Sensitivity analysis is a procedure which adds further information to that derived in clinical trials and cost-effectiveness analyses. There are broadly two kinds: variable-by-variable analysis (sometimes called univariate sensitivity analysis) and scenario analysis (or multivariate sensitivity analysis). In variable-by-variable analysis one lists the important factors that affect the size of the costs and outcomes and for each of them a range of plausible values around the mean (for example, 'optimistic', 'most likely', or 'pessimistic') is specified. Incremental cost-effectiveness ratios (ICERs) are then calculated for each value of each factor, holding all other factors at their expected or most likely values. Thus, if there are three important factors and three estimates for each factor, seven different ICERs will be calculated. In this way one hopes to identify the source(s) of the biggest variations about which decision makers will have to make a judgment (and which may identify priority areas for future research). Scenario analysis allows for the possibility that factors affecting ICERs are not independent of one another, as is assumed in variable-by-variable analysis. In this case, one selects a variety of generalized states of the world (for example, worst case, middling case, best case) and takes all the worst case outcomes, middling, best (as the case may be) to calculate the ICERs that would result under the circumstances specified. Typically this method produces much more extreme variations than the variable-by-variable method. See Modelling.
Was this article helpful?