Probabilistic Analysis: Sensitivity Analysis or Main Result?

According to virtually all published Pharmacoeconomics guidelines, the base case analyses (understood as the main result) must be deterministic. That is, the main result of an economic model(the incremental cost-effectiveness ratio, ICER) is a fixed result and is notsubject to any uncertainty analysis. For example, with deterministic base case analysis we can obtain an ICER of 10,000 € per QALY gained with a new drug versus the former, but we do not know what confidence can we accept this result and what is the likelihood that the drug is cost-effective for a certain willingness to pay. The one-way sensitivity analysis, usually presented as tornado diagram is a deterministic analysis very useful to determine which variables of the study drivers of the result are obtained in the base case. 

According to the NICE guidelines "wherever possible the results of the economic comparisons should be subjected to sensitivity analysis testing. For example, when data are drawn exclusively from clinical trials, 95% confidenceintervals can be calculated for cost-effectiveness ratios. When data are drawn from a variety of sources and used in a modelling framework, probabilistic sensitivity analysis is recommended in order to take account of the uncertainty around data values". According to the ISPOR guidelines "for model-based economic evaluations, parameter uncertainty may be represented for individual parameters in a deterministic sensitivity analysis or across all parameters simultaneously with probabilistic analysis".