After heading back to his job managing a medical device firm in Murray Hill, N.J., Cherry went to see another doctor—and then a third—for relief from a new onslaught of scary symptoms: breathing problems, blocked sinuses, extreme weakness, rapid weight loss, and numbness in his extremities. One doctor "just gave me a nasal spray," he says, and the other suspected some exotic foreign disease when a blood test showed a spike in white blood cells. He told Cherry to see an infectious disease specialist immediately.
Besides relying on first impressions, physicians often put too much stock in their past experience—overdiagnosing tumors, say, because they missed one once. Or, instead of trusting their instincts, they defer to a more experienced colleague or to the adage drummed into them as medical students: "When you hear hoofbeats, think horses, not zebras." Result: They miss the zebra.
Dizziness, for example, frequently is explained by inner ear fluid, migraines, or a medication reaction. But about 5 percent of the time, it indicates a stroke, and these patients may have no other telltale signs like numbness or weakness on one side of the body, says Johns Hopkins neurologist David Newman-Toker, who has studied misdiagnoses in hospitals. A spinning room and pounding headache sent Margaret Powell, 59, to Hopkins last August; with those vague symptoms, Powell had about a 35 percent chance of being misdiagnosed, contends Newman-Toker. Fortunately for Powell, the emergency room doctor asked her the right questions: Was this her first dizzy spell? (Yes.) Lasted for more than two minutes? (Yes.) Not brought on by anything in particular? (Yes.) An MRI confirmed a stroke, and Powell was immediately put on blood-thinner drugs.
Physicians who pride themselves on their diagnostic skills say that, like Olympic athletes, they train constantly. Gurpreet Dhaliwal, an internist at the San Francisco Veterans Affairs Medical Center, hones his skills by devouring medical journal case reports and reviewing patient charts from other departments. While examining a patient, he says, he practices "meta cognition," or thinking about what he's thinking, to look for slip-ups. Finally, he taps into his patients' electronic records days or weeks after he first examined them to check that he made the right call.
Cherry managed to get an appointment with John Mann, a leading infectious disease specialist at Johns Hopkins, who noticed that several things didn't jibe with an infection. For one thing, most infections don't linger so long without putting the patient in the hospital. "For someone with so many things wrong with him, he wasn't feverish and didn't look acutely ill," Mann says. He began to suspect an autoimmune disease, a condition in which Cherry's body was attacking its own tissues.
Some experts believe the solution to reducing diagnostic errors lies in decision-support software like Isabel, which allows doctors to input a medical history and often disparate, puzzling symptoms and get a list of possible causes. The biggest drawback: It often lists too many possibilities because of common symptoms; overlap between a cold and bubonic plague, for example, is 95 percent. But such programs could prove extremely useful within the next few years, Wachter thinks, if a patient's own electronic medical record could be incorporated into the mix.
Newman-Toker is currently testing a computerized questionnaire that asks patients experiencing dizziness to tap in their symptoms while waiting to see an ER physician. The program, which assesses the likelihood of stroke, helps ER docs decide who should be admitted for a full diagnostic work-up. "Some doctors ask why they can't just do an MRI on everyone experiencing dizziness," says Newman-Toker. "But that's not fiscally responsible." He's tested the program on 1,000 patients so far and expects to publish results soon.