survPen (version 2.0.1)

pwcst: Defining piecewise constant (excess) hazard in survPen formulae

Description

Used inside a formula object to define a piecewise constant (excess) hazard. This is useful since it triggers an explicit calculation of cumulative hazard calculation (much more efficient and more precise than Gauss-Legendre quadrature when hazard is constant). The breaks given are used to defined sub-intervals that are left-open (except the first interval which is always left-closed) and right-closed. Internally, this constructor uses the cut function on the follow-up time with options include.lowest=TRUE and right=TRUE Important : this function must not be used with other time-dependent effect functions because the Gauss-Legendre quadrature will not operate correctly. If you really want to fit such a model, please use the cut function with the time variable as an argument to fit a piecewise constant hazard (and do not forget to use a huge number of Gauss-Legendre quadrature nodes, typically n.legendre=500)

Usage

pwcst(breaks)

Value

object of class pwcst.spec

pwcst.breaks

numeric vector that specifies the boundaries of each sub-interval on which the hazard is constant

Arguments

breaks

numeric vector that specifies the boundaries of each sub-interval on which the hazard is constant

Examples

library(survPen)

data(datCancer)

# piece constant hazard on 6 sub-intervals : [0;0.5]; ]0.5;1]; ]1;2]; ]2;3]; ]3;4]; ]4;5]
formula <- ~pwcst(breaks=c(0,0.5,1,2,3,4,5))
mod <- survPen(formula,t1=fu,event=dead,data=datCancer)

# The same but in an inefficient way
formula2 <- ~cut(fu,breaks=c(0,0.5,1,2,3,4,5),include.lowest=TRUE,right=TRUE)
mod.inefficient <- survPen(formula2,t1=fu,event=dead,data=datCancer,n.legendre=500)