broom (version 0.3.4)

survfit_tidiers: tidy survival curve fits

Description

Construct tidied data frames showing survival curves over time.

Usage

## S3 method for class 'survfit':
tidy(x, ...)
## S3 method for class 'survfit':
glance(x, ...)

Arguments

x

"survfit" object

...

extra arguments, not used

Value

• All tidying methods return a data.frame without rownames, whose structure depends on the method chosen.

  tidy returns a row for each time point, with columns

• time
  timepoint
• n.risk
  number of subjects at risk at time t0
• n.event
  number of events at time t
• n.censor
  number of censored events
• estimate
  estimate of survival
• std.error
  standard error of estimate
• conf.high
  upper end of confidence interval
• conf.low
  lower end of confidence interval
• glance returns one-row data.frame with the columns displayed by print.survfit
• records
  number of observations
• n.max
  n.max
• n.start
  n.start
• events
  number of events
• median
  median survival
• conf.low
  lower end of confidence interval on median
• conf.high
  upper end of confidence interval on median

Details

glance does not work on multi-state survival curves, since the values glance outputs would be calculated for each state. tidy does work for multi-state survival objects, and includes a state column to distinguish between them.

Examples

if (require("survival", quietly = TRUE)) {
    cfit <- coxph(Surv(time, status) ~ age + sex, lung)
    sfit <- survfit(cfit)

    head(tidy(sfit))
    glance(sfit)

    library(ggplot2)
    ggplot(tidy(sfit), aes(time, estimate)) + geom_line() +
        geom_ribbon(aes(ymin=conf.low, ymax=conf.high), alpha=.25)

    # multi-state
    fitCI <- survfit(Surv(stop, status * as.numeric(event), type = "mstate") ~ 1,
                  data = mgus1, subset = (start == 0))
    td_multi <- tidy(fitCI)
    head(td_multi)
    tail(td_multi)
    ggplot(td_multi, aes(time, estimate, group = state)) +
        geom_line(aes(color = state)) +
        geom_ribbon(aes(ymin = conf.low, ymax = conf.high), alpha = .25)

    # perform simple bootstrapping
    library(dplyr)
    bootstraps <- lung %>% bootstrap(100) %>%
        do(tidy(survfit(coxph(Surv(time, status) ~ age + sex, .))))

    ggplot(bootstraps, aes(time, estimate, group = replicate)) +
        geom_line(alpha = .25)

    bootstraps_bytime <- bootstraps %>% group_by(time) %>%
        summarize(median = median(estimate),
                  low = quantile(estimate, .025),
                  high = quantile(estimate, .975))

    ggplot(bootstraps_bytime, aes(x = time, y = median)) + geom_line() +
        geom_ribbon(aes(ymin = low, ymax = high), alpha = .25)

    # bootstrap for median survival
    glances <- lung %>% bootstrap(100) %>%
        do(glance(survfit(coxph(Surv(time, status) ~ age + sex, .))))

    qplot(glances$median, binwidth = 15)
    quantile(glances$median, c(.025, .975))
}