The point at which the time-to-event sensitivity designs are simulated

Crossing control-arm heterogeneity, loss to follow-up and the event time distribution with the whole grid multiplies a run by the product of their lengths, and most of that is spent re-simulating designs nobody reads. When sensitivity_reference names a sample size factor and a source denominator change factor, a design other than the primary one is simulated only there, so the sensitivity analysis still varies the drift across its full range at a single trial size. Without the key the axes cross the whole grid, which is what a configuration written before it existed does.

Usage

time_to_event_sensitivity_reference(scenarios_config, case_study_config)

Arguments

scenarios_config

Configuration for the simulation.

case_study_config

Configuration for the case study.

Value

A list with sample_size_factor and denominator_change_factor, or NULL when every design is simulated across the whole grid.