Skip to contents

Predicted median survival time

predict_surv_time.Rd

Get the survival time at a given probability of survival. A convenience wrapper for survival::quantile.survfit.

Usage

predict_surv_time(
  object,
  df = NULL,
  probs = 0.5,
  conf.level = 0.95,
  scale = 1,
  ...
)

Arguments

object

An object of either

  • survival::coxph, or

  • Survival formula using survival::Surv

df

A data frame in which to interpret the variables in object.

  • For object = survival::coxph, may be NULL or the usual newdata style argument.

  • For object = formula, must be the data frame used for fitting in survival::survfit.

probs

A numeric vector of probabilities. This argument applies to the cumulative distribution function F(t) = 1-S(t). For example, prob = 0.9 will return the survival time at the point of 10% probability of surviving.

conf.level

The level for a two-sided confidence interval on the survival time(s). Default is 0.95.

scale

A numeric value to rescale the survival time, e.g., if the input data to survfit were in days, scale = 365.25 would scale the output to years.

...

Optional arguments passed to survival::survfit.

Value

data.frame

Details

If the survival curve or its confidence limits do not cross the probability of interest (i.e. they extend out to infinity time without crossing the chosen probability), then NA is returned for the time estimate.

The probabilities are defined using the cumulative distribution function F(t) = 1-S(t). For example, prob = 0.9 will return the survival time at the point of 10% probability of surviving.

Examples

#----------------------------------------------------------------------------
# surv_time() examples.
#----------------------------------------------------------------------------
library(survival)
library(bkstat)

bkstat::predict_surv_time(
  object = coxph(Surv(time, status) ~ sex + ph.ecog, data = lung),
  df = data.frame(expand.grid(sex = 1:2, ph.ecog = 0:3)),
  probs = c(0.5, 0.9)
)
#>    sex ph.ecog prob time lower_ci upper_ci
#> 1    1       0  0.1  883      731       NA
#> 2    1       1  0.1  655      574      765
#> 3    1       2  0.1  455      363      624
#> 4    1       3  0.1  310      230      524
#> 5    2       0  0.1   NA       NA       NA
#> 6    2       1  0.1   NA      765       NA
#> 7    2       2  0.1  689      558       NA
#> 8    2       3  0.1  473      350      814
#> 9    1       0  0.5  371      310      524
#> 10   1       1  0.5  283      223      329
#> 11   1       2  0.5  186      166      245
#> 12   1       3  0.5  145      105      210
#> 13   2       0  0.5  583      455      791
#> 14   2       1  0.5  390      337      533
#> 15   2       2  0.5  286      222      371
#> 16   2       3  0.5  197      163      340
bkstat::predict_surv_time(
  Surv(time, status) ~ sex + ph.ecog,
  df = lung,
  probs = c(0.5, 0.9)
)
#>    sex ph.ecog prob time lower_ci upper_ci
#> 1    1       0  0.1  883      643       NA
#> 2    1       1  0.1  624      519       NA
#> 3    1       2  0.1  533      329       NA
#> 4    1       3  0.1  118       NA       NA
#> 5    2       0  0.1   NA      705       NA
#> 6    2       1  0.1  765      731       NA
#> 7    2       2  0.1  654      444       NA
#> 8    1       0  0.5  353      303      558
#> 9    1       1  0.5  239      207      363
#> 10   1       2  0.5  166      105      288
#> 11   1       3  0.5  118       NA       NA
#> 12   2       0  0.5  705      350       NA
#> 13   2       1  0.5  450      345      687
#> 14   2       2  0.5  239      199      444
bkstat::predict_surv_time(
  object = coxph(Surv(time, status) ~ 1, lung),
  probs = c(0.5, 0.9)
)
#>   prob time lower_ci upper_ci
#> 1  0.1  765      689       NA
#> 2  0.5  310      285      363
bkstat::predict_surv_time(
  Surv(time, status) ~ 1,
  df = lung,
  probs = c(0.5, 0.9)
)
#>   prob time lower_ci upper_ci
#> 1  0.1  735      689       NA
#> 2  0.5  310      285      363