Function plotCIF plots, for one or more groups,
the cumulative incidence curves for a selected event out of two or more
competing events. Function stackedCIF plots, for one group or population,
the cumulative incidence curves for two or more competing events such that
the cumulative incidences are stacked upon each other. The CIFs are
are estimated by the Aalen-Johansen method.
## S3 method for class 'survfit'
plotCIF( x, event = 1,
xlab = "Time",
ylab = "Cumulative incidence",
ylim = c(0, 1),
lty = NULL,
col = NULL, ... )## S3 method for class 'survfit'
stackedCIF( x, group = 1,
colour = NULL,
ylim = c(0, 1),
xlab = "Time",
ylab = "Cumulative incidence", ... )
An object of class survfit, the type
of event in Surv() being "mstate";
the first level of the event factor represents censoring and the
remaining ones the alternative competing events.
Determines the event for which the cumulative incidence curve is plotted by plotCIF.
An integer showing the selected level of a possible grouping factor
appearing in the model formula in survfit when plotting by stackedCIF
A vector specifying the plotting color(s) of the curve(s)
for the different groups in plotCIF-- default: all "black".
A vector indicating the colours to be
used for shading the areas pertinent to the separate outcomes in stackedCIF --
default: all "white".
Label for the $x$-axis.
Label for the $y$-axis.
Limits of the $y$-axis.
A vector specifying the line type(s) of the curve(s) for the different groups -- default: all 1 (=solid).
Further graphical parameters to be passed.
No value is returned but a plot is produced as a side-effect.
The order in which the curves with stackedCIF are piled
upon each other is the same as the ordering of the values or levels of
the competing events in the pertinent event variable. The ordering can
be changed by permuting the levels as desired using function
Relevel, after which survfit is called with the relevelled
event variable in Surv()
Putter, H., Fiocco, M., Geskus, R.B. (2007). Tutorial in biostatistics: competing risks and multi-state models. Statistics in Medicine, 26: 2389--2430.
# NOT RUN {
library(survival) # requires version 2.39-4 or later
head(mgus1)
# Aalen-Johansen estimates of CIF are plotted by sex for two
# competing events: (1) progression (pcm), and (2) death, in
# a cohort of patients with monoclonal gammopathy.
# The data are actually covering transitions from pcm to death, too,
# for those entering the state of pcm. Such patients have two rows
# in the data frame, and in their 2nd row the 'start' time is
# the time to pcm (in days).
# In our analysis we shall only include those time intervals with value 0
# for variable 'start'. Thus, the relevant follow-up time is represented
# by variable 'stop' (days). For convenience, days are converted to years.
fitCI <- survfit(Surv(stop/365.25, event, type="mstate") ~ sex,
data= subset(mgus1, start==0) )
par(mfrow=c(1,2))
plotCIF(fitCI, event = 1, col = c("red", "blue"),
main = "Progression", xlab="Time (years)" )
text( 38, 0.15, "Men", pos = 2)
text( 38, 0.4, "Women", pos = 2)
plotCIF(fitCI, event = 2, col = c("red", "blue"),
main = "Death", xlab="Time (years)" )
text( 38, 0.8, "Men", pos = 2)
text( 38, 0.5, "Women", pos = 2)
par(mfrow=c(1,2))
stackedCIF(fitCI, group = 1, colour = c("gray80", "gray90"),
main = "Women", xlab="Time (years)" )
text( 36, 0.15, "PCM", pos = 2)
text( 36, 0.6, "Death", pos = 2)
stackedCIF(fitCI, group = 2, colour = c("gray80", "gray90"),
main = "Men", xlab="Time (years)" )
text( 39, 0.10, "PCM", pos = 2)
text( 39, 0.6, "Death", pos = 2)
# }
Run the code above in your browser using DataLab