Heller Explained Relative Risk — coxphERR • clinfun

Calculates the contribution of a subset of covariates to the explained relative risk derived from the full Cox proportional hazards model.

Usage

coxphERR(phfit, ngamma=NULL)

Arguments

phfit: The output from a proportional hazards fit.
ngamma: A vector of indices corresponding to covariates of interest. If missing (default), the explained relative risk is computed for the full model.

Value

The function coxphERR returns the vector (ERR, se.ERR). The first component ERR represents the contribution of a subset of covariates to the explained relative risk estimate of the full model. If a set of covariates is not provided, then it computes the estimate of the full model. The second component se.ERR is the standard error of the estimate.

Details

The object phfit should be the result of a call to coxph with the option x=TRUE.

References

Heller G. (2012) A measure of explained risk in the proportional hazards model. Biostatistics

Examples

if (FALSE) {
 library(survival)
 ovarianfit <- coxph(Surv(futime, fustat) ~ age + resid.ds + rx +
                                           ecog.ps, data=ovarian,x=T) 
 # Compute the explained relative risk (ERR) and
 # its standard error (se.ERR) for the full model.
 coxphERR(ovarianfit)       
 # Compute the contribution of age and ECOG performance status to
 # the explained relative risk. Age and ECOG performance status are 
 # the first and fourth covariates in the model.
 coxphERR(ovarianfit, c(1,4))
}