simple way of break.time.by in ggsurvfit?

[andrewjyee]

In ggsurvplot, break.time.by is a great, simple option. To indicate x axis intervals one month apart, it's

ggsurvplot(result, break.time.by=1)

In ggsurvfit, it seems more complicated and requires scale_ggsurvfit()

ggsurvfit(result) +
  scale_ggsurvfit(x_scales = list(breaks = seq(0, 12, by = 1)))

Is there an easier way of doing this?

[ddsjoberg]

Hi @andrewjyee ,

You don't need to use scale_ggsurvfit() if you don't want to. It's just a wrapper to ggplot2::scale_x_continuous() and ggplot2::scale_y_continuous() with common scale changes for survival plots.

If it's easier for you, you can just call ggplot2::scale_x_continuous() directly to change the x-axis in any way you'd like.

[andrewjyee]

Hi I like the simplicity of scale_x_continuous(n.breaks=6)

But I also like how scale_ggsurvfit() seems to automatically create the y axis in my preferred format where it goes from 0-100% and has the probability in percent. If I use scale_x_continuous(n.breaks=6), the y axis starts at the lowest value which seems non standard for a KM curve.

[ddsjoberg]

Scales are tricky in ggplot, because they don't "add" like the other elements in ggplots. So if you call scale_ggsurvfit() and then scale_x_continuous(n.breaks=6) all the scale changes from the first call will be removed to add the scale updates from the second. That is why we stick it all in one function for you. I understand that the syntax in the other package is simpler for this modification. But I am cautious to change the syntax from the syntax we see in ggplot, which makes it easy to make the all the other changes that are possible in the ggplot scales (of which, there are sooooooo many).