Healthcare stakeholders require accurate and reliable models to aid performance management, but there is an issue as to whether deterministic or stochastic modelling is more appropriate. Deterministic models are usually simpler, more easily understood, and less data intensive. However, stochastic models tend to be more realistic, but require significant data to develop a realistic model. We illustrate that when queueing is a defining property of the problem semantics then we need suitable data and a model that allows us to describe and characterise the relationship between the queueing process, its parameters and delay. By comparing an analytic deterministic model with a stochastic simulation of an orthopaedic Integrated Clinical Assessment and Treatment Service, we show that when uncertainty is ignored in the planning stages then there will be a deficiency in the number of staff allocated. This causes queues to build up and it follows that patients will receive a diminished quality of care.