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.
Data
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"
Single-Arm Plot
We start by creating a KMunicate-style plot for all study subjects.
First, we load the package:
library(KMunicate)
#> Loading required package: survival
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:
ts <- seq(0, max(brcancer$rectime), length.out = 5)
ts
#> [1] 0.00 664.75 1329.50 1994.25 2659.00
Finally, we can pass fit and ts to the
KMunicate() function to simply obtain a plot:
KMunicate(fit = fit, time_scale = ts)
Multiple-Arms 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:
KMunicate(fit = fit2, time_scale = ts)
Customisation
The KMunicate() function contains a few options to
customise the produced plot. First, we can pass a ggplot2
theme to the .theme argument:
KMunicate(fit = fit2, time_scale = ts, .theme = ggplot2::theme_minimal())
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:
KMunicate(fit = fit2, time_scale = ts, .alpha = 0.1)
We can customise the label of the horizontal axis:
KMunicate(fit = fit2, time_scale = ts, .xlab = "New Label (Time in Days, Actually)")