Abstract
A variety of stochastic and deterministic non-optimizing techniques have been used both predictively and for comparing policy options for patient treatment. Models of the system of the treatment of kidney patients are reviewed and are shown to be based on too small a subsystem to be useful for planning and budgeting. Other drawbacks include poor user-credibility and lack of robustness. Discrete-event simulation is shown to be the most appropriate technique which does not limit the type of distribution functions that may be used and can model patient attributes, resource use and constraints.