| Expected
value decision making can help in choosing between treatment options
by averaging together the outcomes experienced by many patients.
However, each patient only experiences one outcome. Expected value
decision making increases the odds that the patient will experience
a good outcome. It does not guarantee a good outome.
As you have seen in
the scenario of the injured rock climber, the difference in the
expected value of the decision alternatives depends on the values
assigned to the probabilities and utilities in the decision tree.
Given the opportunity, two patients might assign very different
values to different outcomes. Furthermore, a probability might
vary from patient to patient. This is why it is important to incorporate
sensitivity analysis into expected value decision making.
Sensitivity analysis
requires a lot of computations, especially when you allow the
value of an uncertain parameter to take on every value in the
range in order to graph the relationship between the value of
the parameter and the expected outcomes of the three decision
alternatives. Therefore, people who do decision analysis rely
on computer programs that allow you to represent a decision tree,
attach values to the parameters of the decision tree, and perform
and then graph sensitivity analysis.
Sensitivity analysis
is fun. Doing it gives you a feeling for the power of decision
analysis. If you are interested in learning more about sensitivity
analysis, the author of this chapter recommends visiting the following
Web site where you can play with the injured rock climber model:
www.treeage.com/demos/actxdemo.htm
If you visit this Web
site, you can:
- Do sensitivity analysis
and see how the preferred option changes as you adjust different
parameters of the model.
- Access and download
a small version of a program to create decision trees. (Have
fun!)
Note: The computational
model at this Web site is slightly different than the one illustrated
in this chapter, so the expected values are slightly different.
|