OBJECTIVES : Probabilistic Sensitivity Analysis (PSA) is a well-established tool to assess parameter uncertainty in Cost-Effectiveness analysis. Yet, parameters are generally varied independently, and a large set of measures of uncertainty need to be input in a Cost-Effectiveness model in the presence of comparators coming from a Network-Neta-Analytic (NMA) framework.

METHODS : We suggest a simple approach based on Singular-Value-Decomposition that quantifies the total uncertainty around the parameters in use in the Cost-Effectiveness model. We assess how the total uncertainty changes when subgroups of the entire population are selected or additional comparators from NMAs are added. A simple cell with a drop-down menu allows for the selection of the relevant total uncertainty values. These are coupled to orthogonal matrices to build the relevant variance-covariance matrices for the set of parameters in use.

RESULTS : We present an Excel-model that performs PSAs for multiple comparators and the option to select many subgroups with a limited set of data (e.g. we don't include variance-covariance matrices from individual parametric extrapolations).

CONCLUSIONS : We show how simple linear algebra methods can significantly simplify the excel-based modeling efforts and model the overall uncertainty in Cost-Effectiveness evaluations more reliably.