Discrete Event Simulation for Health Technology Assessment

186 Discrete Event Simulation for Health Technology Assessment

other such statistical measure. In a model using that clinical trial result as

an input, it would be important to incorporate the parameter uncertainty,

generally based on the reported 95% condence interval. In some cases, the

input value may be known to be uncertain, but there may not be sufcient

empirical data to quantify that uncertainty adequately. This poses a quan-

dary for the analyst as ignoring variability in this parameter estimate is tan-

tamount to assuming it is known with certainty. Applying an arbitrary range

(e.g., ±50%)—although often done—is not an appropriate way to address

this uncertainty, but such analyses can be useful to determine if the model

outputs are ...