plot.bshazard {bshazard}

Plot Method for 'bshazard'

Description

A plot of hazard rate is produced. The overall option allows to plot an hazard rate for each covariate value (assuming proportional hazard).

Usage

## S3 method for class 'bshazard'
plot(x, conf.int = T, overall = T, col = 1, lwd = 1, lty = 1, xlab = "Time", 
ylab = "Hazard rate",border=NA,col.fill="lightgrey",...)

Arguments

x

the result of a call to the bshazard function.

conf.int

Determines whether confidence intervals will be plotted. The default is to do so if there is only 1 curve, i.e., no strata.

overall

Determines whether an overall curve is plotted (default overall=T) or a curve for each covariate value in the data (overall=F). It works only if there are covariates.

col

a vector of integers specifying colors for each curve. The default value is 1.

lwd

a vector of integers specifying line width for each curve. The default value is 1.

lty

a vector of integers specifying line types for each curve. The default value is 1.

xlab

label given to the x-axis.

ylab

label given to the y-axis.

border

the color to draw the border. The deafult value is NA.

col.fill

the color for filling the polygon. The default is lightgrey.

...

other arguments that will be passed forward to the underlying plot method, such as xlab or ylab.

See Also

bshazard,summary.bshazard,print.bshazard

Examples

fit<-bshazard(Surv(time, status==2) ~ 1,data=lung)
plot(fit)


#Example to graphically evaluate the Markov assumpion in an illness-death model
### data simulated under EXTENDED SEMI-MARKOV model
set.seed(123)
n  <- 500
beta<-log(3)
R  <- runif(n, min = 0, max =2)
cat <- cut(R, breaks = seq(0,2,0.5), labels = seq(1,4,1))
T12  <- 1/0.2*( (-log(runif(n, min = 0, max = 1))) / (exp(beta*R)))**(1/0.6)
C  <- runif(n, min =0, max = 10)
T12obs <- pmin(T12, C)
status  <- ifelse(T12obs < C, 1, 0)
T012obs <- R+T12obs
#R: simulated time to illness
#cat: time to illness categorised in 4 classes
#T12obs:  time from illness to death or censoring
#T012obs: time from origin to death or censoring
#status: indicator of death (1=death,0=censoring)

# Hazard estimate in the Clock Reset scale (time from illness) by time to illness class
fit <- bshazard(Surv(T12obs[cat == 1],status[cat==1]) ~ 1)
plot(fit,conf.int=FALSE,xlab='Time since illness',xlim=c(0,5),ylim=c(0,10),lwd=2,col=rainbow(1))
for(i in 2:4) {
  fit <- bshazard(Surv(T12obs[cat == i],status[cat==i]) ~ 1)
  lines(fit$time, fit$hazard, type = 'l', lwd = 2, col = rainbow(4)[i])
}
legend("top",title="Time to illness",c("<=0.5","0.5-|1","1-|1.5","1.5-|2"),col=c(rainbow(4)),lwd =2)

# Hazard estimate in the Clock Forward scale (time from origin) by time to illness class
fit <- bshazard(Surv(R[cat == 1],T012obs[cat == 1],status[cat==1]) ~ 1)
plot(fit,conf.int=FALSE,xlab='Time since origin',xlim=c(0,5),ylim=c(0,10),lwd=2,col=rainbow(1))
for(i in 2:4) {
  fit <- bshazard(Surv(R[cat == i],T012obs[cat == i],status[cat==i]) ~ 1)
  lines(fit$time, fit$hazard, type = 'l', lwd = 2, col = rainbow(4)[i])
}
legend("top",title="Time to illness",c("<=0.5","0.5-|1","1-|1.5","1.5-|2"),col=c(rainbow(4)),lwd =2)

#Alternatively an adjusted estimate can be performed, assuming proportionl hazard 
# this computation can be time consuming!
## Not run: fit.adj <- bshazard(Surv(T12obs,status) ~ cat )
plot(fit.adj,overall=FALSE, xlab = 'Time since illness',col=rainbow(4),lwd=2, xlim = c(0,5), 
ylim = c(0,10))
legend("top",title="Time to illness",c("<=0.5","0.5-|1","1-|1.5","1.5-|2"),col=c(rainbow(4)),lwd =2)
## End(Not run)

### A more general setting with examples of Markov assumption evaluation can be found 
### in Bernasconi et al. Stat in Med 2016