Map of life expectancy at birth from Global Education Project.

Tuesday, October 07, 2008

What are you afraid of?

Since I'm going to give a little talk about this today, I might as well conserve my dwindling intellectual resources and talk about it here. We face innumerable risks in daily life, from being hit by a bus to breathing in invisible ultra-fine particles from the bus's exhaust, to eating trans-fat in our morning muffin.

Scientific and technical type people -- engineers, epidemiologists, policy analysts, physicians, insurance actuaries, etc. -- evaluate those risks using mathematics. They might be thinking about the risk to an individual, as physicians are likely to do, or to populations, as epidemiologists do. There are different complexities in the two cases, but the individual case is perhaps a bit more straightforward. Here I'll concentrate on the population case.

A "scientific" assessment of risk is essentially an equation of the form P X Harm, where P is the probability of harm occurring, and "Harm" is a quantification of its magnitude. To decide whether risks are worth taking, or which are more severe, you need a common metric for "Harm." Popular choices are dollars, and something called Quality Adjusted Life Years, QALYs (pronounced Quahlies).

To value human lives or health in dollars seems intuitively repugnant to most people, but of course we do it all the time. We could be spending four bucks to save a kid in Africa from malaria or starvation, but we don't do it, right? Resources are scarce, we're all gonna die anyway, so yes, ultimately human lives have finite value in dollars. Exactly how to make that calculation, however, and what its ethical implications may be, are highly controversial questions, however.

QALYs, in essence, are years of life, marked down for ill health or disability, for example, a year of life in a wheelchair is some percentage of a year of being able to walk. No, I'm not kidding. But if you ask people -- at least comparatively healthy and young people -- if they would be willing to give up some amount of total life span to avoid a period of disability, most people will answer yes. They'd be willing to die a bit earlier to avoid going blind or whatever. And it's on the basis of that kind of survey data that the QALYs are calculated.

Then there's P. For an individual, it's the probability of it happening to that person. For a population, the probability will be distributed unevenly, so you have to add up all the individual probabilities, multiply them by the lost dollars or QALYs per person, and there's your total cost to the population. If you've made the computation in dollars, you can compare that to the cost of eliminating or reducing the risk, and voila, you know whether it's worth doing. If you've made the computation in QALYs, you can compare the cost of eliminating the risk to other possible expenditures and decide which one is a higher priority.

There are some problems. Valuing lives in dollars generally means figuring out a person's predicted lifetime economic contribution. You can include non-market contributions such as child care and housekeeping, being a friend, etc., but you still end up valuing the lives of higher income and younger people more highly than those of people with lesser earning potential. You could try to fix that by assigning the same dollar value to a QALY, regardless of who it belongs to, but with QALYs, the lives of people with disabilities or ill health, and again, older people, are worth less, whether or not you convert QALYs to dollars.

In fact, while we might say we're willing to give up some life span to stay healthy, once we find ourselves disabled or sick we don't necessarily find that we value our lives less. Whoops, I didn't really mean that after all. Too late, sucker, your insurance won't cover that.

Average people also don't have access to the information to make these calculations anyway, and finally, they don't think in probabilities the same way scientists do -- they buy lottery tickets, for example, and it gets a lot more complicated than that when you start to weight potential gains vs. potential losses. Most people's assessment of risk is based on heuristics -- rules of thumb, readily available decision rules -- that have to do with characteristics of risks that are quite different from their probabilities and quantifiable harms. These include whether they are assumed voluntarily, whether they are catastrophic -- concentrated in a single notable event, such as a plane crash, vs. spread out into smaller events such as automobile accidents -- whether they are natural or human caused, etc.

Many academic analysts call people's views of risk "irrational," but maybe that's arrogant and elitist and all that. Maybe people's views incorporate an instinctive view of fairness and an ethical accounting for agency that ought to be acknowledged in the way we make these calculations.

These are convoluted questions but they are at the heart of much political disagreement, and need to be discussed more openly and in terms that people can understand and relate to.

No comments: