Economic Aspects of Evaluation - Evaluation Methods in Biomedical Informatics - page 323

model with entirely new components in it, and would represent a structural
sensitivity analysis in that the actual structure of the model representing
the strategies is extended to include a different level of detail concerning
the potential outcomes of each strategy.
Parameter Estimate Sensitivity Analyses
Virtually every number involved in an economic analysis is at best a point
estimate of a quantity that could, in reality, be different from the specific
number estimated in that particular instance. For example, even in the sit-
uation in which the estimate of a particular cost is the result of an economic
analysis conducted alongside a randomized controlled trial, the cost esti-
mate is only accurate within some confidence limits; had the trial been con-
ducted in a different set of patients or a different setting, undoubtedly the
estimated costs would be (at least slightly) different. More commonly, the
estimate may come from the literature, and is accompanied by a measure
of its accuracy given by confidence limits or standard deviation. Although
the accuracy with which parameters for a particular analysis are known may
vary, it is rarely (if ever) the case that a parameter is known exactly. There-
fore an analysis of the model outputs using slightly different estimates for
each parameter is necessary.
One of the most common methods of describing the variability of results
in an economic analysis is to provide baseline , best-case , and worst-case
calculations. This is illustrated in Figures 11.4 and 11.5. Assume that a
CEA has been done as described in Figure 11.4. There are two strategies,
A and B, each with different costs in several categories (hospital costs,
ambulatory costs, pharmacy costs, and time costs) and strategy B is both
more expensive and more effective (it adds 5 years of life expectancy for
this particular disease). The straightforward calculation of the ICER is to
FIGURE 11.4. Sample cost-effectiveness analysis, describing the comparison between
two strategies, A and B, in which B is both more expensive and more effective than
A, and calculating a cost-effectiveness (CE) ratio.