Assessment of Economic Model Stability By Repeated One-Way Sensitivity Analysis

Abstract

One-way sensitivity analysis (OWSA) is used to assess parameter uncertainty in health economic modeling by increasing and decreasing the base case value of each parameter and plotting the impact on the outcome in a tornado chart. OWSA provides useful information regarding which parameters are most influential. However, in cost-effectiveness analyses, where the outcome is an incremental cost-effectiveness ratio, OWSA may be challenging to interpret when the sign of the numerator or denominator changes and does not clearly display where these inflection points occur. We present a method to conduct repeated OWSA in order to better characterize the relationship between parameters and results across a range of values, and to identify where inflection points occur. A Microsoft Excel with Visual Basic for Applications (VBA)-based hypothetical model was created to demonstrate the method and to compare and contrast a traditional “fixed” OWSA and repeated OWSA. The repeated OWSA analysis performs a OWSA at multiple intervals within a range specified by the user. Results are plotted on an xy-scatterplot of the Δ % change in parameter value vs the ICER. Repeated OWSA displays which parameters result in a stable, predictable, ICER and also demonstrates at what point certain parameters result in an inflection in the ICER (ie, where the sign of the ICER changed due to a switch in the ∆ cost or ∆ effectiveness). Finally, the analysis displays the relationship between each parameter and the ICER result (eg, linear or non-linear). Repeated OWSA offers a potential improvement to fixed OWSA by providing a more complete view of how each parameter interacts with the ICER across a range rather than one snapshot at a fixed value. This analysis may be particularly useful in cases where there are small differences in costs or effectiveness, resulting in ICER volatility in fixed OWSA.