2.4 Sensitivity analyses
To address the uncertainty of costs and outcomes in our model, we did
several 1-way sensitivity analyses. The parameters listed in Table 1-3
were varied in specified ranges of NAC and AC. In the base case
analysis, a chance of ±20% to the original values were set as the upper
and lower bounds.
We also performed probabilistic sensitivity analysis (PSA) to assess the
combined effect of multiple parameter uncertainty on the increment cost
effectiveness ratio (ICER.) Factors varied in sensitivity analyses
included OS, base-case event probabilities, costs, utility weights and
treatment time frames with distributions selected based on published
literature [30]. Costs and treatment time obeyed gamma distribution,
with a standard error of 10% of the original value. Utilities
distribution reflected a triangular distribution, and base-case utility
weights were set as the most likely value with minimum and maximum
values of ±0.05. Events probabilities distribution were beta
distribution, with alpha and beta parameters calculated through the
event’s incidence in the study of Brandt et al [10]. The PSA was
performed with 10,000 Monte Carlo simulations across all distributions.