A fascinating paper. Healthcare spending has little measurable effect on health or lifespan.
“Gender, exercise and social status, for example, can change lifespans by ten to
fifteen years or more (Lantz, House, Lepkowski, Williams, Mero, & Chen, 1998). When one
looks at medical spending, however, the usual2 finding in such studies is no effect. When
comparing nations or counties or individuals, people who get more medicine have no signif-
icant difference in health when compared to people who get less medicine.
For example, Jonathan Skinner and John Wennberg looked at five million Medicare
patients in about 3500 hospital service areas in 1990. They looked to see if people died
less in areas where, during the last six months of life, Medicare spent more on treatment.
After controlling for age, sex, race, median income, poverty, education, urbanization, and
initial health limitations, Skinner and Wennberg found that the average effect of spending
$1000 more was somewhere between increasing lifespan by about five days and decreasing
it by about fifteen days.3 Areas that kept patients in the intensive care unit one more day
on average reduced lifespan by between fifteen and seventy five days (Skinner & Wennberg,
2000).”
http://hanson.gmu.edu/feardie.pdf
Fear of Death and Muddled Thinking – It Is So Much
Worse Than You Think
Robin Hanson∗
Department of Economics
George Mason University†
August 2005
Abstract
Humans clearly have trouble thinking about death. This trouble is often invoked to
explain behavior like delays in writing wills or buying life insurance, or interest in odd
medical and religious beliefs. But the problem is far worse than most people imagine.
Fear of death makes us spend fifteen percent of our wealth on medicine, from which we
get little or no health benefit, while we neglect things like exercise, which offer large
health benefits.
One cannot look directly at either the sun or death. La Rochefoucauld, 1665.
Any animal that uses emotion to focus its attention must have a healthy fear of death.
When something death-related appears in its environment, the animal must focus on iden-
tifying and responding to possible threats. Humans have inherited this fear of death, but
human intelligence allows many new ways for environmental cues to suggest death to us. So
humans must manage their wandering thoughts, to avoid triggering this fear too often.
We usually do not feel very afraid of death, and we are not aware of thinking about death
much. But apparently we are just not very aware of how much the fear of death directs our
thinking. Scholars assure us that subconscious death fears are a powerful influence on our
thinking (Becker, 1973).
Psychologists have even found that weak death cues, like standing next to a Mortuary or
watching an autopsy video, measurably change our thinking. In the presence of such death
cues we tend to reward heros and punish prostitutes more. We tend more to favor those
who praise our religion and nation, and to favor those who criticize others. We become
more reluctant to treat flags and crucifixes casually. We think we are better drivers, and
that others agree with us more. We try harder to divert attention from our less attractive
features and group affiliations (Soloman, Greenberg, & Pyszczynski, 2000). We believe more
in the supernatural (Norenzayan & Hansen, 2006). (People with high self-esteem, oddly
enough, are mostly immune to these effects.1)
Ah yes, the fear of death, you say. As someone who reads books like “Death and Anti-
Death,” you feel familiar with this concept. You sometimes feel the fear yourself, and you see
that people often delay long before arranging for life insurance, wills, or HIV tests (Kopczuk
& Slemrod, 2005). You may have seen people pushing to do “everything possible”, even
harmful things, for a dying loved one because they could not accept the inevitable. If you
are an atheist, you may think that religion and supernatural beliefs exist primarily to help
people deal with death fears (Atran & Norenzayan, 2004). And if you have an unusual
medical belief, such as in cryonics or life extension, you may invoke the fear of death to
explain the hostility and disinterest you face.
But you probably think that this is about as far as it goes. Sure, among “primitive”
cultures, or the less educated around you, a desperate need to feel control over death may
lead to an overconfident trust in fairy tales and quack cures. But you know that your tales
and cures are supported by the clarity of science and statistics. You live in a rich society
with an advanced understanding of biology, enabling powerful technologies of medicine and
public health, allowing a vast improvement in health and lifespan. The modern medical
miracle, at least, is no mirage. Right?
Sorry to be the bearer of bad news, but, well, maybe you should sit down. Are you sitting
now? O.K., then let us just see how deep this rabbit hole goes. Yes, our society understands
a great deal about biochemistry and the mechanisms of life. This allows us to generate
reasoned hypotheses about causes of our deaths and interventions that might prevent such
deaths. But biological systems are far too complex for us to have much confidence in such
hypotheses. Fundamental principles cannot tell us why exactly any one person dies, or what
would have prevented that death. To confirm or refute such hypotheses, we must look at
who actually lives and dies in the world around us.
But we do look at who lives and dies in our world, do we not? Well individual doctors
have long justified their confidence in the cures they recommend by pointing to their personal
experience seeing patients live and die. Yet many medical historians now say that through
most of human history people would on average have been healthier if they had completely
avoided doctors. So clearly our doctors’ informal experiences can not by itself assure us
that we are avoiding fairy tales and quack cures. You may know someone whose health
dramatically improved after being treated. But usually even untreated people get better
eventually, and you probably also know people whose health dramatically declined after
being treated. Personal anecdotes just do not help much here.
But our doctors rely on scientific studies, not just informal experience, right? Well, in
1
I suspect this behavior was functional for our distant ancestors. By believing more in themselves, the
supernatural, and their nation or religion, our ancestors could credibly signal their ability and loyalty, which
might inspire their neighbors to help more to prevent death.
2
fact the vast ma jority of what doctors do is not based on specific scientific studies. For most
things that doctors do there are no studies, and when there is a study most doctors have not
even heard a summary of the study. Actually, doctors mostly just copy what the doctors
around them do, and try a few new things out on their own. This produces huge practice
variation, where similar patients are treated very differently by doctors in different regions
(Welch, Miller, Welch, Fisher, & Wennberg, 1993).
O.K., this may be true about treatments like surgery, bed rest, and so on, but for drugs
at least regulation forces doctors to attend to studies, right? Well, it is true that doctors can
only prescribe drugs approved by government regulators. However, once a drug has been
approved to treat any condition, doctors are free to use that same drug to treat any other
condition, regardless of what studies say.
How about when regulators approve drugs to treat specific conditions? At least we know
those treatments are more than fairy tales and quack cures, right? After all, drug companies
have to produce not just any study, showing some possibly misleading correlation, but a
blind randomized clinical trial. That is, patients must be randomly assigned to treatments,
must not be told what treatment they are getting, and statistics must then determine if any
differences are “significant.”
Even then three serious problems remain. First, drug companies often do a bunch of
trials, but only report the trials that make their drug look good, invalidating the usual
statistics. Second, drug companies usually run best case trials, on patients most likely to
benefit and under doctors most likely to make good choices. The fact that the best patients
under the best doctors benefit may not say much about how ordinary patients and doctors
will fare. Third, patients can often tell if they are getting the drug being tested instead of
a sugar pill, because most drugs tested have side effects, such as dizziness or dry mouth.
Patients who experience side effects are then assured that they are getting the real treatment,
producing a placebo effect (Kirsch & Sapirstein, 1998).
Well, sure, you can always be skeptical if you try hard enough. But is there any concrete
reason to think that we do not typically benefit from medical treatment? Actually, for
decades researchers have worked to measure the average health effects of medicine. Some of
these studies look at correlations out in the world, seeing if people who get more medical
care tend to be healthier, all else equal.
Correlations-in-the-world studies see many apparently large influences on health, includ-
ing age, gender, exercise, social status, urban location, smoking, sleep, and even church
attendance. Gender, exercise and social status, for example, can change lifespans by ten to
fifteen years or more (Lantz, House, Lepkowski, Williams, Mero, & Chen, 1998). When one
looks at medical spending, however, the usual2 finding in such studies is no effect. When
comparing nations or counties or individuals, people who get more medicine have no signif-
icant difference in health when compared to people who get less medicine.
For example, Jonathan Skinner and John Wennberg looked at five million Medicare
patients in about 3500 hospital service areas in 1990. They looked to see if people died
less in areas where, during the last six months of life, Medicare spent more on treatment.
2
Random errors and bad statistical modeling assumptions should and do give a few contrary results.
3
After controlling for age, sex, race, median income, poverty, education, urbanization, and
initial health limitations, Skinner and Wennberg found that the average effect of spending
$1000 more was somewhere between increasing lifespan by about five days and decreasing
it by about fifteen days.3 Areas that kept patients in the intensive care unit one more day
on average reduced lifespan by between fifteen and seventy five days (Skinner & Wennberg,
2000).
Of course all these correlations-in-the-world studies have flaws. It is never entirely clear
what is the cause and what is the effect, and one can always think the results would have
changed had more variables been included in the study. This is why randomized trials are so
important; randomization controls for everything. And it turns out that we have at least one
large randomized trial on the overall health effects of medicine: the RAND health insurance
experiment.
In the late 1970s, most of 5816 non-elderly adults4 from six U.S. cities were randomly
assigned for three to five years to one of two conditions. Either they had free health care, or
they had to pay a substantial fraction (ranging from 25 to 95 percent) of their health care
costs. People with free health care had more doctor and hospital visits, and those with free
care spent about 75% more than those who paid nearly full price. While this sample was
too small to see effects on death rates, the plan was to look at a general health index based
on over twenty health indicators (Newhouse & Group, 1993).
The bottom line: no significant difference in general health was seen between those with
more health care and those with less. And since this was a randomized, but not blind, clinical
trial, this no-effect result includes any placebo effects, suggesting that free care would look
even worse without placebo effects.
Breaking the data down by four subgroups, the closest result the RAND experiment
found to being significant was that free medicine made initially-well poor people sicker (6%
significant). Breaking the data eighty ways by health indicators, the RAND experiment
found these 5% significant results: free health care produced better vision from free eye-
glasses, fewer decayed and more filled teeth in kids from free dental care, and lower blood
pressure from regular testing.5 Free care also induced more restricted-activity days, when
people could not follow their normal routines. In high risk people, free care made their acne
and hearing worse.
Thus while there may be health benefits from wearing eyeglasses, filling decayed teeth,
and regularly checking blood pressure, the RAND experiment found no net health benefit
from all the other health care that having free care induced people to get. While some
subgroups may have benefited, either those groups were very small, or other subgroups
suffered harms of a similar size.
The RAND experiment and all the other correlations-in-the-world studies only speak
3
To convert risk ratios to lifespans, I use the estimate that risk ratios double roughly every decade.
4
They included more poor people than a random sample would contain.
5
After the experiment some researchers put together a “risk of dying” index based on smoking, cholesterol
and blood pressure. They found no significant effect overall (between eight months longer and three months
shorter lives), but for a certain high-index group lifespan increased by about a year (3% significant).
4
directly to the health effects of “marginal” medical care, i.e., the care that the studies saw
some people get and other people not get. But the RAND experiment does give us some
clues about the health effects of common care, the care that everyone in their experiment got.
Common care (the first 60% of spending for those with free care) and marginal care (the last
40% of spending for those with free care) had the same fraction of “inappropriate” hospital
admissions and care-days, as judged by doctors reviewing cases, and also the same fraction
of ma jor and catastrophic disease presentations, relative to moderate and asymptomatic
manifestations.
Thus it is clearly not the case that marginal care is known to doctors to be less useful and
more frivolous, while the serious situations where doctors know medicine is very valuable are
usually in common care. In fact, doctors do not seem to see a difference between common
and marginal care. So, if common care is much more useful on average than the useless-on-
average marginal care, it must because each patient somehow knows something that doctors
cannot see about when he really needs to see a doctor. Now how likely is that?
But what about those miracles of modern medicine we have all heard so much about.
Did not the introduction of antibiotics, for example, dramatically reduce death rates for
key diseases? Well, not much actually. If you look at graphs of death rates versus time
for those key diseases, you usually cannot see any change in the trend near the dates when
famous treatments were introduced to “cure” those diseases (McKinlay & McKinlay, 1977).
Lifespans have more than doubled in the last century or so, and medicine cannot claim credit
for much more than a year or so of that increase (Bunker, Frazier, & Mosteller, 1994).
If medicine for treating individuals is not quite the miracle we have heard, does public
health make up the difference? Have not we all heard how the introduction of modern water
and sewer systems greatly improved our health? Well, a century ago the U.S. cities with the
most advanced water and sewer systems had higher death rates than the other cities. Also,
we can look today at how individual water sources and sewer mechanisms correlate with
death rates. Even in poor countries with high death rates, once we control for a few other
variables like social status we usually find that water and sewer parameters are unrelated to
death rates (Lee, Rosenzweig, & Pitt, 1997).
Well we must live longer now for some reason, right? Yes, and in fact in the developed
nations it seems that age specific death rates have fallen at a steady exponential rate for at
least a century (Lee & Carter, 1992; Tuljapurkar, Li, & Boe, 2000). But the fact is that we
do not know why we now live so much longer.
Most of the obvious theories have serious problems, you see. For example, exercise,
smoking, social status, and urban living appear to have large effects on individuals. But the
time trends for exercise, urban living, and smoking have been in the wrong directions for
them to explain increasing lifespan. And since social status is usually thought be relative to
contemporaries, so it is hard to see how average social status can increase with time.
Finally, while biggest advances in nutrition, medicine, and public health seem to have
occured during the first two thirds of the twentieth century, death rates have fallen just
as fast during the last third of the twentieth century. Perhaps some new influence rose
in important just as those other influences became less important, but if so it seems a
5
remarkable coincidence that the total rate of improvement has remained pretty steady.
Everything I have told you here has long been well known among the people who specialize
in studying these things. None of this is particular new or controversial to them. And yet
most of it is probably news, perhaps even shocking news, to most of you. Why is this?
The fear of death is a powerful influence on our thinking, even if we are not often conscious
of it. Our society, like all others before it, has a strong need to feel in control of death, even
if we must embrace fairy tales and quack cures to gain that sense of control. The idea that
we mostly do not understand and cannot control death is just not a message that people
want to hear. The message that medical miracles can control death, in contrast, is a message
that people do want to hear.
So we now spend 15% of our national income on medicine, even though half of that
spending has been clearly demonstrated to be on average useless, and even though we have
good reasons to doubt the value of most of the other half. Furthermore, we seem relatively
uninterested in living longer by trying the things that our evidence suggests do work, like
gaining high social status, exercising more, smoking less, and living in rural areas. Such
apparently effective approaches to increasing lifespan just do not have the magic allure of
conquering death via medical miracles. We know that exercise will not save us when we are
dying, in the way that we hope that medicine might.
Unfortunately, even if you believe everything that I have said, your behavior will probably
not change much as a result. You will still spend nearly as much on medicine for yourself
and your family, and spend much less effort on the more effective ways to increase lifespan.
After all, your family would consider it the worst kind of betrayal if you did not provide
them with all the medicine that your doctor recommends (Hanson, 2002). Alas, the problem
of the fear of death muddling our thinking is so much worse than we imagined.
References
Atran, S., & Norenzayan, A. (2004). Why minds create gods: Devotion, deception, death,
and arational decision making. Behavioral and Brain Sciences, 27, 713–770.
Becker, E. (1973). The Denial of Death. Free Press, New York.
Bunker, J. P., Frazier, H. S., & Mosteller, F. (1994). Improving Health: Measuring Effects
of Medical Care. Milbank Quarterly, 72 (2), 225–258.
Hanson, R. (2002). Why Health Is Not Special: Errors in Evolved Bioethics Intuitions.
Social Philosophy and Policy, 19 (2), 153–179.
Kirsch, I., & Sapirstein, G. (1998). Listening to Prozac but Hearing Placebo: A Meta-
Analysis of Antidepressant Medication. Prevention & Treatment, 1 (2a).
Kopczuk, W., & Slemrod, J. (2005). Denial of Death and Economic Be-
havior. Tech. rep. 11485, National Bureau of Economic Research.
http://papers.nber.org/papers/w11485.
6
Lantz, P., House, J., Lepkowski, J., Williams, D., Mero, R., & Chen, J. (1998). Socioeconomic
Factors, Health Behaviors, and Mortality, Results From a Nationally Representative
Study of US Adults. Journal of the American Medical Association, 279 (21), 1703–1746.
Lee, L., Rosenzweig, M., & Pitt, M. (1997). The effects of improved nutrition, sanitation, and
water quality on child health in high-mortality populations. Journal of Econometrics,
77, 209–235.
Lee, R. D., & Carter, L. R. (1992). Modeling and Forecasting U. S. Mortality. Journal of
the American Statistical Association, 87 (419), 659–671.
McKinlay, J., & McKinlay, S. (1977). The Questionable Contribution of Medical Measures
to the Decline of Mortality in the United States in the Twentieth Century. Milbank
Quarterly, 55, 405–428.
Newhouse, J. P., & Group, T. I. E. (1993). Free for Al l? Lessons from the RAND Health
Insurance Experiment. Harvard University Press, Cambridge, Massachusetts.
Norenzayan, A., & Hansen, I. G. (2006). Belief in Supernatural Agents in the Face of Death.
Personality and Social Psychology Bul letin.
Skinner, J., & Wennberg, J. E. (2000). How Much Is Enough? Efficiency and Medicare
Spending in the Last Six Months of Life. In Cutler, D. (Ed.), The Changing Hospi-
tal Industry: Comparing Not-for-Profit and For-Profit Institutions, pp. 169–193. The
University of Chicago Press.
Soloman, S., Greenberg, J., & Pyszczynski, T. (2000). Pride and Prejudice: Fear of Death
and Social Behavior. Current Directions in Psychological Research, 9 (6), 200–204.
Tuljapurkar, S., Li, N., & Boe, C. (2000). A universal pattern of mortality decline in the G7
countries. Nature, 405, 789–792.
Welch, P., Miller, M., Welch, H. G., Fisher, E., & Wennberg, J. (1993). Geographic Variations
in expenditures for physicians’ services in the United States. New England Journal of
Medicine, 328 (9), 621–627.
7
