In raredd/cmprsk2: cmprsk2

Subdistribution Analysis of Competing Risks

• intro
• design
• examples
• install

library('cmprsk2')
library('htmlTable')

intro

Extensions to the cmprsk package:

1. estimation, testing, and regression modeling of subdistribution functions in competing risks, as described in Gray (1988);

2. A class of K-sample tests for comparing the cumulative incidence of a competing risk, Ann. Stat. 16:1141-1154, and Fine JP and Gray RJ (1999); and

3. A proportional hazards model for the subdistribution of a competing risk, JASA, 94:496-509.

design

The two main package functions, cuminc2 and crr2, provide a convenient formula-interface to cmprsk::cuminc and cmprsk::crr to users who are familiar with using survival objects in the survival package.

The syntax for creating survival objects is

$$Surv(time, status) \sim rhs$$ Similarly, the syntax for creating cmprsk objects is

$$Surv(time, status(censor) == event) \sim rhs$$ where status is a categorical variable with censoring indicators, usually "0", "Censor", "Censored", etc., and two or more unique event indicators for each competing risk. event in the formula will be one of the event indicators found in status and will be considered the primary event of interest.

examples

cumulative incidence

Note in the formula that it is not necessary to specify the event of interest for cuminc2.

ci <- cuminc2(Surv(futime, event(censored)) ~ sex, transplant)
summary(ci)

cmprsk::timepoints(ci$cuminc, times = 0:4 * 500)
cmprsk2::timepoints2(ci)
cmprsk2::timepoints2(
  ci, sd = TRUE, html = TRUE,
  htmlArgs = list(caption = 'Table 1: Timepoints.')
)

plot(ci)

ci <- cuminc2(Surv(futime, event(censored)) ~ abo, transplant)
cuminc_pairs(ci)$p.value

plot(ci, split = 'event')

competing risks regression

While it is necessary to specify the event of interest in the crr2 formula, this will be the primary event of interest. That is, by default models are fit for all competing events, but the primary event model will be listed first.

Using crr2(..., which = event) will only fit model(s) for event.

cr <- crr2(Surv(futime, event(censored) == death) ~ sex, transplant)
summary(cr)


cr <- crr2(Surv(futime, event(censored) == death) ~ sex + abo, transplant)
summary(cr, n = TRUE, ref = TRUE)


cr <- crr2(Surv(futime, event(censored) == ltx) ~ sex + abo, transplant,
           cox = Surv(futime, event == 'death') ~ sex + abo)
summary(
  cr, html = TRUE, n = TRUE, ref = TRUE,
  htmlArgs = list(
    caption = 'Table 1: Competing risks.',
    rnames = with(transplant, c(levels(sex), levels(abo)))
  )
)


cr <- crr2(Surv(futime, event(censored) == ltx) ~ sex + abo, transplant,
           which = 'ltx')
summary(
  cr, html = TRUE, n = TRUE,
  htmlArgs = list(
    caption = 'Table 1: Competing risks.',
    tfoot = '<font size=1>Reference groups: Sex, male; blood type, A.</font>',
    rgroup = c('Sex', 'Blood type'),
    rnames = c('Female', paste('Type:', c('B', 'AB', 'O')))
  )
)

install

# install.packages('devtools')
devtools::install_github('raredd/cmprsk2', build_vignettes = TRUE)

session

utils::sessionInfo()

raredd/cmprsk2 documentation built on March 29, 2024, 5:34 a.m.