This is a very good point.

I have redone the analysis with this concern in mind. I also removed the outcomes involving hospitalization from the decision tree, as they were barely making any contribution to the expected values of the outcomes in which an individual either received or did not receive a flu shot.

The contribution of E(death) to E(getting a flu shot) ended up being around - $30-90 for healthy adults of ages 19-64, rather than - $500, mainly because around 90% of deaths from flu are in people of ages 65+.