casewise.test: Estimates the casewise concordance based on Concordance and...

In mets: Analysis of Multivariate Event Times

Description

Estimates the casewise concordance based on Concordance and marginal estimate using timereg and performs test for independence

Usage

casewise.test(conc, marg, test = "no-test", p = 0.01)

Arguments

conc	Concordance
marg	Marginal estimate
test	Type of test for independence assumption. "conc" makes test on concordance scale and "case" means a test on the casewise concordance
p	check that marginal probability is greater at some point than p

Details

Uses cluster based conservative standard errors for marginal

Author(s)

Thomas Scheike

Examples

data(prt);

prt <- prt[which(prt$id %in% sample(unique(prt$id),7500)),]
### marginal cumulative incidence of prostate cancer
times <- seq(60,100,by=2)
outm <- comp.risk(Event(time,status)~+1,data=prt,cause=2,times=times)

cifmz <- predict(outm,X=1,uniform=0,resample.iid=1)
cifdz <- predict(outm,X=1,uniform=0,resample.iid=1)

### concordance for MZ and DZ twins
cc <- bicomprisk(Event(time,status)~strata(zyg)+id(id),
                 data=prt,cause=c(2,2))
cdz <- cc$model$"DZ"
cmz <- cc$model$"MZ"

### To compute casewise cluster argument must be passed on,
###  here with a max of 100 to limit comp-time
outm <-comp.risk(Event(time,status)~+1,data=prt,
                 cause=2,times=times,max.clust=100)
cifmz <-predict(outm,X=1,uniform=0,resample.iid=1)
cc <-bicomprisk(Event(time,status)~strata(zyg)+id(id),data=prt,
                cause=c(2,2),se.clusters=outm$clusters)
cdz <- cc$model$"DZ"
cmz <- cc$model$"MZ"

cdz <- casewise.test(cdz,cifmz,test="case") ## test based on casewise
cmz <- casewise.test(cmz,cifmz,test="conc") ## based on concordance

plot(cmz,ylim=c(0,0.7),xlim=c(60,100))
par(new=TRUE)
plot(cdz,ylim=c(0,0.7),xlim=c(60,100))

slope.process(cdz$casewise[,1],cdz$casewise[,2],iid=cdz$casewise.iid)

slope.process(cmz$casewise[,1],cmz$casewise[,2],iid=cmz$casewise.iid)

mets documentation built on May 2, 2019, 4:43 p.m.