,1,1),c(0,0,0,1)),iid=1)
plot(coxrm)
summary(coxrm)

### aalen model not optimal here 
aalenrm <- restricted.residual.mean(outa,tau=7,
   x=rbind(c(1,0,0,0,0),c(1,0,0,1,0),c(1,0,0,1,1),c(1,0,0,0,1)),iid=1)
with(aalenrm,matlines(timetau,S0tau,type="s",ylim=c(0,1)))
legend("bottomleft",c("baseline","+chf","+chf+vf","+vf"),col=1:4,lty=1)
summary(aalenrm)

mm <-cbind(coxrm$mean,coxrm$se,aalenrm$mean,aalenrm$se)
colnames(mm)<-c("cox-res-mean","se","aalen-res-mean","se")
rownames(mm)<-c("baseline","+chf","+chf+vf","+vf")
mm