crrQR {cmprskQR}

Competing Risks Quantile Regression

Description

quantile regression modeling of subdistribution functions in competing risks

Usage

crrQR(ftime, fstatus, X, failcode=1, cencode=0, 
tau.range=c(0.01,0.99), tau.step=0.01, subset, 
na.action=na.omit, rq.method="br", variance=TRUE)

## S3 method for class 'crrQR'
print(x, ...)

Arguments

ftime

vector of failure/censoring times

fstatus

vector with a unique code for each failure type and a separate code for censored observations

X

matrix (nobs x ncovs) of covariates

failcode

code of fstatus that denotes the failure type of interest

cencode

code of fstatus that denotes censored observations

tau.range

vector of length 2 denoting the range of quantiles

tau.step

grid size on tau.range (spacing between two grid points)

subset

a logical vector specifying a subset of cases to include in the analysis

na.action

a function specifying the action to take for any cases missing any of ftime, fstatus, cov1, cov2, cengroup, or subset.

rq.method

method of computation for quantile regressions. (cf. documentation of method rq.fit in package quantreg for details.)

variance

if FALSE, then suppresses computation of asymptotic variances

x

crrQR object (output from crrQR()) for method print

...

included for compatibility with the generic functions. Not currently used.

Details

Fits the competing risks quantile regression model described in Peng and Fine (2009).

While the use of model formulas is not supported, the model.matrix function can be used to generate suitable matrices of covariates from factors, eg model.matrix(~factor1+factor2)[,-1] will generate the variables for the factor coding of the factors factor1 and factor2. The final [,-1] removes the constant term from the output of model.matrix.

If variance=FALSE, then some of the functionality in summary.crrQR and print.crrQR will be lost. This option can be useful in situations where crrQR is called repeatedly for point estimates, but standard errors are not required, such as in bootstrapping the cumulative incidence function for confidence intervals.

The print method prints the estimated coefficients, the estimated standard errors, and the two-sided p-values for the test of the individual coefficients equal to 0.

A first implementation of the estimation procedure was prepared by Limin Peng and Ruosha Li.

Value

Returns a list of class crrQR, with components

$beta.seq

the estimated regression coefficients

$tau.seq

the sequence of quantiles computed

$var.seq

estimated variance covariance matrix of coef

$inf.func

list of estimated influence functions

$call

the call to crr

$n

the number of observations used in fitting the model

$n.missing

the number of observations removed from the input data due to missing values

$cvt.length

number of covariates (columns of X)

References

Peng L and Fine JP (2009) Competing risks quantile regression. JASA 104:1440-1453.

See Also

predict.crrQR plot.predict.crrQR summary.crrQR rq.fit

Examples

# simulated data to test 
set.seed(10)
ftime <- rexp(200)
fstatus <- sample(0:2,200,replace=TRUE)
X <- matrix(runif(600),nrow=200)
dimnames(X)[[2]] <- c('x1','x2','x3')
#compute model
print(z <- crrQR(ftime,fstatus,X))
summary(z)
# predict and plot cumulative incedences
reference <- as.matrix(rbind(c(.1,.5,.8),c(.1,.5,.2)))
dimnames(reference)[[2]] <- c('x1','x2','x3') 
z.p <- predict(z,reference)
print(z.p)
plot(z.p,lty=1,color=2:3)
crrQR(ftime,fstatus,X,failcode=2)