The {KMunicate} package can be used to produce Kaplan-Meier plots in the style recommended following the KMunicate study by TP Morris et al. (2019).
In this vignette, we’ll see learn how to produce such a plot using the {KMunicate} package.
We will be using the brcancer
dataset, which comes
bundled with {KMunicate}:
data("brcancer", package = "KMunicate")
str(brcancer)
#> Classes 'tbl_df', 'tbl' and 'data.frame': 686 obs. of 14 variables:
#> $ id : num 1 2 3 4 5 6 7 8 9 10 ...
#> $ hormon : num 0 1 1 1 0 0 1 0 0 0 ...
#> $ x1 : num 70 56 58 59 73 32 59 65 80 66 ...
#> $ x2 : num 2 2 2 2 2 1 2 2 2 2 ...
#> $ x3 : num 21 12 35 17 35 57 8 16 39 18 ...
#> $ x4 : num 2 2 2 2 2 3 2 2 2 2 ...
#> $ x5 : num 3 7 9 4 1 24 2 1 30 7 ...
#> $ x6 : num 48 61 52 60 26 0 181 192 0 0 ...
#> $ x7 : num 66 77 271 29 65 13 0 25 59 3 ...
#> $ rectime: num 1814 2018 712 1807 772 ...
#> $ censrec: num 1 1 1 1 1 1 0 0 1 0 ...
#> $ x4a : num 1 1 1 1 1 1 1 1 1 1 ...
#> $ x4b : num 0 0 0 0 0 1 0 0 0 0 ...
#> $ x5e : num 0.698 0.432 0.34 0.619 0.887 ...
#> - attr(*, "label")= chr "German breast cancer data"
We start by creating a KMunicate-style plot for all study subjects. First, we load the package:
Then, we fit a Kaplan-Meier curve using the survfit
function from the {survival} package:
fit <- survfit(Surv(rectime, censrec) ~ 1, data = brcancer)
fit
#> Call: survfit(formula = Surv(rectime, censrec) ~ 1, data = brcancer)
#>
#> n events median 0.95LCL 0.95UCL
#> [1,] 686 299 1807 1587 2030
Then, we need to define the horizontal axis of the plot. For instance, we can define 5 equally-spaced breaks between time zero and the largest observed time:
Finally, we can pass fit
and ts
to the
KMunicate()
function to simply obtain a plot:
A multiple-arms plot will be automatically produced if the
survfit
object has covariates in it. For instance, if we
fit the Kaplan-Meier estimator by treatment arm:
fit2 <- survfit(Surv(rectime, censrec) ~ hormon, data = brcancer)
fit2
#> Call: survfit(formula = Surv(rectime, censrec) ~ hormon, data = brcancer)
#>
#> n events median 0.95LCL 0.95UCL
#> hormon=0 440 205 1528 1296 1814
#> hormon=1 246 94 2018 1918 NA
We will be using the same time scale as before (ts
), and
the call to KMunicate()
is analogous:
The KMunicate()
function contains a few options to
customise the produced plot. First, we can pass a ggplot2
theme to the .theme
argument:
We can also pass custom colour (and fill) scales:
KMunicate(
fit = fit2,
time_scale = ts,
.color_scale = ggplot2::scale_color_brewer(type = "qual", palette = "Set2"),
.fill_scale = ggplot2::scale_fill_brewer(type = "qual", palette = "Set2")
)
We can customise the transparency of the point-wise confidence
intervals via the .alpha
argument:
We can customise the label of the horizontal axis:
We can customise the relative size of risk tables and plots, although the default should generally work fine for most scenarios:
Assuming you have set up your computer and R session to support
custom fonts (e.g. using a ragg
graphics device), you can produce plots with custom fonts via the
.ff
argument of the KMunicate()
function.
Examples of using ragg
can be found here.
We can combine all the above to obtain an (arguably) much better plot:
As of version 0.2.0, {KMunicate} provide additional arguments to customise the final plot.
You can customise the size of each Kaplan-Meier curve via the
.size
argument:
You can customise the linetype scale via the
.linetype_scale
argument:
KMunicate(
fit = fit2,
time_scale = ts,
.linetype_scale = ggplot2::scale_linetype_manual(values = c("dotted", "dashed"))
)
You can customise the location of the legend via the
.legend_position
argument, or you can even suppress it
fully:
You can add custom annotations via the .annotate
argument:
KMunicate(
fit = fit2,
time_scale = ts,
.annotate = ggplot2::annotate(geom = "text", x = 365, y = 0.5, label = "Some annotation")
)
You can add customise labels and title with the .xlab
,
.ylab
, and .title
arguments: