Simulation of trajectories in the illness-death model for chronic diseases: discrete event simulation, Doob-Gillespie algorithm and coverage of Wald confidence intervals

Abstract

AbstractWe compare two approaches for simulating events in the illness-death model in a test example about type 2 diabetes in Germany. The first approach is a discrete event simulation, where relevant events, i.e., onset of disease and death, are simulated for each subject individually. The second approach is the Doob-Gillespie algorithm, which simulates the number of people in each state of the illness-death model at each point in time. The algorithms are compared in terms of bias, variance and speed. Based on the results of the comparison in the test example, we assess coverage of the corresponding Wald confidence intervals.