Parameter Uncertainty in Health Economics Modelling

Should account for variability in model inputs at two levels:

1. The individual ("spatial"): What characteristics does a patient have at the beginning of the model? 
2. Temporal: What happens to a patient in the model over time?

Terminology

• Monte Carlo simulation => use of random numbers in evaluating a model
• Probabilistic Sensitivity Analysis => incorporate all parameter uncertainty through random sampling from probability distributions
• Sampling defines patient characteristics at the start of the model (e.g. age; blood pressure; HbA1c)
• Sometimes referred to as 2nd-order simulation
• Sampling defines patient's transition through the model (at the "chance" nodes)
• Sometimes referred to as 1st-order simulation or microsimulation or random walk