Simulate a single subject's event time under piecewise-constant hazard

Simulates an event time for one subject followed from entry age entry_age to administrative censoring age censor_age under an attained-age baseline hazard that is piecewise constant over age bands. The subject-specific hazard is $$ h(a) = h_0(a) \exp(\eta), $$ where exp_eta = exp(eta) is the Cox proportional hazards multiplier.

Usage

.simulate_one(entry_age, censor_age, exp_eta, lookup_table)

Arguments

entry_age: Numeric scalar giving the subject's age at study entry.
censor_age: Numeric scalar giving the administrative censoring age. Must satisfy censor_age > entry_age.
exp_eta: Positive scalar equal to $\exp(\eta)$, the subject-specific hazard multiplier.
lookup_table: List produced by .prep_lookup(), containing numeric vectors lo, hi, and rate defining the attained-age baseline hazard bands.

Value

A list with components:

age_exit: Numeric scalar giving the exit age (event or censoring).
event: Integer scalar equal to 1 if an event occurred and 0 if administratively censored.

Details

The simulation proceeds band-by-band. At the current attained age current_age, the baseline hazard $h_0(a)$ is obtained from the age band containing that age. A waiting time is drawn from an exponential distribution with rate h_0(a) * exp_eta. If the waiting time occurs before the end of the current age band (or before censor_age), an event occurs. Otherwise the subject advances to the next band boundary and the procedure repeats until censor_age.

Age bands are treated as left-closed, right-open intervals: [age_lo, age_hi). An attained age equal to age_hi is assigned to the next band.

The baseline hazard must be defined for all attained ages reachable in [entry_age, censor_age). If no age band contains the current attained age, the function errors.

The method is exact for piecewise-constant hazards because within each age band the hazard is constant and the waiting time distribution is exponential.

Examples

lk <- list(
  lo = c(50, 60),
  hi = c(60, 70),
  rate = c(0.01, 0.02)
)

set.seed(1)
.simulate_one(entry_age = 55,
              censor_age = 65,
              exp_eta = 1.5,
              lookup_table = lk)
#> Error in .simulate_one(entry_age = 55, censor_age = 65, exp_eta = 1.5,     lookup_table = lk): could not find function ".simulate_one"