The scary topic of rationing seems to be the buzzword of the day in health care. As usual, being a nick-picky statistical type, I am unhappy with the public discussion that I have seen so far. Below I throw in my two cents.

There is a discussion of rationing of healthcare and the value of life by Peter Singer in the New York Times. I think it is a good discussion and urge everyone to read it. It gives a good idea of the general ideas, promises and problems of giving a dollar value to a human life, and its use in making social decisions regarding health and safety.

It is excellent in pointing out that difficult decisions must be made in life, and we are fooling ourselves if we think we can evade them by ignoring the issue, or just assuming that whatever system we have does it best, for somehow magically evades the problem.

Unfortunately it leaves out a crucial proviso, which leaves the wrong impression, and muddles the issue. In particular, it gives the impression that an economist has some economic formula that can determine whether it is socially optimal to approve or not approve a drug that is required to cure a deadly disease. This is in my opinion simply wrong. The economist has no such basis, either theoretical or empirical. Leaving the impression that an economist does have some magic way of estimating the value of life for treatment of deadly diseases leaves the door open to unfounded fears that health reform will produce rationing of life-saving drugs, arbitrary denial of care, and all sorts of other potentially nasty things

I think that in some cases, the idea of estimating “the value of a human life” and using for healthcare decisions is a very good idea and can make us all better off, including the very ill. Some may disagree, but the idea should be presented carefully, so that people really understand what it can and cannot do. Applying the concept, even in an opinion piece, in situations where it is not appropriate will simply give ammunition to obstructionists. . An example is opposition to what I think is the very good idea that there be a research board the estimates the relative effectiveness of different public health and medical therapies. That such a board would wield life and death power over everyone who is very sick is being used to argue against any healthcare reform at all.

Singer gives the scary side of rationing and the concept of the value of a human life right up front. In the UK, a board of health experts sets a limit on the how much should be paid for life-saving drugs. The limit was that a drug should only be used if it produced an additional year of life, on average, at a cost of $49,000 or less. A drug for advanced kidney cancer produced a year of life at a cost higher than that, so the government decided that the UK National Health Service should not pay for it. Singer says that this has something to do with the value of life. To my professional knowledge, it does not.

The idea of the economic ‘value of a human life’ or ‘value of a statistical life’ is simple, and Singer does a good job of describing it, so I will quote him:

“Suppose that there is a 1 in 100,000 chance that an air bag in my car will save my life, and that I would pay $50 – but no more than that – for an air bag. Then it looks as if I value my life at $50 x 100,000, or $5 million.”

Now I am an economic statistician, or to be fancy about it, an econometrician, and I believe that this is a sound way of thinking. There are a number of ways you can estimate such a ‘value of a statistical life’ For example, the change in salary required for a worker to move to a more hazardous job.

I believe that estimates such as these are very useful if prioritizing public health, and preventive health programs, and routine medical care, and deciding which programs to adopt.

But Singer implies in his piece that this kind of reasoning has something to with the decision to approve or not approve a life saving drug for a deadly disease. I do not think it does, and here is why.

All of the empirical estimates we have for the value of a statistical life are from situations where a person makes decisions that concern small variations in probability of death around a low base rate of death that is normally experienced by adults in everyday life. For the air bag example, a one in 100,000 chance that this air bag might save my life. For the worker choosing a new job, increases in the chance of death from an average for the general population of less than 0.0025 per year (the average for 22 to 50 year olds who are likely to have physically hazardous jobs).

A life saving drug for a cancer patient, or a stroke or heart attack victim may very well produce big changes in probability of death (maybe 25% or more) around a large probability of death over the next year, or even weeks for heart attack or stroke (say, over 10%).

Now compare the base probabilities of death, or changes in that probability of death, that the person is considering: for estimates of the value of life, one in 100,000 or one in 400. For a possibly fatal disease: maybe one in ten, and often far more. What does a person’s behavior in the former situation say about a person’s preference in the latter? I think a fair minded answer is ‘who knows?’ At least every good textbook of cost-benefit analysis says that the answer is “who knows?”

Using the value of a statistical life estimated for people facing the normal hazards of everyday life to decide what should be the cut-off for effort in saving an individual who faces a probability of death orders of magnitude higher, and for decisions that can change the chance of death dramatically, has no basis in economics or sound statistical practice.

I do not know whether the UK board uses the ‘value of life’ to justify its decisions. If it thinks it is using the value of life in that decision, it is mistaken and should stop saying wrong things to justify its actions. Most discussions I see of what such a cut-off level should be are based on what, historically, society and the medical profession have considered reasonable costs, on average. And that is actually, in my view, a sounder criterion than misapplying the value of life for difficult treatment decisions such as approving an expensive medicine for a person with a probably fatal cancer.

Perhaps I am being overly wonky about this, and no one cares. But I think that it is important not to discuss these concepts sloppily. Here is why. I think it is very sound and good policy to use the value of a statistical life to make decisions regarding public health and preventive medicine. We do have a basis for deciding which treatments, and public health interventions are in some sense worth it, and which are not. We can use estimates that reflect people’s observable behavior in the face of similar magnitudes of risk in their daily lives.

If we do that job efficiently, for preventive health visits, maternal and perinatal health, smoking, traffic safety, low risk treatment decisions (for example, treatment of high blood pressure or cholesterol) then we will have more resources available for truly difficult social decisions that must be made for difficult cases of deadly disease.

But we should not make the mistake of thinking that economics or statistics contain any magic formula that gives us knowledge of the value of near certain life versus near certain death.

Singer goes into detail about all the agonizing problems of using this value of life concept in different circumstances, and on different kinds of people with different disabilities. Singer is a philosopher who believes in utilitarianism, and I guess he thinks that there is such a thing as a total utility of human life that can be maximized if we can just get some kind of formula right.

In the course of this he muddles up the concept called a Quality Adjusted Life Year (QALY). He may just be slopply, but he gives the impression that a QALY tells us what the relative value of life should be for people with different levels of disability. It does not. QALYs are an attempt to determine the relative welfare of people with different degrees of disability, but it does not estimate what the value of their life is. In a simple example, a perfectly healthy person has a QALY of one, a dead person has a QALY of zero, and a handicapped or sick persona has, say, a QALY of 0.5. But there is nothing in the estimation of QALYs that tell us what that value of one for the perfectly healthy person should be worth in terms of other goods and services, or in dollars. And there is nothing that can tell what that QALY of 0.5 should be in terms of dollars.

Singer may have a theory about it, but I do not know that it is commonly accepted. There is nothing in the estimation of QALYs that ties them to dollars or risks in non-medical situations.

Singer’s other points are very good. In particular, Singer makes that point that we do not, and cannot, evade the problem of making decisions about the value of life and death under our current, supposedly market based, system of healthcare in the US, and if we think we can evade those choices, or that our system will make better choices or easier choices than a more equitable system, then we are just fooling ourselves.

However, I do think that some of the concepts needed to understand the issues involved in healthcare reform are not all that complicated. And that needless ambiguity or carelessness will work against good healthcare reform.

The bottom line is, when making decisions about near certain life or near certain death, a ‘value of life’ may or may not exist, But one thing is for sure, economics and statistics cannot tell you want it is in terms of dollars. We just do not know. There is a basis for estimating public health and routine medical care that involve small changes in the probability of life and death for the average adult, and we should use that to create more opportunities for people whenever possible. That last task his one task of comparative effectiveness research in healthcare, and it is an important one that shuold be done.

Why We Must Ration Health Care
By PETER SINGER
Published: July 15, 2009
http://www.nytimes.com/2009/07/19/magazine/19healthcare-t.html?pagewanted=1&_r=1&ref=magazine

found via
17 Jul 2009 12:18 pm
Peter Singer on Health Care Rationing
By Conor Clarke
http://andrewsullivan.theatlantic.com/the_daily_dish/2009/07/peter-signer-on-health-care-rationing.html#more