KM plot show quantiles

[SHAESEN2]

Add drop lines to show quantiles of the KM distribution.

[SHAESEN2]

Also allow to add horizontal line for median survival using a wrapper around geom_hline(yintercept=xx, linetype = xx)

[timtreis]

We used this as an "exercise to the reader" in the PSI2021 workshop, so we prepared a minimal solution for this. Just dropping this here as a starting point

m = adtte %>% estimate_KM(strata = "SEX")
half_surv_df = m %>% 
  get_quantile(probs = 0.5) %>% 
  dplyr::filter(quantity == "quantile") %>%
  dplyr::mutate(pos1 = `50`, pos2 = 0.5) %>%
  dplyr::select(-c(quantity, `50`))

half_surv_df_horizontal = half_surv_df %>% dplyr::mutate(pos1 = 0)

half_surv_df_vertical = half_surv_df %>% dplyr::mutate(pos2 = 0)

half_surv_df = rbind(half_surv_df, half_surv_df_horizontal, half_surv_df_vertical)  

m %>%
  visr() %>%
  add_CI() %>%
  add_CNSR() +
  ggplot2::geom_line(data = half_surv_df, 
                     ggplot2::aes(pos1, pos2, group = strata), 
                     linetype = "dashed")

[timtreis]

@SHAESEN2 I'll probably have some time during R/Pharma 2021 to implement this, doesn't feel like a lot of work. Do you already have a name in mind? `add_survival_quantiles()' with "half" as the default?

[timtreis]

Unsorted ideas:

• lines hitting the x-axis in the colour of the strata?
• horizontal lines solid?

[SHAESEN2]

add_quantiles and make an S3 method out of it => add_quantiles uses get_quantiles.
I like the dotted lines you show in the example and the line color as per strata.

[timtreis]

library(visR)
library(ggplot2)

theme_set(theme_bw())

adtte %>%
  visR::estimate_KM("SEX") %>%
  visR::visr() %>%
  visR::add_quantiles()

What do you think? The default behaviour is to pick linetype and colour separately for horizontal and vertical lines but are exposed as arguments accepting normal values.

library(visR)
library(ggplot2)

theme_set(theme_bw())

adtte %>%
  visR::estimate_KM("SEX") %>%
  visR::visr() %>%
  visR::add_quantiles(quantiles = c(0.25, 0.5))

library(visR)
library(ggplot2)

theme_set(theme_bw())

adtte %>%
  visR::estimate_KM("SEX") %>%
  visR::visr() %>%
  visR::add_quantiles(quantiles = c(0.25, 0.5),
                      linetype = "dashed",
                      linecolour = "black")

[SHAESEN2]

I think I prefer a default of dotted lines. I asked @AlexandraP-21 to look into it as well. She has a good eye for those things.

[AlexandraP-21]

I took a look at the lines and like them dotted. However, I think that maybe (light) grey (and not black) would be better. The coloring ones I found them a bit confusing and distracting to be honest. Hope this helps!