Economic Aspects of Evaluation - Evaluation Methods in Biomedical Informatics - page 327

FIGURE 11.7. Monte Carlo sensitivity analysis. In this technique, individual variables
in the analysis are replaced with probability distributions, and the analysis is recal-
culated hundreds or thousands of times: for each iteration a different set of data
randomly drawn from the distributions describing the variables is used. The result-
ing plot indicates the various possible cost and effectiveness answers for each iter-
ation. In this figure, most of the iterations result in points lying in the right upper
quadrant, where CEA is appropriate. On only very few iterations does the analysis
indicate the new strategy is worse than the existing strategy. Note that in the CE
plane, the acceptable ICER is represented by the slope of a line through the exist-
ing program. Points below the line represent areas where the intervention is cost-
effective in that the ICER is less than the acceptable range; points above that line
represent combinations of variables in which the ICER is above the acceptable
level.
the input variables drawn from the distributions. The analysis is repeated
many times, represented by the multiple dots. In this analysis, most of the
dots cluster around the baseline, and the majority are found below a line
that represents the acceptable cost-effectiveness threshold, indicating that
there is a high likelihood that this strategy is economically preferred to the
existing strategy.
Self-Test 11.3
1. The implementation of a particular information technology in your
health system will produce the following streams of costs and benefits
(all defined in terms of dollars).