quantileKM: Compute arbitrary quantile of a Kaplan-Meier estimate of a...

In biostatUZH: Misc Tools of the Department of Biostatistics, EBPI, University of Zurich

Description

In survival analysis, often some quantiles (mainly the median) of an estimated survival function are of interest. The survfit function in the 'survival' packages of version < 2.35 computed median time to event. However, the output of that function was such that the median is not accessible by the user (it can only be read off the output). This function makes the median and any other quantile accessible by the user.

Usage

quantileKM(time, event, group = NA, quant = 0.5, conf.level = 0.95, 
    conftype = c("log", "log-log", "plain", "none")[2], conflower = 
    c("usual", "peto", "modified")[1])

Arguments

time	Event times, censored or observed.
event	Censoring indicator, 1 for event, 0 for censored.
group	Quantiles can be computed for several survival curves, defined by group.
quant	Quantile to be computed. Real number in [0, 1].
conf.level	Significance level for confidence interval for the time to event quantile.
conftype	Type of confidence interval to be computed. For possible choices see above, and for specifications regarding the different options confer the help file of the function survfit.
conflower	Controls modified lower limits to the curve, the upper limit remains unchanged. See survfit for details.

Value

n	Number of observations used.
events	Number of events.
quantile	Quantile estimate.
lower.ci	Lower limit of confidence interval.
upper.ci	Upper limit of confidence interval.

Author(s)

Kaspar Rufibach
[e-mail]

References

Computation of confidence intervals is done according to

Brookmeyer, R. and Crowley, J. (1982). A Confidence Interval for the Median Survival Time. Biometrics, 38, 29–41.

See Also

Partly based on the function survfit.

Examples

## use Acute Myelogenous Leukemia survival data contained in package 'survival'
time <- leukemia[, 1]; status <- leukemia[, 2]; x <- as.factor(leukemia[, 3])
plot(survfit(Surv(time, status) ~ x, conf.type = "none"), mark = "/", col = 1:2)

## median time to event
qKM <- quantileKM(time, status, group = x, quant = 0.5, conf.level = 0.95, 
    conftype = "log")
qKM

## extract results
qKM$quantities

## comparison to standard function (median time to event not accessible by user)
quant.surv <- survfit(Surv(time, status) ~ x, conf.int = 0.95)
quant.surv

## compute 0.25 quantile
qKM2 <- quantileKM(time, status, group = x, quant = 0.25, conf.level = 0.95, 
    conftype = "log")
qKM2

biostatUZH documentation built on May 2, 2019, 6:06 p.m.