by Atul Gawande
He was right. No more surgery, Sykes told me. He’d been through enough. We called Mrs. Sykes, who was staying with a friend about two hours away, and she set out for the hospital.
It was about midnight. I sat with him as he lay silent and bleeding, his arms slack at his sides, his eyes without fear. I imagined his wife out on the Mass Pike, frantic, helpless, with six lanes, virtually empty at that hour, stretching far ahead.
Sykes held on, and at 2:15 A.M. his wife arrived. She turned ashen at the sight of him, but she steadied herself. She gently took his hand in hers. She squeezed, and he squeezed back. I left them to themselves.
At 2:45, the nurse called me in. I listened with my stethoscope, then turned to Mrs. Sykes and told her that he was gone. She had her husband’s Yankee reserve, but she broke into quiet tears, weeping into her hands, and seemed suddenly frail and small. A friend who had come with her soon appeared, took her by the arm, and led her out of the room.
We are instructed to request an autopsy on everyone as a means of confirming the cause of death and catching our mistakes. And this was the moment I was supposed to ask—with the wife despondent and reeling with shock. But surely, I began to think, here was a case in which an autopsy would be pointless. We knew what had happened—a persistent infection, a rupture. We were sure of it. What would cutting the man apart accomplish?
And so I let Mrs. Sykes go. I could have caught her as she walked through the ICU’s double doors. Or even called her on the phone later. But I never did.
Such reasoning, it appears, has become commonplace in medicine. Doctors are seeking so few autopsies that in recent years the Journal of the American Medical Association has twice felt the need to declare “war on the nonautopsy.” According to the most recent statistics available, autopsies have been done in fewer than 10 percent of deaths; many hospitals do none. This is a dramatic turnabout. Through much of the twentieth century, doctors diligently obtained autopsies in the majority of all deaths—and it had taken centuries to reach this point. As Kenneth Iserson recounts in his fascinating almanac, Death to Dust, physicians have performed autopsies for more than two thousand years. But for most of history they were rarely performed. If religions permitted them at all—Islam, Shinto, orthodox Judaism, and the Greek Orthodox Church still frown on them—it was generally only for legal purposes. The Roman physician Antistius performed one of the earliest forensic examinations on record, in 44 B.C., on Julius Caesar, documenting twenty-three wounds, including a final, fatal stab to the chest. In 1410, the Catholic Church itself ordered an autopsy—on Pope Alexander V, to determine whether his successor had poisoned him. No evidence of this was apparently found.
The first documented postmortem examination in the New World was actually done for religious reasons, though. It was performed on July 19, 1533, on the island of Espanola (now the Dominican Republic), upon conjoined female twins connected at the lower chest, to determine if they had one soul or two. The twins had been born alive, and a priest had baptized them as two separate souls. A disagreement subsequently ensued about whether he was right to have done so, and when the “double monster” died at eight days of age an autopsy was ordered to settle the issue. A surgeon, one Johan Camacho, found two virtually complete sets of internal organs, and it was decided that two souls had lived and died.
Even in the nineteenth century, however, long after church strictures had loosened, people in the West seldom allowed doctors to autopsy their family members for medical purposes. As a result, the practice was largely clandestine. Some doctors went ahead and autopsied hospital patients immediately after death, before relatives could turn up to object. Others waited until burial and then robbed the graves, either personally or through accomplices, an activity that continued into the twentieth century. To deter such autopsies, some families would post nighttime guards at the grave site—hence the term “graveyard shift.” Others placed heavy stones on the coffins. In 1878, one company in Columbus, Ohio, even sold “torpedo coffins,” equipped with pipe bombs rigged to blow up if they were tampered with. Yet doctors remained undeterred. Ambrose Bierce’s The Devil’s Dictionary, published in 1906, defined “grave” as “a place in which the dead are laid to await the coming of the medical student.”
By the turn of the twentieth century, however, prominent physicians such as Rudolf Virchow in Berlin, Karl Rokitansky in Vienna, and William Osler in Baltimore began to win popular support for the practice of autopsy. They defended it as a tool of discovery, one that had already been used to identify the cause of tuberculosis, reveal how to treat appendicitis, and establish the existence of Alzheimer’s disease. They also showed that autopsies prevented errors—that without them doctors could not know when their diagnoses were incorrect. Moreover, most deaths were a mystery then, and perhaps what clinched the argument was the notion that autopsies could provide families with answers—give the story of a loved one’s life a comprehensible ending. Once doctors had insured a dignified and respectful dissection at the hospital, public opinion turned. With time, doctors who did not obtain autopsies were viewed with suspicion. By the end of the Second World War, the autopsy was firmly established as a routine part of death in Europe and North America.
So what accounts for its decline? In truth, it’s not because families refuse—to judge from recent studies, they still grant that permission up to 80 percent of the time. Instead, doctors, once so eager to perform autopsies that they stole bodies, have simply stopped asking. Some people ascribe this to shady motives. It has been said that hospitals are trying to save money by avoiding autopsies, since insurers don’t pay for them, or that doctors avoid them in order to cover up evidence of malpractice. And yet autopsies lost money and uncovered malpractice when they were popular, too.
Instead, I suspect, what discourages autopsies is medicine’s twenty-first-century, tall-in-the-saddle confidence. When I failed to ask Mrs. Sykes whether we could autopsy her husband, it was not because of the expense, or because I feared that the autopsy would uncover an error. It was the opposite: I didn’t see much likelihood that an error would be found. Today, we have MRI scans, ultrasound, nuclear medicine, molecular testing, and much more. When somebody dies, we already know why. We don’t need an autopsy to find out.
Or so I thought. Then I had a patient who changed my mind.
He was in his sixties, whiskered and cheerful, a former engineer who had found success in retirement as an artist. I will call him Mr. Jolly, because that’s what he was. He was also what we call a vasculopath—he did not seem to have an undiseased artery in him. Whether because of his diet or his genes or the fact that he used to smoke, he had had, in the previous decade, one heart attack, two abdominal aortic aneurysm repairs, four bypass operations to keep blood flowing past blockages in his leg arteries, and several balloon procedures to keep hardened arteries open. Still, I never knew him to take a dark view of his lot. “Well, you can’t get miserable about it,” he’d say. He had wonderful children. He had beautiful grandchildren. “But, aargh, the wife,” he’d go on. She would be sitting right there at the bedside and would roll her eyes, and he’d break into a grin.
Mr. Jolly had come into the hospital for treatment of a wound infection in his legs. But he soon developed congestive heart failure, causing fluid to back up into his lungs. Breathing became steadily harder for him, until we had to put him in the ICU, intubate him, and place him on a ventilator. A two-day admission turned into two weeks. With a regimen of diuretics and a change in heart medications, however, his heart failure reversed, and his lungs recovered. And one bright Sunday morning he was reclining in bed, breathing on his own, watching the morning shows on the TV set that hung from the ceiling. “You’re doing marvelously,” I said. I told him we would transfer him out of intensive care by the afternoon. He would probably be home in a couple of days.
Two hours later, a code-blue emergency call went out on the overhead speakers. When I got to the ICU and saw the nurse hunched over Mr. Jolly, doing chest compression
s, I blurted out an angry curse. He’d been fine, the nurse explained, just watching TV, when suddenly he sat upright with a look of shock and then fell back, unresponsive. At first, he was asystolic—no heart rhythm on the monitor—and then the rhythm came back, but he had no pulse. A crowd of staffers set to work. I had him intubated, gave him fluids and epinephrine, had someone call the attending surgeon at home, someone else check the morning lab test results. An X-ray technician shot a portable chest film.
I mentally ran through possible causes. There were not many. A collapsed lung, but I heard good breath sounds with my stethoscope, and when his X ray came back the lungs looked fine. A massive blood loss, but his abdomen wasn’t swelling, and his decline happened so quickly that bleeding just didn’t make sense. Extreme acidity of the blood could do it, but his lab tests were fine. Then there was cardiac tamponade—bleeding into the sac that contains the heart. I took a six-inch spinal needle on a syringe, pushed it through the skin below the breastbone, and advanced it to the heart sac. I found no bleeding. That left only one possibility: a pulmonary embolism—a blood clot that flips into the lung and instantly wedges off all blood flow. And nothing could be done about that.
I went out and spoke to the attending surgeon by phone and then to the chief resident, who had just arrived. An embolism was the only logical explanation, they agreed. I went back into the room and stopped the code. “Time of death: 10:23 A.M.,” I announced. I phoned his wife at home, told her that things had taken a turn for the worse, and asked her to come in.
This shouldn’t have happened; I was sure of it. I scanned the records for clues. Then I found one. In a lab test done the day before, the patient’s clotting had seemed slow, which wasn’t serious, but an ICU physician had decided to correct it with vitamin K. A frequent side effect of vitamin K is blood clots. I was furious. Giving the vitamin was completely unnecessary—just fixing a number on a lab test. Both the chief resident and I lit into the physician. We all but accused him of killing the patient.
When Mrs. Jolly arrived, we took her to a family room where it was quiet and calm. I could see from her face that she’d already surmised the worst. His heart had stopped suddenly, we told her, because of a pulmonary embolism. We said the medicines we gave him may have contributed to it. I took her in to see him and left her with him. After a while, she came out, her hands trembling and her face stained with tears. Then, remarkably, she thanked us. We had kept him for her all these years, she said. Maybe so, but neither of us felt any pride about what had just happened.
I asked her the required question. I told her that we wanted to perform an autopsy and needed her permission. We thought we already knew what had happened, but an autopsy would confirm it, I said. She considered my request for a moment. If an autopsy would help us, she finally said, then we could do it. I said, as I was supposed to, that it would. I wasn’t sure I believed it.
I wasn’t assigned to the operating room the following morning, so I went down to observe the autopsy. When I arrived, Mr. Jolly was already laid out on the dissecting table, his arms splayed, skin flayed back, chest exposed, abdomen open. I put on a gown, gloves, and a mask, and went up close. The assistant began buzzing through the ribs on the left side with the electric saw, and immediately blood started seeping out, as dark and viscous as crankcase oil. Puzzled, I helped him lift open the rib cage. The left side of the chest was full of blood. I felt along the pulmonary arteries for a hardened, embolized clot, but there was none. He hadn’t had an embolism after all. We suctioned out three liters of blood, lifted the left lung, and the answer appeared before our eyes. The thoracic aorta was almost three times larger than it should have been, and there was a half-inch hole in it. The man had ruptured an aortic aneurysm and had bled to death almost instantly.
In the days afterward, I apologized to the physician I’d reamed out over the vitamin, and pondered how we had managed to miss the diagnosis. I looked through the patient’s old X rays and now saw a shadowy outline of what must have been his aneurysm. But none of us, not even the radiologists, had caught it. Even if we had caught it, we wouldn’t have dared to do anything about it until weeks after treating his infection and heart failure, and that would have been too late. It disturbed me, however, to have felt so confident about what had happened that day and to have been so wrong.
The most perplexing thing was his final chest X ray, the one we had taken during the code blue. With all that blood filling the chest, I should have seen at least a haze over the left side. But when I pulled the film out to look again, there was nothing.
How often do autopsies turn up a major misdiagnosis in the cause of death? I would have guessed this happened rarely, in 1 or 2 percent of cases at most. According to three studies done in 1998 and 1999, however, the figure is about 40 percent. A large review of autopsy studies concluded that in about a third of the misdiagnoses the patients would have been expected to live if proper treatment had been administered. George Lundberg, a pathologist and former editor of the Journal of the American Medical Association, has done more than anyone to call attention to these figures. He points out the most surprising fact of all: the rates at which misdiagnosis is detected in autopsy studies have not improved since at least 1938.
With all the recent advances in imaging and diagnostics, it’s hard to accept that we not only get the diagnosis wrong in two out of five of our patients who die but that we have also failed to improve over time. To see if this could really be true, doctors at Harvard put together a simple study. They went back into their hospital records to see how often autopsies picked up missed diagnoses in 1960 and 1970, before the advent of CT, ultrasound, nuclear scanning, and other technologies, and then in 1980, after those technologies became widely used. The researchers found no improvement. Regardless of the decade, physicians missed a quarter of fatal infections, a third of heart attacks, and almost two-thirds of pulmonary emboli in their patients who died.
In most cases, it wasn’t technology that failed. Rather, the physicians did not consider the correct diagnosis in the first place. The perfect test or scan may have been available, but the physicians never ordered it.
In a 1976 essay, the philosophers Samuel Gorovitz and Alasdair MacIntyre explored the nature of fallibility. Why would a meteorologist, say, fail to correctly predict where a hurricane was going to make landfall? They saw three possible reasons. One was ignorance: perhaps science affords only a limited understanding of how hurricanes behave. A second reason was ineptitude: the knowledge is available, but the weatherman fails to apply it correctly. Both of these are surmountable sources of error. We believe that science will overcome ignorance, and that training and technology will overcome ineptitude. The third possible cause of error the philosophers posited, however, was an insurmountable kind, one they termed “necessary fallibility.”
There may be some kinds of knowledge that science and technology will never deliver, Gorovitz and MacIntyre argued. When we ask science to move beyond explaining how things (say, hurricanes) generally behave to predicting exactly how a particular thing (say, Thursday’s storm off the South Carolina coast) will behave, we may be asking it to do more than it can. No hurricane is quite like any other hurricane. Although all hurricanes follow predictable laws of behavior, each one is continuously shaped by myriad uncontrollable, accidental factors in the environment. To say precisely how one specific hurricane will behave would require a complete understanding of the world in all its particulars—in other words, omniscience.
It’s not that it’s impossible to predict anything; plenty of things are completely predictable. Gorovitz and MacIntyre give the example of a random ice cube in a fire. Ice cubes are so simple and so alike that you can predict with complete assurance that an ice cube will melt. But when it comes to inferring exactly what is going on in a particular person, are people more like ice cubes or like hurricanes?
Right now, at about midnight, I am seeing a patient in the emergency room, and I want to say that she is an ice cube. T
hat is, I believe I can understand what’s going on with her, that I can discern all her relevant properties. I believe I can help her.
Charlotte Duveen, as we will call her, is forty-nine years old, and for two days she has had abdominal pain. I begin observing her from the moment I walk through the curtains into her room. She is sitting cross-legged in the chair next to her stretcher and greets me with a cheerful, tobacco-beaten voice. She does not look sick. No clutching the belly. No gasping for words. Her color is good—neither flushed nor pale. Her shoulder-length brown hair has been brushed, her red lipstick neatly applied.
She tells me the pain started out crampy, like a gas pain. But then, during the course of the day, it became sharp and focused, and as she says this she points to a spot in the lower right part of her abdomen. She has developed diarrhea. She constantly feels as if she has to urinate. She doesn’t have a fever. She is not nauseated. Actually, she is hungry. She tells me that she ate a hot dog at Fen-way Park two days ago and visited the exotic birds at the zoo a few days before that, and she asks if either might have anything to do with this. She has two grown children. Her last period was three months ago. She smokes half a pack a day. She used to use heroin but says she’s clean now. She once had hepatitis. She has never had surgery.
I feel her abdomen. It could be anything, I think: food poisoning, a virus, appendicitis, a urinary-tract infection, an ovarian cyst, a pregnancy. Her abdomen is soft, without distension, and there is an area of particular tenderness in the lower right quadrant. When I press there, I feel her muscles harden reflexively beneath my fingers. On the pelvic exam, her ovaries feel normal. I order some lab tests. Her white blood cell count comes back elevated. Her urinalysis is normal. A pregnancy test is negative. I order an abdominal CT scan.
