Details of a Cox Model Fit

Returns the individual contributions to the first and second derivative matrix, at each unique event time.

Usage

coxph.detail(object, riskmat=FALSE, rorder=c("data", "time"))

Arguments

object

a Cox model object, i.e., the result of coxph.

riskmat

include the at-risk indicator matrix in the output?

rorder

should the rows of x, y and riskmat be returned in the original data order, or sorted by time within strata.

Value

a list with components

time

the vector of unique event times

nevent

the number of events at each of these time points.

means

a matrix with one row for each event time and one column for each variable in the Cox model, containing the weighted mean of the variable at that time, over all subjects still at risk at that time. The weights are the risk weights exp(x %*% fit$coef).

nrisk

number of subjects at risk.

score

the contribution to the score vector (first derivative of the log partial likelihood) at each time point.

imat

the contribution to the information matrix (second derivative of the log partial likelihood) at each time point.

hazard

the hazard increment. Note that the hazard and variance of the hazard are always for some particular future subject. This routine uses object$means as the future subject.

varhaz

the variance of the hazard increment.

x,y

copies of the input data.

strata

only present for a stratified Cox model, this is a table giving the number of time points of component time that were contributed by each of the strata.

wtrisk

the weighted number at risk

riskmat

a matrix with one row for each observation and one colum for each unique event time, containing a 0/1 value to indicate whether that observation was (1) or was not (0) at risk at the given time point. Rows are in the order of the original data (after removal of any missings by coxph), or in time order.

Details

This function may be useful for those who wish to investigate new methods or extensions to the Cox model. The example below shows one way to calculate the Schoenfeld residuals.

See also

coxph, residuals.coxph

Examples

fit   <- coxph(Surv(futime,fustat) ~ age + rx + ecog.ps, ovarian, x=TRUE)
fitd  <- coxph.detail(fit)
#  There is one Schoenfeld residual for each unique death.  It is a
# vector (covariates for the subject who died) - (weighted mean covariate
# vector at that time).  The weighted mean is defined over the subjects
# still at risk, with exp(X beta) as the weight.

events <- fit$y[,2]==1
etime  <- fit$y[events,1]   #the event times --- may have duplicates
indx   <- match(etime, fitd$time)
schoen <- fit$x[events,] - fitd$means[indx,]