Bug Report: `add_quantile()` discrepancy compared to `survival::quantile.survfit()` when there is a plateau at the specified quantile

[KennethATaylor]

Problem

There is an inconsistency in how the median survival time is calculated between cardx::ard_survival_survfit() and gtsummary::tbl_survfit() compared to the median from survival::quantile.survfit() when there is a plateau in survival at the specified quantile.

Cause

The issue is occurring because cardx::ard_survival_survfit() and gtsummary::tbl_survfit() both rely on survival::quantile.survfit() to estimate the median, while ggsurvfit::add_quantile() uses its own internal logic based on the values plotted to add the reference line to the plot.

According to the documentation for survival::quantile.survfit():

The kth quantile for a survival curve S(t) is the location at which a horizontal line at height p= 1-k intersects the plot of S(t). Since S(t) is a step function, it is possible for the curve to have a horizontal segment at exactly 1-k, in which case the midpoint of the horizontal segment is returned. This mirrors the standard behavior of the median when data is uncensored...

When a horizontal segment of the survival curve exactly matches one of the requested quantiles the returned value will be the midpoint of the horizontal segment; this agrees with the usual definition of a median for uncensored data...

Because of the discrepancy, users who want to use add_quantile() when there is a plateau at the specified value will get a discrepancy between values generated via other common functions despite using feeding in the exact same survfit object.

Reproducible Example

In the example below, I show the discrepancy between the quantile value plotted for the median from add_quantile() compared to the median reported via cardx::ard_survival_survfit() and gtsummary::tbl_survfit() (which both use survival::quantile.survfit()).

I also show an example plot with a reference line when pulling the median from quantile.survfit() (what most users would probably expect given the behavior of the other functions.

# Load all required libraries
library(survival)
library(ggsurvfit)
#> Loading required package: ggplot2
library(ggplot2)
library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union

# --- Create a dataset with a plateau at S(t) = 0.5 ---
df_reprex <- data.frame(
  time = c(1, 2, 3, 4, 5, 10, 12, 12, 12, 12),
  status = c(1, 1, 1, 1, 1, 1, 0, 0, 0, 0)
)

# --- Create the survfit object ---
fit_reprex <- survfit(Surv(time, status) ~ 1, data = df_reprex)

# --- Get the `quantile.survfit()` result ---
median_interpolated <- quantile(fit_reprex, probs = 0.5)$quantile %>%
  tibble::enframe(value = "median_time")

print(paste("Median from quantile.survfit()", median_interpolated$median_time))
#> [1] "Median from quantile.survfit() 7.5"

# --- Plot with ggsurvfit::add_quantile() ---
plot_add_quantile <- ggsurvfit(fit_reprex) +
  add_quantile() +
  scale_ggsurvfit() +
  labs(
    title = "add_quantile() uses earliest time at 0.5",
    subtitle = "Note: Median line is at t = 5"
  )

# --- Manually plot the `quantile.survfit()` result ---
# This plots the median generated from quantile.survfit()
plot_quantile_survfit <- ggsurvfit(fit_reprex) +
  scale_ggsurvfit() +
  # Manually add segments using the median from `quantile.survfit()`
  geom_segment(
    data = median_interpolated,
    aes(x = 0, xend = median_time, y = 0.5, yend = 0.5),
    linetype = 2, color = "red"
  ) +
  geom_segment(
    data = median_interpolated,
    aes(x = median_time, xend = median_time, y = 0.5, yend = 0),
    linetype = 2, color = "red"
  ) +
  labs(
    title = "quantile.survfit() uses midpoint of the plateau for median",
    subtitle = paste0("Note: Red dashed line is at t = ", median_interpolated$median_time)
  )

# --- Show the conflicting results ---
plot_add_quantile

plot_quantile_survfit

# --- Value from gtsummary::tbl_survfit() / cardx::ard_survival_survfit()
cardx::ard_survival_survfit(fit_reprex, probs = 0.5) %>% 
  filter(stat_name == "estimate")
#> {cards} data frame: 1 x 9
#>   variable variable_level   context stat_name stat_label stat
#> 1     prob            0.5 survival…  estimate  Survival…  7.5
#> ℹ 3 more variables: fmt_fun, warning, error


gtsummary::tbl_survfit(fit_reprex, probs = 0.5) %>% 
  gtsummary::as_kable()
#> Warning in attr(x, "align"): 'xfun::attr()' is deprecated.
#> Use 'xfun::attr2()' instead.
#> See help("Deprecated")
#> Warning in attr(x, "format"): 'xfun::attr()' is deprecated.
#> Use 'xfun::attr2()' instead.
#> See help("Deprecated")

Characteristic	50% Percentile
Overall	7.5 (3.0, —)

# Median survival calculated from cardx::ard_survival_survfit() - and by extension, gtsummary::tbl_survfit()
#    conflicts with the value used by ggsurvplot::add_quantile()

^{Created on 2025-11-10 with reprex v2.1.1}

[ddsjoberg]

Thanks for the heads up. We'll add it to the list!

[KennethATaylor]

Got some extra time and motivation this weekend to circle back on this issue. Submitted a PR with the fix!

Cheers! 🍻