groupci: make calibration curve for competing risks endpoint
In jixccf/QHScrnomo: Construct nomograms for competing risk models

Description Usage Arguments Details Value Note Author(s) See Also Examples

Cumulative Incidence Estimates vs. a Continuous Variable

groupci(
  x,
  ftime,
  fstatus,
  cencode = 0,
  failcode = 1,
  ci = TRUE,
  m = 50,
  g,
  cuts,
  u,
  pl = TRUE,
  conf.int = 0.95,
  xlab,
  ylab,
  xlim = c(0, 1),
  ylim = c(0, 1),
  lty = 1,
  add = FALSE,
  cex.subtitle = FALSE,
  ab = TRUE,
  ...
)

`x`	a continuous variable
`ftime`	vector of follow-up time
`fstatus`	vector of failure status
`cencode`	value indicating cencering.
`failcode`	value indicating event of interest
`ci`	logical flag to output event free probability if setting `FALSE`
`m`	desired minimum number of observations in a group
`g`	number of quantile groups
`cuts`	actual cuts in `x`, e.g. `c(0,1,2)` to use [0,1), [1,2].
`u`	time for which to estimate cumulative incidence
`pl`	TRUE to plot results
`conf.int`	defaults to `.95` for 0.95 confidence bars. Set to `FALSE` to suppress bars
`xlab`	if `pl=TRUE`, is x-axis label. Default is `label(x)` or name of calling argument
`ylab`	if `pl=TRUE`, is y-axis label. Default is constructed from `u` and time `units`
`xlim`	range of x axis
`ylim`	range of y axis
`lty`	line time for primary line connecting estimates
`add`	set to `TRUE` if adding to an existing plot
`cex.subtitle`	character size for subtitle. Default is `.7`. Use `FALSE` to suppress subtitle.
`ab`	`TRUE` to add a 45 degree line
`...`	plotting parameters to pass to the plot and errbar functions

Function to divide a continuous variable x (e.g. age, or predicted cumulative incidence at time u created by predict.cmprsk into g quantile groups, get cumulative incidence estimates at time u (a scaler), and to return a matrix with columns x=mean x in quantile, n=number of subjects, events=no. events, and ci= cumulattive incidence at time u, std.err = standard error. Instead of supplying g, the user can supply the minimum number of subjects to have in the quantile group (m, default=50). If cuts is given (e.g. cuts=c(0,.1,.2,...{},.9,.1)), it overrides m and g.

matrix with columns named x (mean predictor value in interval), n (sample size in interval), events (number of events in interval), ci (cumulative incidence estimate), std.err (standard error of cumulative incidence)

This function is adapted from Harrell's function groupkm.

Changhong Yu, Michael Kattan, Ph.D
Department of Quantitative Health Sciences
Cleveland Clinic

groupkm, cuminc,pred.ci

data(prostate.dat)
dd <- datadist(prostate.dat)
options(datadist = "dd")
prostate.f <- cph(Surv(TIME_EVENT,EVENT_DOD == 1) ~ TX  + rcs(PSA,3) +
           BX_GLSN_CAT +  CLIN_STG + rcs(AGE,3) +
           RACE_AA, data = prostate.dat,
           x = TRUE, y= TRUE, surv=TRUE,time.inc = 144)
prostate.crr <- crr.fit(prostate.f,cencode = 0,failcode = 1)

## ten fold cross validation
prostate.dat$preds.tenf.cv.prostate.crr.120 <-
                                       tenf.crr(prostate.crr,time = 120)

with(prostate.dat,
     groupci(preds.tenf.cv.prostate.crr.120 , ftime = TIME_EVENT,
             fstatus =EVENT_DOD, g = 5, u = 120,
             xlab = "Nomogram predicted 10-year cancerspecific mortality",
             ylab = "Observed predicted 10-year cancerspecific mortality")
)