This looks at the costs of extending the end of life by a short period of time and tries to draw a boundary between those cases when treatment should be applied, and those when it is not worth it to do so (the conclusion is that "studies powered to detect a survival advantage of two months or less should test only interventions that can be marketed at a cost of less than $20,000 for a course of treatment").
How do we draw this line (and if you don't think we should, how do we avoid drawing it)? Usually, I would give the standard answer that we should employ these life-extending procedures up until the point where the marginal cost of the treatment equals the marginal benefit, and let someone else worry about how to actually measure the costs and benefits. But in this case the measurement of the benefits - life itself - seems particularly hard to quantify, and trying to account for quality of life complicates it further, and it is not clear to me that a market test is even appropriate when there may not be a tomorrow and standard opportunity cost tradeoffs are missing from the evaluation (update: thinking more, I suppose this is just "cap-T" in our models, which isn't too hard to handle in the deterministic case, i.e. where T is known in advance with certainty, but the evaluation still seems problematic due to the other reasons that are cited). So I don't think there is a good answer to this question, at least not one that standard economic models can deliver:
How much is life worth? The $440 billion question, EurekAlert: The decision to use expensive cancer therapies that typically produce only a relatively short extension of survival is a serious ethical dilemma in the U.S. that needs to be addressed by the oncology community, according to a commentary published online June 29 in the Journal of the National Cancer Institute.
Tito Fojo, M.D., Ph.D., of the Medical Oncology Branch, Center of Cancer Research at the National Cancer Institute, in Bethesda, Md., and Christine Grady, Ph.D., of the Department of Bioethics, the Clinical Center at the National Institutes of Health, ... illustrate cost-benefit relationships for several cancer drugs, including cetuximab for treatment of non-small cell lung cancer, touted as "practice changing" and new standards of care by professional societies, including the American Society of Clinical Oncology. ...
According to Fojo and Grady,... 18 weeks of cetuximab treatment for non-small cell lung cancer, which was found to extend life by 1.2 months, costs an average of $80,000, which translates into an expenditure of $800,000 to prolong the life of one patient by 1 year. At this rate, it would cost $440 billion annually ... to extend the lives of 550,000 Americans who die of cancer annually by 1 year.
To address the issue, the commentators recommend that studies powered to detect a survival advantage of two months or less should test only interventions that can be marketed at a cost of less than $20,000 for a course of treatment.
Every life is of infinite value, the authors say, but spiraling costs of cancer care makes this dilemma inescapable.
"The current situation cannot continue. We cannot ignore the cumulative costs of the tests and treatments we recommend and prescribe. As the agents of change, professional societies, including their academic and practicing oncologist members, must lead the way," the authors write. "The time to start is now."