plot.survfit: Plot method for `survfit` objects

Description

A plot of survival curves is produced, one curve for each strata. The log=T option does extra work to avoid log(0), and to try to create a pleasing result. If there are zeros, they are plotted by default at 0.8 times the smallest non-zero value on the curve(s).

Curves are plotted in the same order as they are listed by print (which gives a 1 line summary of each). This will be the order in which col, lty, etc are used.

Usage

# S3 method for survfit
plot(x, conf.int=, mark.time=FALSE, 
 pch=3, col=1, lty=1, lwd=1, cex=1, log=FALSE, xscale=1, yscale=1,  
 xlim, ylim, xmax, fun, 
 xlab="", ylab="", xaxs="r",  conf.times, conf.cap=.005,
 conf.offset=.012,
 conf.type = c("log", "log-log", "plain", "logit", "arcsin"),
 mark, noplot="(s0)", cumhaz=FALSE,
 firstx, ymin, cumprob=FALSE, ...)

Value

a list with components x and y, containing the coordinates of the last point on each of the curves (but not the confidence limits). This may be useful for labeling. If cumprob=TRUE then y will be a matrix with one row per curve and x will be all the time points. This may be useful for adding shading.

Arguments

x: an object of class survfit, usually returned by the survfit function.
conf.int: determines whether pointwise confidence intervals will be plotted. The default is to do so if there is only 1 curve, i.e., no strata, using 95% confidence intervals Alternatively, this can be a numeric value giving the desired confidence level.
mark.time: controls the labeling of the curves. If set to FALSE, no labeling is done. If TRUE, then curves are marked at each censoring time. If mark is a numeric vector then curves are marked at the specified time points.
pch: vector of characters which will be used to label the curves. The points help file contains examples of the possible marks. A single string such as "abcd" is treated as a vector c("a", "b", "c", "d"). The vector is reused cyclically if it is shorter than the number of curves. If it is present this implies mark.time = TRUE.
col: a vector of integers specifying colors for each curve. The default value is 1.
lty: a vector of integers specifying line types for each curve. The default value is 1.
lwd: a vector of numeric values for line widths. The default value is 1.
cex: a numeric value specifying the size of the marks. This is not treated as a vector; all marks have the same size.
log: a logical value, if TRUE the y axis wll be on a log scale. Alternately, one of the standard character strings "x", "y", or "xy" can be given to specific logarithmic horizontal and/or vertical axes.
xscale: a numeric value used like yscale for labels on the x axis. A value of 365.25 will give labels in years instead of the original days.
yscale: a numeric value used to multiply the labels on the y axis. A value of 100, for instance, would be used to give a percent scale. Only the labels are changed, not the actual plot coordinates, so that adding a curve with "lines(surv.exp(...))", say, will perform as it did without the yscale argument.
xlim,ylim: optional limits for the plotting region.
xmax: the maximum horizontal plot coordinate. This can be used to shrink the range of a plot. It shortens the curve before plotting it, so that unlike using the xlim graphical parameter, warning messages about out of bounds points are not generated.
fun: an arbitrary function defining a transformation of the survival (or probability in state, or cumulative hazard) curves. For example fun=log is an alternative way to draw a log-survival curve (but with the axis labeled with log(S) values), and fun=sqrt would generate a curve on square root scale. Four often used transformations can be specified with a character argument instead: "S" gives the usual survival curve, "log" is the same as using the log=T option, "event" or "F" plots the empirical CDF \(F(t)= 1-S(t)\) (f(y) = 1-y), and "cloglog" creates a complimentary log-log survival plot (f(y) = log(-log(y)) along with log scale for the x-axis). The terms "identity" and "surv" are allowed as synonyms for type="S". The argument "cumhaz" causes the cumulative hazard function to be plotted.
xlab: label given to the x-axis.
ylab: label given to the y-axis.
xaxs: either "S" for a survival curve or a standard x axis style as listed in par; "r" (regular) is the R default. Survival curves have historically been displayed with the curve touching the y-axis, but not touching the bounding box of the plot on the other 3 sides, Type "S" accomplishes this by manipulating the plot range and then using the "i" style internally. The "S" style is becoming increasingly less common, however.
conf.times: optional vector of times at which to place a confidence bar on the curve(s). If present, these will be used instead of confidence bands.
conf.cap: width of the horizontal cap on top of the confidence bars; only used if conf.times is used. A value of 1 is the width of the plot region.
conf.offset: the offset for confidence bars, when there are multiple curves on the plot. A value of 1 is the width of the plot region. If this is a single number then each curve's bars are offset by this amount from the prior curve's bars, if it is a vector the values are used directly.
conf.type: One of "none", "plain", "log" (the default), "log-log" or "logit". Only enough of the string to uniquely identify it is necessary. The first option causes confidence intervals not to be generated. The second causes the intervals curve +- k *se(curve), where k is determined from conf.int. The log option calculates intervals based on the cumulative hazard or log(survival). The log-log option bases the intervals on the log hazard or log(-log(survival)), and the logit option on log(survival/(1-survival)) The default is "plain" for a plot of the cumulative hazard.
mark: a historical alias for pch
noplot: for multi-state models, curves with this label will not be plotted. (Also see the istate0 argument in survcheck.)
cumhaz: plot the cumulative hazard rather than the probability in state or survival. Optionally, this can be a numeric vector specifying which columns of the cumhaz component to plot.
ymin: this will normally be given as part of the ylim argument
firstx: this will normally be given as part of the xlim argument.
cumprob: for a multi-state curve, plot the probabilities in state 1, (state1 + state2), (state1 + state2 + state3), .... If cumprob is an integer vector the totals will be in the order indicated.
...: other arguments that will be passed forward to the underlying plot method, such as xlab or ylab.

Details

If the object contains a cumulative hazard curve, then fun='cumhaz' will plot that curve, otherwise it will plot -log(S) as an approximation. Theoretically, S = \(\exp(-\Lambda)\) where S is the survival and \(\Lambda\) is the cumulative hazard. The same relationship holds for estimates of S and \(\Lambda\) only in special cases, but the approximation is often close.

When the survfit function creates a multi-state survival curve the resulting object also has class `survfitms'. Competing risk curves are a common case. In this situation the fun argument is ignored.

When the conf.times argument is used, the confidence bars are offset by conf.offset units to avoid overlap. The bar on each curve are the confidence interval for the time point at which the bar is drawn, i.e., different time points for each curve. If curves are steep at that point, the visual impact can sometimes substantially differ for positive and negative values of conf.offset.

Examples

Run this code

leukemia.surv <- survfit(Surv(time, status) ~ x, data = aml) 
plot(leukemia.surv, lty = 2:3) 
legend(100, .9, c("Maintenance", "No Maintenance"), lty = 2:3) 
title("Kaplan-Meier Curves\nfor AML Maintenance Study") 
lsurv2 <- survfit(Surv(time, status) ~ x, aml, type='fleming') 
plot(lsurv2, lty=2:3, fun="cumhaz", 
	xlab="Months", ylab="Cumulative Hazard")