Shot in the dark: why doctors know less than you think.

How Doctors Think By Jerome Groopman Houghton Mifflin, 320 pp.

When Anne Dodge was twenty years old, food suddenly stopped agreeing with her. She would get hungry, she would eat, then a feeling like something gripping her stomach would come over her, and she would retreat to the bathroom to vomit. Suspecting a mental disorder, her primary-care physician referred her to a psychiatrist, who diagnosed anorexia nervosa with bulimia. But despite weekly therapy sessions, Anne continued to lose weight.

As her health deteriorated, she consulted more and more specialists: hematologists because her red blood cell count was perilously low; nutritionists because she was malnourished; orthopedists when she developed osteoporosis. When she began suffering from bouts of constipation alternating with diarrhea, various gastroenterologists diagnosed irritable bowel syndrome, a disease of unknown etiology and even more uncertain treatment. By the time Dodge was thirty years old, she had seen at least two dozen different specialists. She was put on a 3,000-calorie-a-day diet that consisted of mostly bread and pasta, which she swore she was forcing down, yet her doctors suspected she was lying as she continued to lose weight. By December 2004, Dodge was down to eighty-two pounds and deeply despondent about her condition. At her boyfriend's insistence, she drove into Boston shortly after Christmas Day for a consultation with yet another specialist.

This specialist was different. Before performing a physical exam, Dr. Myron Falchuk, a gastroenterologist at Boston's Beth Israel Deaconess Medical Center, asked Dodge to do something no other physician had requested. "Let's go back to the beginning," he said. "Tell me about when you first didn't feel good." The longer Falchuk observed Anne Dodge, writes Jerome Groopman, "the more he listened to her, the more disquiet he felt." Everyone else had written her off as some neurotic, a mental case who not only wasn't eating enough but lied about it, yet Falchuk's intuition told him the picture didn't fit. And once he felt that way, he told Groopman, "I began to wonder: What was missing?" As it turned out, Dodge did not have anorexia nervosa, or bulimia, or irritable bowel syndrome. In asking this young woman simply to recount her story, Falchuk was able to come to the diagnosis that had eluded the specialists who came before him. He was able to restore her to health.

In telling Falchuk's tale, and those of other difficult-to-diagnose cases, Groopman offers a riveting view of the intricate and not always logical thought processes that lie behind the decisions that doctors make. On average, a doctor interrupts a patient who is describing her symptoms within about twelve seconds. By then, many doctors have decided upon a diagnosis and treatment. Often they're right, and the ability to "shoot from the hip" is much admired in medical circles, especially in high-pressure specialties like emergency medicine and radiology. But Groopman, who is himself a physician as well as an accomplished writer, is more interested in cases when doctors get it wrong. In How Doctors Think, he uses a series of case histories and interviews with physicians and patients to explore the nature of diagnostic error, and to come up with suggestions for how both doctors and patients can avoid it.

Anne Dodge's various physicians, for instance, had fallen prey to a "framing" error; as soon as the first physician labeled her with anorexia, each subsequent doctor viewed her various symptoms within that framework, and looked no further for an alternate diagnosis--at least not until she met Falchuk. Attribution error was behind an emergency physician's failure to diagnose unstable angina in a forest ranger in his forties, Unstable angina provokes a crescendo of chest pain that often presages a heart attack, but in this case, the patient appeared far too fit and trim to be suffering from heart disease. After performing several tests that seemed to rule out angina, the doctor concluded that the shooting pain in his patient's chest was nothing more serious than a pulled muscle, and he sent the man home. Such mistakes are called attribution errors because doctors attribute the symptoms they see to the wrong diagnosis. The doctor only learned of his mistake when the patient was admitted to his emergency room the next day, in the midst of a full-blown heart attack. Then there's the type of error that results from confirmation bias, a kind of "cognitive cherry-picking," as Groopman puts it, the tendency to see only the symptoms and test results that support the presumed diagnosis.

With this multitude of ways to be led astray, how on earth do doctors ever manage to come up with the right diagnosis? They are most likely to do so, in Groopman's view, by maintaining a sense of humility in the face of the deep and abiding uncertainty of medicine. For all of modern medicine's high-tech gadgetry and sophisticated tests, doctors are still limited by their lack of knowledge about the body and all of its various ailments. The causes of many kinds of cancer, for example, remain an enigma, as does heart disease in a patient like the forest ranger, who was young and fit. Doctors only recently learned that the ailment afflicting Anne Dodge is remarkably widespread. Most of the time, they don't know the source of lower back pain--whether it's caused by strained muscles, a damaged disk, or a pinched nerve. On top of all of this, every human body is slightly different, and no two people respond in exactly the same way to the same drug or medical procedure. Doctors face a vast sea of unknowns, and Groopman's obvious compassion for this potentially unnerving state of uncertainty is evident in his stories. (Indeed, Groopman comes through so clearly in this book and his previous work, The Anatomy of Hope, as the kind of doctor we all wish we had. He's a cancer specialist who is at once intelligent, kind, intuitive, and profoundly comforting in the face of an often desperate illness.)

The other form of uncertainty in medicine is more pernicious. Many if not most of the treatments that medicine has to offer are not based on scientific evidence. Back surgeons, for example, have no idea if a commonly performed, potentially risky--and highly lucrative--surgical procedure called spinal fusion is actually any better at relieving lower back pain than nonsurgical remedies. Yet they keep doing spinal fusions, though many of their patients wind up no better than they were before the surgery--while some are left in worse shape. There's still no clear evidence which form of treatment for prostate cancer is most effective. Is minimally invasive hip replacement better than standard surgery? Does lung reduction surgery reduce the risk of dying from emphysema? Nobody knows, despite recent headlines, whether or not screening for lung cancer with CAT scans will save lives. The list of what we don't know about new as well as commonly performed procedures and widely prescribed drugs goes on and on.

The notion that modern medicine isn't grounded in science is probably hard for many readers to believe. We want to trust that our doctors know what they're doing, that our surgeon is certain that the painful, risky procedure we are about to undergo is likely to restore us to health, that our internist can point to solid scientific research showing that the drug she is about to prescribe will improve our symptoms. Yet as Groopman makes clear, a great deal of medical knowledge is based not on careful testing, but on supposition and belief. It's handed down, "as if from the heavens," as he pus it, during every young doctor's apprenticeship. Different physicians often treat the same condition in different ways, depending on where they were trained, or what they happened to hear from a respected authority in the field.

Yet all too often, physicians act as if there isn't a doubt in their minds about what ails a body, and what to do about it. Groopman recounts a conversation with a spine surgeon who told him he was a "believer" in spinal fusion. "My outcomes are better than anything in the published literature," the surgeon boasted. Yet when pressed, writes Groopman, the man confessed that he rarely followed up on patients over the long term; nor had he participated in any randomized controlled clinical trials--the gold standard of evidence-based medicine--to test whether or not spinal fusion is more effective than nonsurgical therapy. In many ways, this is the most surprising aspect of all when it comes to the question of how doctors think. Though medicine has wrapped itself in the mantle of science, most doctors aren't trained to think very scientifically--especially when doing so is in conflict with their financial interests.

The obvious remedy for the lacunae in medical evidence is more clinical research. Currently, the National Institutes of Health barely look at the efficacy of existing medical practices. The Food and Drug Administration generally requires only minimal testing of medical devices. And the industry-funded clinical drug trials the FDA does oversee generally compare new drugs to placebos, not to other similar drugs to see which ones work better. As I and others have written in these pages and elsewhere, the United States needs a federally funded institute, dedicated to research that will show what works in medicine, what doesn't, and for which patients.

Putting all of medicine on firmer scientific footing will require dozens of trials, and millions of dollars in funding. Until then, patients can help their doctors at least come to the right diagnosis, even if they don't always know what treatment is best. Near the end of How Doctors Think, Groopman tells the story of a man who consulted six different hand surgeons and received four different opinions about the source of a painful, debilitating problem with his right hand. The patient was Groopman himself, and his three-year odyssey in search of effective treatment is a testament to the importance of avoiding physicians who jump to conclusions, or ignore symptoms that don't happen to fit their predetermined notion of what's wrong. Read this eloquent little book to enjoy Groopman's fine storytelling, or to find out how he regained the use of his hand and what disease afflicted Anne Dodge. But read it especially if you want to learn how to encourage your own doctors to arrive at the right diagnosis.

Shannon Brownlee is a Schwartz senior fellow at the New America Foundation and a health care policy writer. Her new book, Over-treated: Why Too Much Medicine Is Making Americans Sicker and Poorer, will be published by Bloomsbury Press in 2007.