pairtest: Paired Two-Sample Tests for Survival Endpoints

In umich-biostatistics/corrsurv: Two-Sample Tests for Correlated Survival Endpoints

Description

Perform two-sample tests for treatment effects with paired censored survival data. The method requires the time to event or a censored observation and the event indicator for each pair of observations.

Usage

pairtest(
  x1,
  delta1,
  x2,
  delta2,
  n,
  tm = sort(unique(c(0, x1, x2))),
  weights = "left",
  maxtau = 1e+05
)

Arguments

x1	numeric, time to event for treatment group 1
delta1	numeric, 1 if subject died, 0 if censored
x2	numeric, time to event for treatment group 2
delta2	numeric, 1 if subject died, 0 if censored
n	the integer number of correlated individuals and we assume they are sorted as in the paper, so that correlated observations come first up to the nth observation and then uncorrelated observations follow.
tm	sorted vector of times to events
weights	string, options are "left" and "right". Option "left" will give pf and yls weighting using upperlimit similar to logrank. "right" will use weights that are similar, but right continuous.
maxtau	numeric, maximum tau cutoff

Value

a list object which contains named two-sample statistics and p-values for the independent and correlated cases, and other auxiliary information relevant to the analysis.

References

Murray, Susan. Nonparametric Rank-Based Methods for Group Sequential Monitoring of Paired Censored Survival Data. 2000. Biometrics, 56, pp. 984-990.

See Also

TM(), TM2()

Examples

data(pairdata)
eyeresults = pairtest(x1 = pairdata$x1, delta1 = pairdata$delta1, 
                      x2 = pairdata$x2, delta2 = pairdata$delta2, n = 3711)
summary(eyeresults)

umich-biostatistics/corrsurv documentation built on Jan. 11, 2020, 2:03 a.m.