Chances Are (29 page)

Gerd Gigerenzer, in
Calculated Risks,
posed a simple question to doctors in a German teaching hospital: You're in charge of a mammogram screening program covering women between 40 and 50 who show no symptoms. You know that the overall probability that a woman of this age has breast cancer is 0.8 percent. If a woman
has
breast cancer, the probability that she will show a positive mammogram result is 90 percent. If she does
not
have breast cancer, the probability that she will have a falsely positive mammogram result is 7 percent. Your patient, Ursula K., has a positive mammogram result. What is the probability that she has breast cancer?

The doctors were baffled: a third of them decided the probability was 90 percent; a sixth thought it was 1 percent. It would have made a big difference to Ursula K. which doctor was on duty the day she came in for her results.

As Gigerenzer explains it, the problem is not with the doctors but with the percentages. Phrase the question again, using plain numbers: out of 1,000 women of this age, 8 will have breast cancer. When you screen those 8, 7 will have a positive mammogram result. When you screen the remaining 992 women
without
breast cancer, 69 or 70 of them will have a false-positive mammogram. Ursula K. is one of the 7 + 70 women with a positive result; how likely is she to have breast cancer?

It is a lot easier to compare 7 and 77 than to figure out

When the problem was explained this way, half the doctors got the answer right: Ursula's chance of having cancer is less than 1 in 10. That said, half of them still got it wrong, and two even said her chance of having cancer was 80 percent. Maybe part of the problem
is
the doctors.

 

So far, we have been talking about patients and their diseases as the raw data—but in a modern health-care system doctors, too, are objects of collective scrutiny. Spending on health care has reached an annual level of $1.5 trillion in the United States; staving off mortality costs every American $5,267 a year—more than 14 percent of the gross domestic product. Meanwhile, the UK has had fifty years of a state-funded universal health-care system—the world's third-largest employer, after the Chinese army and the Indian railroads—which an anxious electorate alternately praises and abuses.

How do you gauge the success of such an enterprise? All lives eventually end: medicine wins many battles but must lose the war. Can you define “unnecessary deaths prevented”? Like existence, health care lacks an ultimate goal. Its paymasters, however, have to describe and regulate the movements of this vast collective organism—and their method is necessarily statistical.

The variables most often used to define quality of care are an uneasy mixture of the practical and the political. People want to be treated soon, so “time to diagnosis” and “time to surgery” are variables to be minimized. People want to know they are going to a good hospital, so it is important to publish mortality rates—but the rates must be risk-adjusted; otherwise, advanced critical-care facilities would rank around the level of a Levantine pest-house. Governments want to pay as little as they can for routine procedures; more complex treatments require more money. It's like a continuous clinical trial, with funding as the drug: where it does most good, dosage is increased; where the Null Hypothesis prevails, society can save money. The problem is that this experiment is neither randomized nor double-blind. Doctors and hospital administrators are entirely aware of the criteria and of their implications for future funding. It is as if, in a cross-over experiment, the patients in the control group were told, “You're getting placebo now, but if you show the right symptoms we'll switch you into treatment.” The very sources of data are given both the opportunity and the incentive to manipulate it.

Given this remarkable arrangement, it's surprising how few institutions have been caught fiddling, but the examples that have come to light are worrying enough. In the UK, some administrators reduced apparent waiting times for operations by finding out when patients were going on vacation and then offering appointments they knew wouldn't be taken up. Departments hired extra staff and canceled routine procedures for the one week they knew waiting-time figures were being collected. In the United States, the “diagnosis-related group” reimbursement system for Medicare/ Medicaid has produced what is called “DRG creep,” where conditions are “upcoded” to more complex and remunerative classes. Hospitals anxious to achieve good risk-adjusted mortality figures can do so by sending home the hopelessly moribund and classifying incoming patients as higher-risk: in one New York hospital the proportion of preoperative patients listed as having “chronic obstructive pulmonary disease” rose from 2 percent to 50 percent in two years. In heart surgery, adding a little valve repair to a bypass operation for a high-risk patient could take the whole procedure off the mortality list, improving the bypass survival figures at the expense of a few extra deaths in the “other” column.

Of course, many more frightening things happened in the swaggering days when health care was left to regulate itself. The point is that as long as funding depends on statistics, the temptation to doctor the numbers as well as the patients will be strong. Moreover, ranking asks us to do something all people find difficult: to accept that half of
any
ranking will be below the median. How will you feel when you learn that your surgeon ranks in the 47th percentile? Does that help cement the relationship of trust?

 

Jeremy Bentham described the role of society as providing the greatest good for the greatest number—a difficult ratio to maximize. The “good” of medical science, based on experiment and statistics, consists of matching potential cures to existing illnesses. This model worked well when the bully diseases were still in charge: the constant threats that filled up the middle of our normal curve of deaths. Now, smallpox is gone, polio almost gone, TB generally under control, measles manageable. We are increasingly faced with diseases that conceal huge variety under a single name, like cancer—or mass illnesses caused, on average, by our own choices, like obesity, diabetes, or heart disease. The problem with these isn't finding a cure—if ever there were a magic bullet, vigorous exercise would be it—it's being willing to take it.

A more easily swallowed remedy for the diseases of affluence is the Polypill, proposed in 2003 by Nicholas Wald and Malcolm Law of the Wolfson Institute of Preventative Medicine. Combining aspirin and folic acid with generic drugs that lower cholesterol and blood pressure, this would be given to everyone over 55, and (
assuming
the benefits are multiplicative) should cut the risk of heart attacks by 88 percent, and strokes by 80 percent. Average life span could be increased by 11 years at very little cost.

But some of those drugs have side effects; some people could have problems. So, again, the question is whether you think of
all
of your patients or
each
of them. A National Health Service, dealing with a whole population, would probably favor the Polypill, but American researchers say that variable response among different groups requires tailoring the dose—there would be at least 100 different Polypills. It seems that we will still need to jog up the steep path of self-denial.

“Patients very rarely fit the picture in the textbook. How do you treat an individual?” Dr. Timothy Gilligan of the Harvard Medical School is both a scientist and a practicing surgeon; his life is lived on the interface between the general and the particular. “In something like chemotherapy or radiotherapy, the published results don't tell us anywhere near enough. We are trying to take into account genetic variations in metabolizing certain drugs, or the effects of different social environments on whether someone can get through a difficult therapy successfully. Individual differences can make all the difference to the outcome.”

His hope is that increasing knowledge of the human genome will return medicine to the idea of a unique solution for every patient, custom-built around genetic predispositions. “Cancers are genetic diseases, and ultimately we should be able to define cancers by a series of specific mutations in genes. Right now, we have a hundred people coming in with lung cancer—which we
know
is a variety of diseases—and they all get put in the same basket. Some will respond to chemotherapy and some won't, and one of the reasons is probably the specific character of the cancer, its genetic component. If we understood that, we could tailor treatment to it.” For the moment, though, the complexity of the way the human body expresses its genome makes this still a distant dream.

Improved statistical evaluation may sharpen prognosis, however. Instead of being told that half the people with your disease die within a year, leaving you to wonder which half you are in, more sophisticated computer algorithms take account of several variables about your disease and give a more specific estimate. Dr. Gilligan elaborates: “You're an individual with this type of lung cancer and this was the size of your tumor and this is where the metastases are located, and this is how fit you are right now, and if we plug all these numbers into our computer we can say not just what everyone with lung cancer does, but what people like
you
do. Again, though, you end up with a percentage: it may be you have a 75 percent chance of living a year—but we still can't tell you whether you are in that 75 percent or the 25 percent who don't. We're a long way from the 100 percent or 0 percent that tells you you're going to be cured or you're not going to be cured—but if we have nasty chemotherapy and we are able to say this group of people has a 90 percent chance of benefiting and this group has only a 10 percent chance, then it would be easier to decide whom to treat.”

As the human genome reveals its secrets, many of our assumptions about it begin to unravel. The first to go seems to be the idea of a universal genome from which any mutation represents a potential illness. As methods of studying DNA improve both in their resolution and their signal-to-noise ratio, they reveal more and more variation between “normal” people—not just in the regions of apparent nonsense between known functional stretches, but in the placement and number of copies of functional genes themselves. How this variation affects the potential for disease, developmental differences, or response to drugs becomes a deepening mystery. So not only is there no Death, only I, who am dying—there may be no Malady, only I, who am sick, and no Treatment, only what might work for me.

Where does this leave us? Well short of immortality, but longer-lived and better cared-for than we were not so long ago, when a child on crutches meant polio, not a soccer injury. Our knowledge is flawed, but we can know the nature of its flaws. We take things awry, but we are learning something about the constants of our errors. If we remain aware that the conclusions we draw from data are inherently probabilistic and as interdependent as the balanced weights in a Calder mobile, we can continue the advance as better doctors—and as better patients. The inevitability of uncertainty is no more a reason for despair than the inevitability of death; medical research continues the mission set long ago by Fisher: “We have the duty of formulating, of summarizing, and of communicating our conclusions, in intelligible form, in recognition of the right of
other
free minds to utilize them in making
their own
decisions.”

8

Judging

—W. H. Auden, “Law Like Love”

 

 

 

 

 

A
round 1760 B.C.—a century or so after the departure of Abraham, his flocks, and family—Hammurabi established his supremacy in Mesopotamia, modern-day Iraq. He realized that leaving each mud-walled city under the protection of its own minor god and vassal kinglet was a sure prescription for treachery and disorder—so, as a sign of his supremacy, he gave to all one code of law: a seven-foot black pillar of basalt, densely covered with cuneiform writing, was set up in every marketplace.

Here is the source for “an eye for an eye, a tooth for a tooth” (articles 197 and 200), but here also are set the seller's liability for faulty slaves and the maximum damages for surgical malpractice. Much of the code seems astonishingly modern: it requires contracts, deeds, and witnesses for all transactions of sale or inheritance; merchant's agents must issue receipts for goods entrusted them; wives may tell husbands “you are not congenial to me” and, if otherwise blameless, go their ways in full possession of their dowries. On the other hand, there is also the full complement of loppings, burnings, drownings, and impalements typical of a society without effective police, where deterrence from serious crime rests on visceral fear of pain rather than certainty of detection.