I take the canned data that is used to demonstrate the coxph function in R, and then I look at the cumulative hazard.
require(survival)
options(na.action=na.exclude) # retain NA in predictions
fit <- coxph(Surv(time, status) ~ age + ph.ecog + strata(inst), lung)
x <- basehaz(fit)
p <- x[x$strata=='inst=1',]$hazard
plot(p)
basehaz is reporting a cumulative hazard function that moves greater than 1.
How is this possible?