The Paradox of Modern Medicine
In her first year of medical school, Diana Cejas discovered a lump in her neck. She went to the student medical center to have it evaluated and was told that it was likely benign. But the lump kept growing, and she returned to her doctors, who reassured her that it was just a large lymph node. One night, following a 36-hour shift in her residency, the lump hurt so much she couldn’t sleep. The next day, after she begged for help, a doctor finally ordered a CT scan. She looked up her results on the hospital computer system. There, on the screen, was a large mass in her neck. It turned out to be cancerous. Even as she had been learning how to correctly diagnose others, she had not been able to get an accurate diagnosis herself.
Cejas was a victim of what Alexandra Sifferlin, in her new book The Elusive Body: Patients, Doctors, and the Diagnosis Crisis, calls “multiple kinds of diagnostic error—both incorrect and delayed.” These types of errors are surprisingly common; an estimated 5 percent of Americans—about 13 million people—experience a diagnostic error each year. A 2023 study concluded that more than three-quarters of a million Americans are permanently disabled or die each year as the result of a misdiagnosis.
Diagnosis is, as Sifferlin writes, “the most important piece of medical information a person can receive.” In light of this, the health-care system’s failure to seriously investigate diagnostic error—which can include a wide range of mistakes—is both mystifying and dismaying.
In 2015, a landmark report from the National Academies of Science, Engineering, and Medicine found that most people will be incorrectly diagnosed at least once in their lifetime. The report was intended to catalyze reform, much as a groundbreaking 1999 analysis of medical errors led health-care institutions to reduce mistakes in areas such as surgery. But more than a decade after the NASEM report, little has changed, and Sifferlin writes that no major U.S. health system tracks diagnostic error systematically. When Hardeep Singh, a physician and patient-safety researcher, tried to start a national diagnostic-error-measurement program, he was able to get only nine hospital systems to sign up.
You might think that AI, with its copious processing power, would help solve this problem. But it’s unlikely to do so in the absence of a broad reimagining of how medicine engages with patients. Misdiagnosis, The Elusive Body shows, is not just an occasional glitch in the system but the result of several interrelated and intrinsic flaws—flaws that are, in surprising ways, the product of medical science’s evolution over the past century and a half.
[Read: Hypochondria never dies]
Chief among these weaknesses are the system’s obsession with metrics and testing and its expectation that patients’ bodies will behave roughly alike. The advent of germ theory in the 19th century introduced the idea of diseases as specific entities that cause distinct symptoms and antibodies, and new technologies such as the X-ray and laboratory blood work gave physicians the means to identify them. This was a marvelous advance. It brought us longer lives and eventually led to the idea of evidence-based medicine. But it turned testing into the focal point of the diagnostic process. The result is that doctors are less attuned to what patients describe than to what tests determine.
This problem is compounded by the system’s reluctance to engage in self-reflection, and the myth of the infallible doctor who just knows things. As Sifferlin documents, medical schools fail to teach doctors to be aware of their propensity for error; one study found that more than half lack a curriculum that addresses the fact that physicians frequently make cognitive mistakes. And according to a 2020 survey to which Singh contributed, “Clinicians were significantly less comfortable discussing diagnostic issues than other medical errors,” likely because acknowledging the possibility of such mistakes cuts to the core of who doctors think they are (and in a litigious culture, it can expose them to lawsuits). Without built-in feedback loops—patients rarely return to say You got it wrong—doctors operate in a system that rarely reckons with its own fallibility.
On top of this, as Sifferlin documents, doctors are overwhelmed by the demands of hospital systems and profit-driven insurance companies that reward speed over attention: 15-minute appointments, denied insurance claims, bureaucracy that erodes time with patients. Squeezed on all sides, physicians must make critical decisions under conditions that all but guarantee mistakes. One study at Johns Hopkins University found that the medical school’s first-year interns spent a mere 13 percent of their time in patients’ rooms.
The physical exam—once the key to diagnosis—is in decline; doctors almost immediately default to ordering tests in its stead. Yet a 2011 study suggested that at least 80 percent of diagnoses could be made by using basic clinical skills: Reporting on a bedside-medicine program at Johns Hopkins, Sifferlin documents a retired doctor with heart disease who served as a teaching patient. The intern tasked with diagnosing him wanted to order lab tests, an EKG, and an echocardiogram; her supervisor showed her that with a careful physical exam—checking his pulse, listening to his chest—and a look at his medical history, she could reach the diagnosis on the spot, with 99 percent certainty, as opposed to slowing down the path to diagnosis by deferring to tests.
In the public imagination, conditioned by the 20th century’s embrace of expertise (and pop medical mysteries such as House), diagnostic challenges are in many cases thought to revolve around rare diseases—a zebra where you should expect a horse. But Sifferlin reveals that the problem may be worse, paradoxically, when dealing with everyday illnesses—because it is more widespread. The larger crisis concerns patients such as Cejas, people with treatable problems that were missed because no one listened long enough to think again. As Sifferlin reports, “Between 65 and 80 percent of diagnostic errors can be attributed to breakdowns between the doctor and patients during clinical interactions.”
Still, beyond these unforced errors, there are more zebras than physicians are often willing to admit. Perhaps the worst mistake a doctor can make is not their believing that they know everything but their acting as if everything is known. Today, millions of people live with poorly understood chronic illnesses—myalgic encephalomyelitis / chronic fatigue syndrome, autoimmune diseases, long COVID—that don’t fit cleanly into established categories or show up on tests in their early stages. Sifferlin quotes from the work of the anthropologist Abigail Dumes, who has argued that the rise of evidence-based medicine had the unintended effect of excluding these conditions from “the trappings of medical legibility.”
[Read: The making of a diagnostic mind]
By learning to see inside the body, that is, medicine also learned to overlook patients whose illnesses aren’t already well studied. Gulf War veterans, for example, were told that their symptoms were caused by stress before researchers showed that nerve-gas exposure had caused real biological damage. And long-COVID patients still routinely have their symptoms ascribed to anxiety, despite rapidly growing evidence of the disease’s physical pathology. In such cases, uncertainty leads to dismissiveness. A medical system that can’t yet quantify what is wrong concludes that nothing is wrong, that it’s just in your head—even though, as doctors now hasten to add, the symptoms you are feeling are real.
It has become fashionable, in some medical-humanities circles, to question the importance of diagnosis itself. At a talk I gave about searching for a diagnosis as a patient living at what I called “the edge of medical knowledge,” a professor asked me, “What is the point of a diagnosis, anyway?” I know what he meant: A diagnosis is not the last word, just a step on a complicated journey. But as someone who went without one for more than a decade, I also know that getting a diagnosis matters. In our bureaucratized medical system, it unlocks treatment, insurance coverage, clinical-trial eligibility, and—perhaps most fundamental—recognition, a validation of your illness. In a study that Sifferlin cites, a woman searching for a diagnosis put it simply: “I just want permission to be ill.” So a better question is not Why do we care so much about diagnosis?, but rather Why isn’t the system doing more to get diagnosis right?
Sifferlin suggests that the answer isn’t primarily technological. The Undiagnosed Diseases Program, a program at the National Institutes of Health that identifies rare, new diseases, works not because its providers can sequence genomes at the NIH—though they do—but because of the time devoted to each case, and because of the collaboration and ferment that results from experts conversing outside of their silos, honing and exercising their clinical judgment. But most doctors today are not trained in a system that allows them to work this way.
In her book How Doctors Think, the medical-humanities scholar Kathryn Montgomery defines this kind of judgment as “an intellectual capacity carefully cultivated through the rigors of a long apprenticeship spent dealing with radical uncertainty.” But Brian Garibaldi, who directs the Center for Bedside Medicine at Northwestern University, tells Sifferlin that some medical students make it all of the way through their training without ever having been supervised while examining a real patient. Sifferlin describes an intern who aces an ultrasound assessment but fails to do a basic check on a patient’s reflexes; he didn’t spot the reflex hammer in the exam room.
Many people believe that artificial intelligence—with its extraordinary ability to assimilate information, identify patterns, and make quick judgments—will be just the thing to fix the diagnosis crisis. It is true that AI, by transcribing appointments, frees up doctors from sitting behind their computers and taking notes. It can also help physicians overcome unconscious bias, as the UC Berkeley researcher Ziad Obermeyer and others have shown. But The Elusive Body suggests (correctly, I think) that AI won’t be a panacea—at least not on its own. Obermeyer’s work also shows that AI as easily replicates existing bias as corrects it. Technology is deployed according to human values. The values that our medical system currently rewards—speed, volume, cost reduction—are the same ones that created the crisis we are now in.
Today, Diana Cejas works in pediatric neurology. Her experience changed the kind of doctor she is. “Sometimes it’s all I feel like I do, listen to patients. I just let them talk,” she tells Sifferlin. Her experience of misdiagnosis left her in a paradoxical position: “I want my patients to trust me. At the same time, I have a hard time trusting physicians.”
The word diagnosis comes from the Greek for “knowledge.” To offer a diagnosis is not just a medical act but an epistemological and moral one—a bridge from the world of the well to the world of the sick. Diagnosis requires recognition—the willingness to say: I believe your body is telling us something, even if I don’t know what it is. The crisis that Sifferlin documents—and that I’ve experienced as a patient—is that the system has made this act of witness, a foundational act of thousands of years of medicine, exceedingly difficult to perform.
The Elusive Body shows that it doesn’t have to be this way: The solutions are around us, and all that they require is a new orientation toward uncertainty, attention, and time—as well as the political will to solve the problem. Under the current federal assault on science, and in a moment when AI threatens to make each encounter even more remote, that person-to-person reorientation feels long overdue.