Event instantaneous hazard based on Kaplan-Meier survival estimates
Compute event instantaneous hazard on the basis of a Kaplan-Meier survival function.
epi.insthaz(survfit.obj, conf.level = 0.95)
survfit.obj |
a |
conf.level |
magnitude of the returned confidence interval. Must be a single number between 0 and 1. |
Computes the instantaneous hazard of the event of interest, equivalent to the proportion of the population failing per unit time.
A data frame with three or four elements: strata the strata identifier, time the observed failure times, sest the observed Kaplan-Meier survival function, slow the lower bound of the confidence interval for the observed Kaplan-Meier survival function, supp the upper bound of the confidence interval for the observed Kaplan-Meier survival function, hest the observed instantaneous hazard (the proportion of the population at risk experiencing the event of interest per unit time), hlow the lower bound of the confidence interval for the observed instantaneous hazard, and hupp the upper bound of the confidence interval for the observed instantaneous hazard.
Venables W, Ripley B (2002). Modern Applied Statistics with S, fourth edition. Springer, New York, pp. 353 - 385.
Singer J, Willett J (2003). Applied Longitudinal Data Analysis Modeling Change and Event Occurrence. Oxford University Press, London, pp. 348.
library(survival)
dat <- lung
dat$status <- ifelse(dat$status == 1, 0, dat$status)
dat$status <- ifelse(dat$status == 2, 1, dat$status)
dat$sex <- factor(dat$sex, levels = c(1,2), labels = c("Male","Female"))
lung.km01 <- survfit(Surv(time = time, event = status) ~ 1, data = dat)
lung.haz01 <- epi.insthaz(lung.km01, conf.level = 0.95)
lung.shaz01 <- data.frame(
time = lowess(lung.haz01$time, lung.haz01$hlow, f = 0.20)$x,
hest = lowess(lung.haz01$time, lung.haz01$hest, f = 0.20)$y,
hlow = lowess(lung.haz01$time, lung.haz01$hlow, f = 0.20)$y,
hupp = lowess(lung.haz01$time, lung.haz01$hupp, f = 0.20)$y)
plot(x = lung.haz01$time, y = lung.haz01$hest, xlab = "Time (days)",
ylab = "Daily probability of event", type = "s",
col = "grey", ylim = c(0, 0.05))
lines(x = lung.shaz01$time, y = lung.shaz01$hest,
lty = 1, lwd = 2, col = "black")
lines(x = lung.shaz01$time, y = lung.shaz01$hlow,
lty = 2, lwd = 1, col = "black")
lines(x = lung.shaz01$time, y = lung.shaz01$hupp,
lty = 2, lwd = 1, col = "black")
## Not run:
library(ggplot2)
ggplot() +
theme_bw() +
geom_step(data = lung.haz01, aes(x = time, y = hest), colour = "grey") +
geom_smooth(data = lung.haz01, aes(x = time, y = hest), method = "loess",
colour = "black", size = 0.75, linetype = "solid",
se = FALSE, span = 0.20) +
geom_smooth(data = lung.haz01, aes(x = time, y = hlow), method = "loess",
colour = "black", size = 0.5, linetype = "dashed",
se = FALSE, span = 0.20) +
geom_smooth(data = lung.haz01, aes(x = time, y = hupp), method = "loess",
colour = "black", size = 0.5, linetype = "dashed",
se = FALSE, span = 0.20) +
scale_x_continuous(limits = c(0,1000), name = "Time (days)") +
scale_y_continuous(limits = c(0,0.05), name = "Daily probability of event")
## End(Not run)
## Stratify by gender:
lung.km02 <- survfit(Surv(time = time, event = status) ~ sex, data = dat)
lung.haz02 <- epi.insthaz(lung.km02, conf.level = 0.95)
## Not run:
library(ggplot2)
ggplot() +
theme_bw() +
geom_step(data = lung.haz02, aes(x = time, y = hest), colour = "grey") +
facet_grid(strata ~ .) +
geom_smooth(data = lung.haz02, aes(x = time, y = hest), method = "loess",
colour = "black", size = 0.75, linetype = "solid",
se = FALSE, span = 0.20) +
geom_smooth(data = lung.haz02, aes(x = time, y = hlow), method = "loess",
colour = "black", size = 0.5, linetype = "dashed",
se = FALSE, span = 0.20) +
geom_smooth(data = lung.haz02, aes(x = time, y = hupp), method = "loess",
colour = "black", size = 0.5, linetype = "dashed",
se = FALSE, span = 0.20) +
scale_x_continuous(limits = c(0,1000), name = "Time (days)") +
scale_y_continuous(limits = c(0,0.05), name = "Daily probability of event")
## End(Not run)Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.